Dissertation Prospectus

4 | THE NEW REPUBLIC

ONE DAY IN 2012, Aileen Rizo, a math consultant in the Fresno, California, education system, overheard a recently hired male colleague talking about his salary. Rizo was “floored,” she said, to learn that although she had the same job title as he did, was better educated, and had more experience, he was paid more. After asking around, Rizo learned that this was no coincidence: Several of her male colleagues were earning significantly more than her as well.

“I felt like I was part of a picture and someone cut me out of the picture,” she said, describing what it was like to realize she was being paid so much less. “It was almost surreal.”

Still, Rizo figured it was a mix-up that could be remedied. When she complained to Human Res- ources, though, she was told that the county had relied on her contract with her previous employer— she’d been a schoolteacher in Arizona—to set her pay. “I thought it was being used to confirm my employment and years of experience,” she said. “I didn’t know they were using the number for my salary.” When the county refused to change her pay, Rizo sued, saying the policy discriminated against women. In April 2017, the U.S. Court of Appeals for the Ninth Circuit ruled against her, saying that companies could use

Closing the Wage Gap for Women The case against employers asking what you made at your last job

BY BRYCE COVERT

ILLUST R AT IO NS BY ALEX NABAU M

U.S. & THE WORLD

JULY/AUGUST 2018 | 5

prior salary to determine wages as long as it “was reasonable and effectuated a business policy.”

There is nothing new about using previous salary information to set pay. “That’s been around probably as long as the job interview,” said Deborah Vagins, a senior vice president at the American Association of University Women. In a survey earlier this year, 80 percent of hiring managers and recruiters said they relied on past salary to come up with an offer. It might seem like a neutral practice, but it can per- petuate the inequities that mean women and people of color are paid less, on average, than white men. Even women fresh out of college make less than their male peers in their first jobs. If future pay is based on previous earnings, then the original sin of an initial hire taints a woman’s entire working life. “If this disparity can begin from the moment you go to your first job, and it follows you throughout your career, it will never be rectified, and the wage gap itself will never be rectified,” the District of Columbia’s Representative Eleanor Holmes Norton, a leading advocate to end the practice, told me.

Rizo wasn’t the first woman to challenge salary- history policies in court. In 2000, 37 female employ- ees at Boeing’s Puget Sound headquarters filed a class action lawsuit against the company, alleging that its use of pay history to determine their salaries was among the factors that stopped them from advancing within the company. (Boeing settled the suit and agreed to use new methods to determine starting sal- aries.) Within the last few years, the issue has made its way into legislatures. In 2016, Massachusetts became the first state to forbid employers from requesting salary histories before they make an official job offer. Since then, Connecticut, Delaware, Oregon, Vermont, and Rizo’s home state of California have all passed similar laws, while versions have been introduced in Maine, New York, and Washington, D.C. New York City and Salt Lake City have passed their own bills. Major corporations have joined in: Amazon, Amer- ican Express, Bank of America, Facebook, Google, Starbucks, and Wells Fargo have all announced that they will stop using prior pay to set compensation.

IT HAS BEEN illegal to pay women less than men since 1963. So are salary-history statutes and new hiring practices a solution to a problem that has already been solved? Far from it. Women who work full time still make, on average, 80 percent of what men make, and women of color make even less. It often falls to them to fix the gap by driving a harder bargain with their bosses and, if that fails, to sue employers if they aren’t paid fairly. Lawsuits take time and cost money, however, and women often lose. Salary-history bans, on the other hand, not

only give them more power in a negotiation, they place the responsibility for ensuring fair pay where it belongs: not on women and people of color, but on the employers who perpetuate inequality.

This approach hasn’t always been successful. Illi- nois’s governor, Bruce Rauner, last summer vetoed a salary-history ban passed by the state legislature. In Philadelphia, the first city to pass a salary-history ban, a federal judge struck the law down because he said it violated employers’ First Amendment rights. Federal solutions, including proposed bills from Representative Norton, which she introduced in 2016 and again in 2017, and the oft-revisited Pay- check Fairness Act, have gone nowhere in Congress.

Critics, including commentators and academics, draw a parallel between salary-history bans and the shortcomings of laws that prevent employers from asking about criminal histories. Studies have shown that in places that have banned criminal history questions, employers may simply assume that all black men have records and end up offering them fewer jobs as a result. The concern is that companies will likewise undercut women in salary negotiations based on the assumption that they all earn low pay. One study found that women who refused to disclose their salary histories during the hiring process did indeed get lower offers than those who did.

But that study looked at a scenario where there was no ban in place. A recent field experiment found that employers who couldn’t see salary histories actually did more individualized research into candidates, and candidates were better able to bargain for higher start- ing salaries. This suggests that the value of preventing

employers from asking about salary history extends beyond women and people of color. All workers are placed at a disadvantage when an employer knows what they are already willing to work for—in other words, how low they’ll go. Employers “can drive the [salary] conversation down,” said Hannah Riley Bowles, an expert in how gender influences pay ne- gotiations at the Harvard Kennedy School.

The debate may be ongoing, but Rizo has won her argument. In April, the Ninth Circuit reversed its earlier decision. “Women are told they are not worth as much as men,” Judge Stephen Reinhardt wrote. “Allowing prior salary to justify a wage differential perpetuates

Number of states with no equal pay protections 2 (Mississippi, Alabama)

Number that have passed salary-history bans: 6 (California, Connecticut, Delaware, Massachusetts, Oregon, Vermont)

Change in the amount of money candidates were able to get in salary negotiations: +9 percent

For every dollar a white man makes, women make: 87 cents (Asian) 79 cents (white) 63 cents (black) 54 cents (Hispanic)

Source: AAUW; Georgetown and NYU Stern

The laws place the responsibility for ensuring fair pay where it belongs: not on women and people of color, but on employers who perpetuate inequality.

6 | THE NEW REPUBLIC

ROYALIST MANIA TRANSCENDS traditional political divisions in the United States. Liberals, who decry entrenched privilege at home, seem strangely OK with a British aristocracy that conveys titles and estates through bloodlines. Fox News talking heads, who denounce coastal “elites” and the Ivy League, nonetheless carried breathless live coverage of Prince Harry and Meghan Markle’s wedding in May. A 2015 YouGov poll found that Americans, Republicans and Democrats alike, held more favorable opinions of the British queen, Prince William, Prince Harry, and the Duchess of Cambridge than of their own politi­ cians. Even the most popular American politician, Barack Obama, had a favorability that fell below their net rating by a considerable 34 points.

Donald Trump, with his penchant for Versailles­ style gilded furniture and his predilection for stamp­ ing the family crest on his properties, seems to have

a particularly bad case of this national affliction. In April 2017, The Times of London reported that White House staffers had demanded the full Cinderella treatment for his planned state visit: a gold­ plated carriage ride to meet the queen at Buckingham Palace. (Alas, Trump will have to make do with a more subdued meeting with Theresa May in July, and British officials have hedged on committing to a royal audience, concerned about the possibility of mass protests.)

Very little seems to unite Americans these days— except, apparently, their enjoyment in fawning over the rulers the Founding Fathers waged war to over­ throw. Once, the United States claimed egalitarian­ ism as a central ideal. What happened?

It’s not difficult to see how nostalgia for a system that finds dignity in stasis could take hold. American social mobility, depending on which economist you favor, has either been in steady decline for decades or has at the very least failed to keep up with widen­ ing inequality. Today, those born without privilege face daunting barriers to wealth and advancement. And even in the privileged upper class, the scale of competition—plummeting acceptance rates at elite universities, for example—makes it hard to live up to the assumption, hammered into American children from an early age, that they are “special.” Sleep depri­ vation, which affected 11 percent of Americans in the 1940s, is now a “public health epidemic,” according to the Centers for Disease Control and Prevention. The percentage of people who “worry a lot,” Pew Research analysis shows, has been rising for all income levels since 2003. And prescriptions for both stimulant medications, to keep up in an increasingly chaotic and distracting world, and sedatives, to unwind when it overwhelms, have jumped accordingly.

“This permanent struggle—between the instincts inspired by equality and the means it supplies to satisfy them—harasses and wearies men’s minds,” Alexis de Tocqueville wrote of the United States in the early 1800s. Americans may believe in equality and meritocracy, but if their obsession with the royal family is any guide, they yearn for a time when fulfillment wasn’t quite so much work.

THE WESTERN WORLD has long seen upticks in nostalgia and reactionism when people are frustrated,

this message, entrenching in salary systems an obvious means of discrimination.” When Jerry Brown signed California’s Fair Pay Act into law in 2015, Rizo and her nine­month­old daughter were there with him. “Sometimes I think things are too hard and never going to change, but it did,” she said. Today, Rizo is running

Americans may believe in meritocracy, but if their obsession with the royals is any guide, they yearn for a time when fulfillment wasn’t so much work.

U.S. & THE WORLD

for the state legislature—against an incumbent who has voted against bills that would have forbidden em­ ployers from asking questions about salary history. a

Bryce Covert is a contributing writer at The New York Times and The Nation.

Wet, Hot, Aristocratic Summer Donald Trump, Meghan Markle, and America’s enduring obsession with the British royals

BY HEATHER SOUVAINE HORN

Copyright of New Republic is the property of TNR II, LLC and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Student Loan Repayment W ind Energy Production

Many Americans have outstanding undergraduate stu­dent loans with interest rates of 7 percent or higher; however, those who took out loans during the 2013-2014 school year pay a rate of 3.86 percent under the Biparti­ san Student Loan Certainty Act, passed by Congress in 2013.

On March 18, 2014, Senator Elizabeth Warren (MA- D) and Representative Joe Courtney (CT-D) introduced the Bank on Students Emergency Loan Refinancing Act. The legislation would allow those with college loan debt to refi­ nance at the lower interest rates. The rates would be slightly higher for graduate student loans. The cost would be paid for by requiring millionaires to pay at least a 30 percent ef­ fective Federal tax rate.

A similar bill came to the floor in the last Congress, but fell short of breaking a Republican filibuster.

“Since last year, nearly a million more borrowers have fallen behind on their student loan payments,” said Sena­ tor Warren. “The Bank on Students Emergency Loan Re­ financing Act would give much-needed to relief to millions of borrowers, help boost our economy, and strengthen America’s middle class.”

On March 25, Senator Warren offered the bill on the Senate floor as an amendment to the budget resolution, but senators rejected it by a vote of 46 to 53.

Senate Budget C om m ittee C hair Mike Enzi (WY- R) objected to the mechanism for considering the bill, saying, “Addressing college costs and the burden of high stu d en t debt loans has to be done, b ut it can’t be done on a budget bill. You can’t have policy on a b u d ­ get resolution.”

Instead, the Senate approved, by voice vote, an amend­ ment introduced by Senator Richard Burr (NC-R) that would consolidate various Federal student loan programs and give students a choice of accepting a payment plan over 10 years or repaying loans based on income. Senator Burr explained that, with his legislation, “students will know, prior to entering college, based on the amount that they borrow, what options will be available to them once they graduate from college.”

Senator Mark Warner (VA-D), a cosponsor of the amendment, said, “It’s time to replace our complicated ar­ ray of loans, subsidies, deferments, and forbearances with streamlined, improved repayment options where graduates repay what they borrow based on what they earn.”

For more background and legislative history, see the November 2009 issue of Congressional Digest on “Federal Student Loans.” ■

Federal subsidies for wind production began under the Administration of President Jimmy Carter with passage of the Public Utility Regulatory Policy Act and the Energy Tax Act. When these subsidies failed to make the industry competitive, Congress, in 1992, created the Production Tax Credit (PTC) to give it a boost. The PTC gives wind en­ ergy producers a tax credit of 2.2 cents per kilowatt hour of electricity generated.

Although originally intended as a temporary measure, the PTC has been continually extended by Congress un­ der pressure from the wind industry and renewable energy advocates.

A setback occurred on January 29, 2015, however, when the Senate defeated, 47 to 51, an amendment by Senator Heidi Heitkamp (ND-D) to extend the PTC for another five years. The proposal was considered as an amendment to a bill to approve the Keystone XL pipeline, which was passed but later vetoed by the President.

Following the vote, Senator Heitkamp stated:

There are a lot of Senators that talk about support­ ing an all-of-the-above energy strategy, but clearly many don’t actually mean it. As we continue to cal­ culate a path forward for our energy infrastructure, and for fossil fuels like oil, gas, and coal... we sim­ ply cannot leave wind power out of the equation.

Opponents argued that after more than 20 years of being subsidized by taxpayers, it was time for the wind in­ dustry to stand on its own.

Meanwhile, in his 2015 budget proposal, President Obama called for a permanent extension of the PTC as well as the solar energy Investment Tax Credit (ITC). While the President’s proposals are unlikely to become law, they may serve as a high starting point for negotiating an extension of subsidies for renewal energy.

For more background, see the February 2013 issue of Congressional Digest on “Wind Energy.” ■

Equal Pay Protections

The Equal Pay Act of 1964 made it illegal to pay em­ployees different wages based on their sex. In addi­ tion, the National Labor Relations Act says that employ­ ers cannot prevent employees from discussing wages and other issues. And in 2009, Congress passed, and President Obama signed, the Lilly Ledbetter Fair Pay Act, which

14 C o n g re s s io n a l D ig e s t ■ w w w .C o n g re s s io n a lD ig e s t.c o m ■ M a y 2 0 1 5

amends the 1964 Civil Rights Act to state that the 180- day statute o f lim itations for fding an equal pay lawsuit resets w ith each new paycheck affected by that discrim i­ natory action.

Nevertheless, the gap between men’s and women’s wages has persisted. Although the wage gap varies by State and race, women are nationally estimated to earn 78 cents for every dollar earned by a man. (That figure was about 59 cents when the Equal Pay Act became law.) As a result, pro­ posals have been put forth to reaffirm and better enforce laws that are already on the books.

T he Senate budget resolution passed in M arch 2015 included an am endm ent clarifying that employees should n o t be penalized for discussing wages. D uring the same debate, senators rejected a proposal offered by Senator Bar­ bara M ikulski (M D -D ) requiring stronger penalties for unequal pay.

O n March 26, Senator Deb Fischer (NE-R) introduced S. 875, to make it illegal for employers to retaliate against employees who discuss or ask about comparative compen­ sation. T he bill would also prohibit pay discrim ination unless the differential could be justified by seniority, merit, or some other factor.

O ne year ago, President O bam a took two Executive actions aimed at narrowing the wage gap:

• An Executive order prohibiting Federal contractors from retaliating against workers who discuss their sala­ ries with one another.

• A m em orandum requesting new rules to require Fed­ eral contractors to submit data on employee compen­ sation by race and gender.

T he um brella bill currently supported by equal pay advocates is S. 862, the Paycheck Fairness Act, sponsored by Senator M ikulski. T he measure is designed to help those who believe th at they have experienced pay dis­ crim ination by m aking wages m ore transparent, requir­ ing th at employers prove th at wage discrepancies are tied to legitim ate business qualifications, and p ro h ib itin g companies from taking retaliatory action against employ­ ees who raise concerns.

O pponents o f the bill argue that the statistic that women earn 78 cents for every dollar a man earns is mis­ leading and misapplied because it is based on the wrong comparisons. They point to studies showing that when the job itself, experience, and hours o f work are taken into ac­ count, women make about the same am ount as men. They also say that there are already sufficient laws under which women can seek justice for wage discrimination.

For more background, see the May 2014 issue o f Con­ gressional Digest on “Women’s Pay Equity.” ■

Dietary Guidelines

In February, the Dietary Guidelines Advisory Com m it tee subm itted its Scientific Report to the Secretary of Health and H um an Services and the Secretary o f Agricul­ ture. The committee was charged with examining where sufficient “new scientific evidence is likely to be available that may inform revisions to the current guidance or sug­ gest new guidance.”

Based on their research, the committee concluded that

... a healthy dietary pattern is higher in vegetables, fruits, whole grains, low- or non-fat dairy, seafood, legumes, and nuts; moderate in alcohol (among adults); lower in red and processed meat; and low in sugar sweetened foods and drinks and refined grains.

The report continued:

Consistent evidence indicates that, in general, a dietary pattern that is higher in plant-based foods ... and lower in animal-based foods is more health promoting and is associated with a lesser environ­ mental impact (G H G [greenhouse gas] emissions and energy, land, and water use) than is the current average U.S. diet.

T he emphasis on lower red and processed meat con­ sum ption provoked a strong negative response from the meat industry, and a request that the public com m ent pe­ riod for the report be extended from 45 days to 120 days. T he deadline has since been moved to May 8.

N orth American Meat Institute (NAMI) Vice Presi­ dent o f Scientific Affairs Betsy Booren argued that meat and poultry products are an im portant com ponent o f a health­ ful American diet. NAM I has announced a petition drive opposing the recommendations.

Meanwhile, Senate Agriculture C om m itte Chair Pat Roberts (KS-R) has said that he supports the industry po­ sition and that the recommendations “make a certain seg­ m ent of agriculture a target.”

Concerns on both sides were voiced at a House Agri­ culture Appropriations Subcommittee on March 17.

T he Agriculture and Health and H um an Services D e­ partments will review the report, along with public com­ ments and input from other agencies, before beginning the process o f updating the guidelines. ■

C o n g r e s s io n a l D ig e st ■ w w w .C o n g re s s io n a lD ig e s t.c o m ■ M a y 2 0 1 5 15

Copyright of Congressional Digest is the property of Congressional Digest and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1224

www.internationaljournalofcaringsciences.org

Original Article

Gender Equality in Academia, Business, Technology and Health Care: A WomEnPower view in Cyprus

Christiana Kouta, PhD Associate Professor, Department of Nursing, Cyprus University of Technology, Limassol, Cyprus

Antigoni Parmaxi, PhD Research Associate, Cyprus Interaction Lab, Department of Multimedia and Graphic Arts, Cyprus University of Technology, Limassol, Cyprus

Irene Smoleski, MSc Research Assistant, Department of Nursing, Cyprus University of Technology, Limassol, Cyprus

Correspondence: Christiana Kouta, Associate Professor, Department of Nursing, Cyprus University of Technology, 15 Vragadinou str, 3041, Limassol, Cyprus E-mail: christiana.kouta@cut.ac.cy

Abstract

The aim of this article is to present the findings of a qualitative study aiming at understanding women’s perceptions with regard to a) gender equality at workplace; b) experiences at workplace with regard to gender; c) gender wage gap; d) use of technology for addressing issues of inequality and e) suggestions for the development of an e-mentoring community platform. This study sketches the current situation of gender equality in the fields of academia, business, technology and health care, and provides deep understanding of the difficulties that women with different levels of experience and expertise encounter in their workplace as well as how technology could help them overcome these issues. Data collected demonstrate a variety of challenges faced by women in workplace as well as the need for role models that will allow young women to overcome the stereotypical woman profile as excluded from economic, political and professional life.

Key words: Gender, technology, discrimination, empowerment, perceptions, qualitative

Introduction

Gender inequality can be defined as the lack of “discrimination in relation to opportunities, allocation of resources or benefits and access to services for women or men” (Elwer et al., 2012, p.1). In all EU Member States, female employment rates are lower than those for males. When employment is measured in full-time equivalent, the picture is even worse (OECD, 2012).

Despite the efforts made for shrinking the gap between men and women, the underrepresentation of women in higher positions still exists. Across the European economy women earn on average 16.4% less than men, whilst in USA working women earn 77 cents for every dollar earned by men (EU Equality Challenge Unit, 2014; Smith, 2014; Bryant et al., 2015). Neyer et al. (2013b) conceptualize gender equality beyond ‘‘sameness

of distribution’’, providing three dimensions of gender inequality related to employment, economic resources and the division of housework and family care. Gender equality is achieved when one is able to access and enjoy the same resources, opportunities and rewards regardless their gender (Workplace, Gender Equality Agency, Australian Government, 2013). This is a complex matter, involving economical, demographic and behavioral factors that may contribute to increase gender-based gaps in the labor market (ILO, 2012). The newly adopted UN agenda for 2030 highlights the importance of women’s empowerment in employment, salaries and working environment as a basic human right (UN news center, 2015).

Research studies demonstrated that women suffer from low rates of participation in the workforce, decision making and unequal value of their work (Monroe, et al. 2008; Loscocco & Bird, 2012;

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1225

www.internationaljournalofcaringsciences.org

Blackburn, Jarman, & Racko, 2015). Yet, missing women from professional careers affects both the workforce, as it misses women’s perspective and expertise; and women themselves. Further, most technology is designed by men and one need to consider that technology then reflects those who make it (IGNITE, 2014).

Despite the emphasis given in high level political decisions for encouraging women to reach equality, researchers and practitioners often lack understanding of women’s perspective with regard to gender equality and value of their work. This paper aims at portraying the current situation in gender inequality by taking a snapshot on the way women experience and ascribe meaning to it in the fields of academia, business, technology and health care. The paper presents a qualitative study that brought together women from different areas of work with different levels of experience to voice their views regarding to the status of women in their work area, obstacles that they encounter as well as how they perceive technology as a means for overcoming obstacles in their professional development. Authors provide an overview of the state-of-the art of gender equality in the workforce; methodology follows. The article concludes by linking the empirical results to the existing literature.

Gender equality in the workforce

The under-representation of women in high- ranked positions is a pattern that occurs across several occupations across the globe including health care, academia, entrepreneurship, business; Science, Technology, Engineering and Math (STEM). Although the number of women in authority positions increases, there is still a continuation of discrimination and women experience with regard to downplaying (Monroe et al. 2008). Loscocco and Bird (2012) demonstrated that women are more likely to work in part time works because of childcare, so there is a limited chance to have a director position due to reduced work’s hours. According to Kogut et al. (2014) this is the case in Norway, where, one woman to seven men holds a director position and a percentage of 20% retain structural equality. As indicated by Beede et al. (2011, p. 1), “although women fill close to half of all jobs in the U.S. economy, they hold less than 25 percent of STEM jobs. This has been the case throughout the past decade, even as college educated women have increased their share of

the overall workforce”. Similarly, recent research evidence points systemic gender discrimination and inequality in health workforce. Health care professionals’ work is traditionally associated to femininity as women constitute the majority of health care workers (WHO, 2002; 2008), yet women’s salary in such positions is devaluated in the labor market (Tijdens, De Vries & Steinmetz, 2013). As pointed out by Newman (2014) more attention needs to be paid by governance and human resource for health (HRH) leaders on understanding inequality in the health care domain. Newman (2014) provides a number of specific actions to be carried out which include a unified conceptual framework for gender inequality in the health workforce, research guidance and improvement of HRH policies and practices.

Women’s’ representation in the workforce is decisive to a country’s social, economic and innovation competitiveness. Higher capacity innovation, financial and political growths are amongst the benefits reported for drawing policies that promote equal opportunities. It is a rather constricted view to believe that increase of women’s participation in workforce will reveal novel economic and political growth. However, encouraging and supporting women in the organizational agenda will allow for a different perspective to be heard in social, political and economic discussions.

Womenpower platform

In an attempt to give women a voice in the arena of workplace, Womenpower (WE-ME) was developed. Womenpower is a community platform aiming to connect different generations of women for addressing issues related to women equality in workplace. It embarks to assist young women to receive support and solidarity from women with expertise in their area. Ultimately, through Womenpower a network of women will be developed that will enable women to join forces for achieving their goals.

For the development of Womenpower platform a user-centered design (UCD) approach was followed which aspired to contribute towards a user-friendly system that will encourage young women to receive support for breaking the unseen barriers in their professional development, and eventually reach higher levels in the corporate ladder. UCD is a framework for hardware and software development that ensures maximum involvement of key players (Norman

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1226

www.internationaljournalofcaringsciences.org

& Draper, 1986). Thus, users will be an integral part of any software or hardware development.

For the development of Womenpower platform the research engaged in state-of-the-art-research in the area of gender equality in academia, business, technology and health care. Data from research manuscripts formed an interview protocol that was used for elucidating information from women in lower and higher ranks in academia, health care, technology and business throughout three focus groups that were held (Chen et al., 2013; Elwer et al., 2012; Ding et el., 2006) These data enabled the research team to depict the current situation in academia, business and health care as well as to elucidate different views with regard to the use of technology for mentoring and support. Mentoring provides opportunities for women for professional development as well as personal achievements (Mentoring Women’s Network, 2015). Moreover, building on women’s views, a working prototype of the platform was developed that enabled users to provide feedback on how the e-mentoring platform would work (Parmaxi & Vasiliou, 2015).

Methodology

Study Design

To gain an in-depth understanding of participants’ views of the role of women in the workplace a qualitative methodology was employed.

Sampling

Three focus groups were implemented. The focus groups involved both women in senior and junior positions in the areas of academia, business and health care. Three focus groups took place, two with junior participants (focus group 1, n=10; focus group 2, n=6) and one with senior participants (focus group 3, n= 8). The aim was for all four workplaces (health care, academia, business and technology) to be represented in both senior and junior participants. A convenience sample was used. Participants were recruited though researchers’ personal and professional contacts with key people in these fields. Researchers contacted the interviews though did not know the participants personally and no conflicting interest or relation existed.

Participants’ ages and career stage varied among the groups. The inclusion criteria were the participants to be females, from the fields of

business, health care, academia and technology. In addition for seniors to have a managerial, decision making position for more than 5 years. Moreover, for the junior participants other criteria were to enter the profession the past 5 years and not to have a managerial or an authority position.

Tool

A focus group guide was designed based on the literature review (Ritchie, 2013). The following thematic areas were revealed: a) gender equality at workplace; b) experiences at workplace with regard to gender; c) gender wage gap; d) use of technology for addressing issues of inequality and e) suggestions for the development of an e- mentoring community platform.

Data Collection

The focus groups were conducted in three different dates in agreement with participants. All authors facilitated the focus group discussion; two authors participated in each discussion. Each focus group lasted approximately 60-80 minutes. The facilitators followed the focus group guide with the thematic areas mentioned above. Discussion was recorded with the permission of the participants.

Data Analysis

Thematic analysis was performed in order to extract key themes related to the areas mentioned earlier. Although thematic analysis is generally understood as an analytic technique used in the context of different qualitative methodological approaches e.g. grounded theory, phenomenology etc., it can also be used independently as a flexible method of analyzing qualitative data guiding the search for themes or patterns within the data (Braun & Clarke, 2006). Further, this kind of research involves that the research team studies the data in their natural settings in order to interpret the results and ascribe meaning to them to make sense (Denzin & Lincoln 2005, 2009). The analysis was based on six thematic analysis steps: Familiarizing researcher with data, generating codes, searching for themes and reviewing themes, defining themes and produce the report (Braun & Clarke, 2006). The recorded focus groups were transcribed verbatim by the research team. To guide the systematic analysis the topics guiding the interviews were used thematic categories. Data were repeatedly read and no other categories were developed.

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1227

www.internationaljournalofcaringsciences.org

Ethical Considerations

All participants were informed about the research study in person verbally and in writing. Each participant signed a consent form. It was also explained that participants had the right to withdraw at any time of the study. During the focus groups all principles of Belmont report were followed and applied. The principles of confidentiality, anonymity and personal data were also taken in consideration.

Results

Data collected indicated similarities as well as discrepancies between women in different areas and different years of expertise. Figure 1 provides an overview of the stances voiced by participants in the areas of gender equality at workplace, experiences at workplace in regards to gender, gender wage gap, and use of technology. Participants in mentors’ focus group expressed similar views regarding gender equality articulating equal opportunities in their workplace. On the other hand, junior participants from the field of health care had a different view of gender equality at workplace. In the following section we present the analysis of the data categorised in the five areas mentioned earlier: Gender equality at workplace, experience at workplace in regards to gender, gender wage gap and the use of technology for addressing gender equality issues.

1. Gender equality at workplace

Most of the participants in the senior focus group expressed similar views with regard to gender equality. They stated that they were given equal opportunities and employers did not discriminate due to gender. However, it was noted from some senior participants the general feeling end experience that sometimes things are not as equal as they seem.

“It is obvious that our directors believes that men can do better management than us.” (Participant health - senior 3)

Junior participants from the health field had a different view of gender equality at workplace.

2. Experiences at workplace in regards to gender

Participants from the senior focus group expressed the influence of Cypriot culture at

work place in association to gender equality. A senior participant from the field of business expressed the feeling that Cypriot societal and cultural influences are strong and men hold most managerial positions.

“… I realized that as a woman I could never hold a managerial position. I think that our society is one of the communities in which men are thought to be remarkable and capable enough to hold managerial positions.” (Participant business - senior 3)

Junior participants agreed that there is inequality at workplace, however there was a strong discussion with regard to woman’s role at work and family. The “glass ceiling” appears in the Cypriot society, as women seem incapable of reaching high level positions in their workplace. In such a society, unseen barriers prevent women to claim higher positions. For example, a junior participant from the field of business noted that in a company aiming at the greatest possible profit, men are preferred since women are more emotional and may not be able to cope with difficult situations or hard decisions. Sometimes, even women employers are been more suspicious towards women employees.

“…Every problem we have with machines we are looking for a man to fix it. We find a male colleague to do it…we believe that we are not good in engineering. And I wonder if women do not have the inclination to technology or if we prefer not to deal with.” (Participant health - junior 10)

Participants seemed to agree that women face challenges at work, however in a different degree.

“I often feel that not only my boss, but staff also, expect me to do something more to prove my abilities to manage difficult tasks …” (Participant business- senior 3)

Participants also voiced pregnancy as a barrier that women need to address in their workplace:

“In some cases if you are pregnant and you take sick leaves/time off for breastfeeding the baby, they will fire you...or cut off a large part of your salary…There are companies that have in their requirements that the woman needs to sign that she will not get pregnant for 3 years.” (Participant health - junior 11)

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1228

www.internationaljournalofcaringsciences.org

Figure 1. Overview of the stances voiced by participants in the areas of gender equality at workplace, experiences at workplace in regards to gender, gender wage gap, and use of technology.

3. Gender wage gap

Participants from all groups expressed contradicting views with regard to women’s earnings vis-à-vis men’s. Moreover, different experiences were revealed between women working in public vis-à-vis private sector. For women in private sector gender inequality and wage gap was more visible in their workplace.

“…I have a salary difference (less 20%) from my male colleagues” (Participant business - senior 1- private sector)

4. Use of technology and safety issues

With regard to the use of technology for developing a community e-mentoring platform, all participants demonstrated a positive stance. Controversies arose for issues of anonymity, safety and privacy. Participants in the senior group unanimously agreed that they would prefer being anonymous on the platform. Participants from the technology industry indicated that there must be a name to increase credibility - pseudonym. There is a possibility to have a list of mentors, but mentors to be anonymous. Mentoring can be take place both publicly or privately -starting from the platform and then expanding to the real world. All participants

were struggling in regards to the fact that Cyprus is a small society and most people know each other and this may influence their work.

Cultural underpinnings need to be taken into account, as culture is deeply embedded and difficult to be reformed. Although improvements have been made, still very few women are in decision making positions. According to Cuddy et al. (2010) culture can shape the contents of gender stereotypes.

Discussion

Despite the growing social and political effort to establish gender equality women still experience inequality in their professional development. There are many to be done yet. Some of the obstacles in effective pursuit of gender mainstreaming and equality policies include limited accountability mechanisms in public agencies, lack of awareness on the different effect that policies may have on men and women and lack of an effective monitoring system in evaluating gender equality initiatives and actions (OECD, 2012). All these need to have a good coordination system as to have a useful and meaningful result.

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1229

www.internationaljournalofcaringsciences.org

In this study, there are similarities but also differences in the views amongst the groups. However, in principle there is an agreement that gender inequality exists in many sectors and within daily life in Cyprus. It seems that women who work in the public sector have experienced less gender inequality compared to those working in the private sector. This is understandable in regards to payment as public sector has payment scales with no gender differentiation. In private sector, although scales may exist, the employer may alter salaries based on different arguments such as productivity, years of experience, education and so on.

Reconciling family and work is an urgent need to be applied in the Cyprus context. Considering gender and employment puzzle, tensions between family and work life consist the heart of it (OECD, 2012). Neyer et al. (2013a) noted that mothers often come across difficult dilemmas and need to choose between maintaining their job and having another child. Parenthood is one of the main factors underlying the gender employment gaps. In most EU Member States, the employment rate for women who have children is much lower than for women without children; while this is the opposite for men (OECD, 2012). Cyprus social welfare support is almost nonexistence in regards to family friendly policies such as provision of part time jobs in public sector, working from home, nurseries at workplace. This becomes more difficult with women in more needs such as single parent families, where women are the majority in caring and providing for these families.

These actions enhance reconciliation of family and working life and allow women to be productive and take the chance of decision making and/or managerial positions. Further, strong and sustainable balanced economic growth can be achieved by promoting and improving female working opportunities (OECD, 2012).

Findings in the Neyer et al. (2013a) study demonstrated that directors believe that women are more emotional and may not be able to cope with difficult situations or decisions. However, investing in women’s leadership has essential effects on a country’s Gross Domestic Product GDP and the welfare of next generations (Booz and Co, 2012). A research of 7280 leaders conducted by Zenger Folkman (2012) shows that women excel at most leadership competencies.

Women’s ideas and business receive less start up investment and venture (IGNITE, 2014), while may provide creative and innovative approaches.

Ronnblom et al. (2005) analyzed the gender mainstreaming in regional policies reported that due to economic growth they did not give senior positions to women.

It seems that the participants would like to use the technology and are positive for the development of a community platform that would bring together women mentors and mentees. This reinforces the usability of the platform. According to Kogut et al. (2014), the explosion of data nowadays through the use of social networks can improve the structural equality. Small changes can have big achievements that could be a remarkable improvement for women. With regard to safety and security issues, the senior group would like to have an anonymous profile while junior participants would like to know the name or at least the status/specialty of the mentor.

Overall, focus groups revealed that the platform will be useful to both groups of women- mentor and mentees. The senior group stated that the platform will respond to the needs of the participants to be available at home page with their qualifications. It will be important that questions from mentees to be accepted in both languages, Greek and English, otherwise a number of young women in need will be excluded.

Further, there is a need to express their thoughts, discuss the problems and challenges of women at workplace, and also success stories. As there is the chance of a foreign mentor or mentee, translation should be available and needs to be done probably by the research team considering gender and cultural sensitivity. Cultural norms and discriminatory social institutions often restrict the economic and social role of women worldwide (OECD, 2012).

In WomEnpower platform, all questions and answers need to be filtered by the platform’s coordinator for reasons of anonymity, confidentiality and safety. It is important to note the women’s willingness to advice and support other women in all sectors, in a professional and supportive manner. This highlights the felt need from mentees and mentors and at the same time recognizing the usefulness of such platform. The platform provides the mean to set the bases for a

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1230

www.internationaljournalofcaringsciences.org

productive and creative community of women and men at the same time.

Conclusion

Crossing the 21st century with no doubt there is gender inequality in workplace. Employers should be flexible, provide and support opportunities and initiatives at work regardless gender. Employers should provide equal opportunities and salaries for men and women. Empowering women to participate in all aspects of everyday life can achieve locally and globally agreed goals for development and sustainability in many aspects. This may improve the quality of life for women, men, families and consequently communities. Local policies and strategies need to be revisited and enhanced. The capacity of government needs to be strengthening in the application of gender responsive and sensitive approach throughout the local financial management starting from the public sector. There are many leading women figures in all sectors that can be used as paradigms and/or success stories. Within a time of economic crisis, women can have their chance and role. This is a collective benefit and needs to be seen as such- supporting gender equality is the adoption and implementation of a human right’s principle- equality.

Limitations of the study

The findings of this study cannot be generalized. The convenience sample used in forming the focus groups and the fact that participants represented three sectors only consist also a limitation.

Acknowledgements

The work is funded by Mahallae, United Nations Development Programme (UNDP), Action for Cooperation and Trust (Agreement Number: 87733-MAHALLAE-01).

References

Beede, D.N., Julian, T.A., Langdo,n D., McKittrick, G., Khan, B., & Doms, M.E. (2011). Women in STEM: A gender gap to innovation. Economics and Statistics Administration Issue Brief, (04-11).

Booz and Co (2012) Empowering the Third Billion: Women and the World of Work in 2012. Available at: http://www.booz.com/med ia/uploads/BoozCo_Empo wering-the- ThirdBillion_Full-Report.pdf (accessed 4 February, 2017)

Blackburn, R. M., Jarman, J., & Racko, G. (2015). Understanding gender inequality in employment

and retirement. Contemporary Social Science, 1- 15. DOI:10.1080/21582041.2014.981756

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77-101.

Bryant, M., Winters, P., & Matson, J. (2015). Gender pay gap. Uhuru, Available at: http://lib.dr.iastate.edu/uhuru/vol10/iss3/10/?utm_ source=lib.dr.iastate.edu%2Fuhuru%2Fvol10%2Fi ss3%2F10&utm_medium=PDF&utm_campaign= PDFCoverPages (accessed 6 May 2017)

Chen, Z., Ge, Y., Lai, H., & Wan, C. (2013) Globalization and Gender wage inequality in China. World Development, 44, 256-266.

Coontz, S. (2013). Why gender equality stalled. New York Times, Available at: http://www.nytimes.com/2013/02/17/opinion/sund ay/why-gender-equality-stalled.html?_r=0 (accessed 1 December 2016)

Cuddy, J. C. A., Crotty, S., Chong, J., & Norton, I. M. (2010). Men as Cultural Ideals: How Culture Shapes Gender Stereotypes. Harvard Business School. Available at: http://www.hbs.edu/faculty/Publication%20Files/1 0-097.pdf (Accessed 1 December 2016)

Denzin, N. K., & Lincoln, Y. S. (2009). Handbook of Qualitative Research (Edisi Bahasa Indonesia). Yogyakarta: Pustaka Pelajar

Denzin, N. K., & Lincoln, Y. S., (2005). Qualitative research. 3rd Edition. California: Sage.

Ding, W. W., Murray, F., & Stuart, T. E. (2006). Gender differences in patenting in the academic life sciences. Science 313(5787), 665-667.

Elwer, S., Alex, L., & Hammarstrom, A. (2012). Gender (in) equality among employees infielder care: implications for health. Int J Equity Health 11, 1-10.

European Union Equality Challenge Unit (2014). Tackling the gender pay gap in the European Union. Belgium: European Union. Available at: http://ec.europa.eu/justice/gender- equality/files/gender_pay_gap/140319_gpg_en.pdf (accessed 4 February, 2017)

Foundations of qualitative research in education. Thematic analysis. (2015). Available at: http://isites.harvard.edu/icb/icb.do?keyword=quali tative&pageid=icb.page340897 (1 December, 2016)

IGNITE- Women fueling science and technology (2014). Access to technology and gender equality. Available in: http://ignite.globalfundforwomen.org/gallery/infog raphic-how-does-access-technology-lead-gender- equality (accessed 1 December, 2016)

ILO (2012). ILO Action Plan for Gender Equality 2010–15. Geneva: ILO Available at: http://www.ilo.org/wcmsp5/groups/public/--- dgreports/--- gender/documents/publication/wcms_179106.pdf (Accessed 1 December 2016)

International Journal of Caring Sciences September-December 2017 Volume 10 | Issue 3| Page 1231

www.internationaljournalofcaringsciences.org

Kogut, B., Colomer, J., & Belinsky, M. (2014). Structural equality at the top of the corporation: mandated quotas for women directors. Strategic Management Journal, 35(6), 891-902.

Loscocco, K., & Bird, S .R. (2012). Gendered Paths Why Women Lag Behind Men in Small Business Success. Work and Occupations, 39(2), 183-219.

Mentoring Women's Network (2015). Mentoring women. Available at: http://mentoringwomensnetwork.com/ (accessed 1 December 2016)

Milgram, D. (2011). How to Recruit Women and Girls to the Science, Technology, Engineering, and Math (STEM) Classroom. Technology and Engineering Teacher, 71(3), 4-11.

Monroe, K., Ozyurt, S., Wrigley, T., & Alexander, A. (2008). Gender equality in academia: Bad news from the trenches, and some possible solutions. Perspectives on politics, 6 (02), 215-233.

Newman C (2014) Time to address gender discrimination and inequality in the health workforce. Human resources for health, 12(1), 25.

Neyer, G., Lappegard, T., & Vignoli, D. (2013a). Gender equality and fertility: Which equality matters? European Journal of Population/Revue européenne de Démographie, 29(3), 245-272.

Neyer, G., Lappegard, T., & Vignoli, D. (2013b). Gender equality and fertility: Which equality matters? European Journal of Population/Revue européenne de Démographie, 29(3), 245-272.

Norman, D.A. & Draper, S.W. (1986). User Centred Systems Design. Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

OECD (2012). Closing the gender gap: Act now. OECD Publishing http://dx.doi.org/10.1787/9789264179370-en

Parmaxi, A., & Vasiliou, C. (2015). Communities of interest for enhancing social creativity: the case of Womenpower platform. Proceedings of INTED2015 conference (pp. 2838-2847). Madrid, Spain.

Ritchie, J., Lewis, J., Nicholls, C.M., & Ormston, R. (2013). Qualitative research practice: A guide for social science students and researchers. London: SAGE.

Ronnblom, M. (2005). Letting women in? Gender mainstreaming in regional policies. Nordic Journal of Women's Studies, 13(3), 164-174.

Smith, K. (2014). The Gender Pay Gap Is Just the Beginning of America's Pay Inequity Problem. Forbes. Available at: http://www.forbes.com/sites/kylesmith/2014/04/10 /the-gender-pay-gap-is-just-the-beginning-of- americas-pay-inequity-problem/ (Accessed 6 May, 2017)

Tijdens, K., De Vries, D. H., & Steinmetz, S. (2013). Health workforce remuneration: comparing wage levels, ranking, and dispersion of 16 occupational groups in 20 countries. Hum Resour Health, 11:11 doi:10.1186/1478-4491-11-11

United Nations news center (2015). Valuing women’s work key to achieving new global goal on gender equality – UN labour agency. United Nations. Available at: http://www.un.org/apps/news/story.asp?NewsID= 52378#.Vl8vqNIrKM8 (Accessed 1 December 2016)

Vives-Cases, C., Gil-Gonzalez, D., Ruiz-Perez, I., Escriba-Aguir, V., Plazaola-Castano, J., Montero- Pinar, M.I., & Torrubiano-Dominguez, J. (2010). Identifying sociodemographic differences in Intimate Partner Violence among immigrant and native women in Spain: A cross-sectional study. Preventive medicine, 51(1), 85-87.

Yang, T., & Aldrich, H.E. (2014). Who’s the boss? Explaining gender inequality in entrepreneurial teams. American Sociological Review, 79, 303- 327

Workplace, Gender Equality Agency, Australian Government (2013). About workplace gender equality. Available at: https://www.wgea.gov.au/learn/about-workplace- gender-equality (Accessed 2 December 2016)

World Health Organization (2002). The world health support 2002: reducing risks promoting healthy life. Geneve: World Health Organization.

World Health Organization (2008). Gender and health workforce statistics: spotlight on statistics: A fact file on health workforce statistics. Genève: World Health Organization.

Zenger, F. (2012). A Study in Leadership: Women do it Better than Men. Available at: http://www.zfco.com/media/articles/ZFCo.WP.W omenBetterThanMen.033012.pdf. (accessed 6 May, 2017)

Copyright of International Journal of Caring Sciences is the property of International Journal of Caring Sciences and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

R E G U L A R A R T I C L E

Can labor provisions in trade agreements promote gender equality? Empirical evidence from Cambodia

Elva L�opez Mourelo | Daniel Samaan

International Labour Organization

Correspondence Daniel Samaan, ILO, Research Department, Route des Morillons 4, CH-1211 Geneva, Switzerland. Email: samaan@ilo.org

Abstract In recent years, an increasing number of regional and bilateral

trade agreements have emerged that include provisions on labor

standards. The claimed purpose of these labor provisions is to

improve working conditions in developing and emerging econo-

mies. However, little is known about whether such provisions

actually do impact working conditions. This paper conducts an

econometric study on the effectiveness of labor provisions in

trade agreements. In particular, we evaluate the impact of the

1999 Bilateral Textile Agreement between Cambodia and the

United States (CUSBTA) on both the gender wage gap and dis-

crimination. The agreement combined the incentive of higher

exports with the obligation of textile manufacturers to comply

with international core labor standards, which include the elimi-

nation of discrimination in respect of employment and occupa-

tion. Using data from the Cambodia Socioeconomic Survey and

applying a difference-in-difference estimation, we find a statisti-

cally significant reduction of the gender wage gap in the textile

sector that can be attributed to the CUSBTA.

1 | INTRODUCTION

Efforts to promote global trade on the multilateral level through the Doha Development Agenda (DDA) have not produced much progress over the last decade. At the same time, the number of regional and bilateral free trade agreements (FTAs) has increased significantly. While the prolifera- tion of FTAs has been well documented (see, for example, Egger and Larch, 2008; Baldwin, 2012; Kohl, Brakman, and Garretsen, 2016; ILO, 2016), less attention has been given to the fact

DOI: 10.1111/rode.12347

404 | © 2017 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/rode Rev Dev Econ. 2018;22:404–433.

that a rising number of North–South agreements contain certain legal clauses on labor standards (so-called “labor provisions”).1

The purpose of the inclusion of labor provisions is usually the claim that minimum standards on labor are needed to share the benefits of globalization with the workforce in developing coun- tries through higher wages and better working conditions. Besides the long-standing theoretical arguments for or against labor provisions in FTAs, and whether they promote development or are hidden forms of protectionism,2 another important question remains unanswered: there exists basi- cally no empirical evidence on whether labor provisions actually have an effect on working condi- tions in developing countries when included in a trade agreement or not.

There are several reasons for this lack of empirical evidence. Firstly, while the theoretical debate about the potential impact of labor provisions is quite old, the actual inclusion of labor pro- visions in trade agreements is a relative new phenomenon. In most of these cases, there is simply not enough data available to conduct such empirical analysis. Secondly, labor provisions in trade agreements are quite diverse and may affect a variety of working conditions in different ways, which, in turn, would make it difficult to generalize their potential impacts on working conditions. Instead, analyses need to be made for specific working conditions that can be measured. Finally, there exist considerable methodological challenges since the causality between labor provisions in an international agreement and working conditions at the micro level is usually difficult to estab- lish.

This paper aims to contribute to closing this gap in the literature and addresses the above-men- tioned issues by selecting the 1999 Cambodia–US Bilateral Textile Agreement (CUSBTA) as an example of a trade agreement with labor provisions, and by looking at gender inequality at the workplace as a specific working condition.

Several reasons make the CUSBTA an especially interesting case study which can shed light on the causal relationship between the implementation of a trade agreement with labor provisions and the actual change in working conditions. The CUSBTA includes a labor provision that explic- itly refers to some identifiable aspects of working conditions. More specifically, Section 10 (B) of the CUSBTA requires that “internationally recognized core labor standards” are implemented. While the International Labour Organization (ILO) is not specifically mentioned, “internationally recognized core labor standards” are generally understood as synonyms for ILO’s core labor stan- dards that are summarized in the 1998 ILO Declaration on Fundamental Principles and Rights at Work. The 1998 Declaration mentions the elimination of discrimination with respect to employ- ment and occupation, of which gender discrimination is an important aspect.

Nevertheless, even when the labor provisions are clear and make reference to identifiable work- ing conditions, the causality chain from including the respective provision in an FTA to actual (measurable) impacts on working conditions is still complex. In particular, in order to expect an impact on the de facto working conditions, the respective labor provision needs to entail some form of legal or institutional changes. This conjunction between a change in norms and institutions and the conditions of workers is a crucial link in the causality chain. The weaker the first link is (i.e. from the FTA containing the provision to the legal and institutional change), the harder it is to establish a causality between the labor provision and changes in working conditions. In this regard, this first link in the causality chain is guaranteed in the case of the CUSBTA as legal and institutional changes in response to the agreement have already been sufficiently documented (see, for example, Polaski, 2006; Fenwick, 2016).3 As described in more detail in Section 2, the annual increase in export quotas was made subject to compliance with certain labor standards by the appa- rel factories and the ILO was responsible for the evaluation of this compliance.

L�OPEZ MOURELO and SAMAAN | 405

While labor law and institutional changes affected all sectors of the Cambodian economy, specific policies were created through the CUSBTA that applied only to the apparel sector. In par- ticular, factory inspections were carried out with the aim of sanctioning violations of core labor standards, including cases of unequal remuneration for work of equal value. Thus, in the case of the CUSBTA, there are reasons to believe that a causal link exists between the conclusion of the agreement and the improvement of working conditions in Cambodian factories, in particular a reduction in the gender wage gap. In this paper, we aim to find empirical evidence for such a link. However, we do not discuss in detail how the causal mechanism works. A very detailed discussion about how labor provisions in trade agreements can potentially translate into improved de facto working conditions and labor rights can be found in Aissi, Peels, and Samaan (2017).

From a methodological point of view, in order to estimate the effect of an FTA with labor pro- visions on working conditions, it is crucial to be able to control for other factors apart from the agreement that could have had an impact on working conditions. The fact that the CUSBTA affected only the textile sector allows for setting up a control group to account for those changes in working conditions that are unrelated to the trade agreement.

With respect to the selection of gender inequality as the outcome of interest, we consider that looking at gender discrimination may provide some understanding about the broader impact of FTAs on the socioeconomic status of women in a society, which, in turn, is often a quite good indicator for a country’s state of development. Another reason for concentrating on the gender effects of labor provisions in FTAs is that gender inequality and gender discrimination (as opposed to many other dimensions of working conditions) can be measured quantitatively through the gen- der wage gap.

Regarding the empirical strategy, this paper takes advantage of both facts: that Cambodia went through different phases of trade liberalization very rapidly; and that labor provisions embedded in the CUSBTA applied only to one sector of the economy (the textile sector). These specific circum- stances allow us to overcome some of the identification challenges related to the estimation of the impact of labor provisions on the gender wage gap. An important methodological concern, how- ever, is that the effects of the CUSBTA may not be orthogonal to observable and unobservable factors that also affect the gender wage gap. We solve this issue through different steps that lead us to believe that the relationship between the signature of the CUSBTA and the improvement in the gender wage gap found in this paper is causal. Using data from the Cambodia Socioeconomic Survey, we apply a difference-in-difference estimation of the adjusted and unadjusted gender wage gap, which allows a comparison of outcomes of the non-textile manufacturing sector and the tex- tile sector before and after the agreement, thereby controlling for those differences in the gender wage gap across sectors that are due to time-invariant characteristics. Our difference-in-differences models find a significant reduction of the gender wage gap in the textile sector that can be attribu- ted to the CUSBTA.

Our analysis builds on existing literature on working conditions in Cambodia, which often employs survey data from the Better Factories Cambodia (BFC) program set up by the ILO and the International Finance Corporation. One element of the BFC program is monitoring of working conditions at the factory level.4 Robertson (2011) uses the Cambodian Socioeconomic Survey to compare movements of wages over time. He finds that wage differentials increased dramatically after the trade agreement, and, as predicted by trade theory, they follow changes in unit values of apparel. He uses factory-level data from BFC to assess the impact of factory audits on working conditions. He shows that working conditions in audited firms have improved. Savchenko and Acevedo (2012) analyse the evolution of female wages in Cambodia and Sri Lanka after the multi- fibre arrangement (MFA) was phased out. They use a similar approach and find a decline of the

406 | L�OPEZ MOURELO and SAMAAN

gender wage gap in both countries in the apparel sector after the MFA. Our contribution to the lit- erature is to identify the effects of labor provisions on gender discrimination by setting up a con- trol group in other manufacturing industries and compare the results with the textile sector.7 We use an extended dataset and present a variety of model modifications to estimate and decompose the gender wage gap in Cambodia over four phases: (i) pre-agreement, (ii) agreement without ILO monitoring, (iii) agreement with monitoring, (iv) post-agreement. We also include different ver- sions of the Blinder–Oaxaca decomposition and show that our findings are robust to model varia- tions. Using the Blinder–Oaxaca decomposition, we can show that the decline of the gender wage gap reflects a decrease in gender discrimination.

The paper is structured as follows. Section 2 provides a brief overview of Cambodia’s experi- ence with trade liberalization and describes the main characteristics of the CUSBTA. Section 3 dis- cusses the data used in the analysis, as well as important descriptive statistics of our sample. In Section 4 we discuss some methodological issues related to the research question and propose an econometric model to address them. We then go on to estimate the adjusted gender wage gap in the textile and other manufacturing sectors in Section 5. In Section 6 we perform a difference-in- difference estimation in order to identify the impact of the CUSBTA on the gender wage gap. The paper ends with a discussion of our findings and some conclusions drawn from the analysis in Section 7. Detailed estimation results and other supplementary information are provided in the Appendices.

2 | TRADE LIBERALIZATION AND THE CAMBODIAN LABOR MARKET

Cambodia’s labor market experience in an environment of growing international trade can be seen as a special case and has the potential to give some instructive insights about the relationship between labor provisions in trade agreements and (de facto) working conditions. What makes Cambodia a special case is the fact that trade liberalization took place in a relatively short period of time and was accompanied by changing policy regimes that affected different economic sectors to various degrees. In the following, we give a brief summary of this evolution and highlight some key developments in the recent past.

Cambodia is currently one of the 48 countries formally designated as least developed countries by the United Nations (UN). Before 1989, Cambodia was run as a socialist economy with little to no free trade existing. In 1989, the government allowed privatization of enterprises and adopted free market measures that boosted international trade. As a result, Cambodia’s economic perfor- mance accelerated in the 1990s after decades of relative poor economic growth, also partly due to civil strife from the 1960s through the 1980s. During the past 20 years, real GDP growth has aver- aged 7.5 percent a year, slightly below the average growth rate of 7.9 percent for developing Asia as a whole. Over the same period, income rose from less than US$300 per capita to roughly US $1,100 in 2015 (IMF, 2013; World Development Indicators).5

Although the role of agriculture has been reduced over the years, it remains the main sector of the Cambodian economy. According to the World Development Indicators database, it accounted for 28.2 percent of GDP in 2015, despite a steep decrease of more than 20 percentage points dur- ing the previous 20 years. This decrease in relative importance of agriculture occurred as a result of the growth of the manufacturing sector, which has increased its share of GDP by 7.5 percentage points over the past two decades – from 9.5 percent in 1995 to 17 percent in 2015. This growth in manufacturing has been led mainly by the development of garment manufactures. Indeed, thanks

L�OPEZ MOURELO and SAMAAN | 407

to an impressive growth rate of 30 percent per year, the garment sector has gone from accounting for 16 percent of gross value added in manufacturing to becoming one of the largest components with estimates of about 65 percent in 2015. It is estimated that about 25 percent of Cambodia’s GDP growth in 2014 stemmed from activities in the textile industry, surpassed only by the con- struction sector (Mejija-Mantilla and Woldemichael, 2017).

The prevalence of manufacturing and, particularly, the textile sector is also observed in terms of employment. According to data from the Cambodian Socioeconomic Survey, manufacturing as a whole accounted for 12 percent of the approximately 8 million people employed in Cambodia in 2012, and more than 60 percent of these manufacturing workers were employed in the textile sec- tor. More recent numbers from ILO, based on data from Cambodia’s Ministry of Commerce, report between 500,000 and 600,000 workers in the garment and footwear factories over the period 2014–2016 (ILO, 2017). It is known that the workforce in the apparel sector is dominated by female workers and generally provides employment opportunities for low-skilled female workers. According to our own estimates, 75–85 percent of the workers in the textile sector are female. Overall employment opportunities for women in Cambodia are generally worse than for men. The Asian Development Bank (2013) finds various gender gaps in several Cambodian labor market indicators including employment rate (7.5 percentage points), vulnerable employment rate (9 per- centage points) and wages. Depending on occupation, female wages can be up to 42 percent lower than male wages. On average, the unadjusted gender wage gap in Cambodia is estimated at around 30 percent for Cambodia’s economy (Asian Development Bank, 2013; World Economic Forum, 2016).

From the 1990s, the share of goods and services exports in GDP increased from 16 percent of GDP in 1993 to 58 percent of GDP in 2012, peaking at 65.5 percent in 2008. While garments and textiles6 constitute the main export sector in Cambodia, the country was a latecomer to apparel exports. In fact, according to data from UNCTAD, apparel production exploded from 17 percent of all exports in 1995 to more than 65 percent in 2010. By 2015 the value of garments had reached US$1.4 billion and represented about 75 percent of Cambodia’s exports (see also Mejija- Mantilla and Woldemichael, 2017).

Cambodia’s trade liberalization process was accompanied by several trade political steps aimed at integrating its industries into the global economy (see Neak and Robertson, 2009). During 1993 and 1994, most trade restrictions still existing in Cambodian domestic law were removed. In 1996, Cambodia obtained Most Favored Nation status from the United States, and joined the Generalized System of Preferences of the European Union (EU GSP). In 1999, Cambodia became a member of the Association of Southeast Asian Nations (ASEAN), which requires the implementation of trade liberalization and tariff reduction toward an integrated economy under the ASEAN Free Trade Area (AFTA). As a member of ASEAN, Cambodia has been involved in the negotiation of several FTAs. In particular, according to the World Trade Organization (WTO), Cambodia is currently engaged in six regional trade agreements (with Australia, ASEAN, China, India, Japan, New Zeal- and and South Korea). Cambodia joined the WTO in October 2004. In this context of liberaliza- tion, special attention should be given to the CUSBTA, concluded in 1999.7 The original agreement was active from January 1999 to December 2001, and was subsequently extended for another three years (2002–2004).

A unique feature of the CUSBTA was to link the annual increase in the export quotas to com- pliance by the Cambodian apparel factories with certain labor standards (see Polaski, 2006). Fur- thermore, a system of monitoring working conditions carried out under ILO supervision was implemented. The monitoring system was established in a separate agreement between the ILO, the Cambodian government and social partners, and was signed in May 2000. Factories in the

408 | L�OPEZ MOURELO and SAMAAN

apparel industry needed to sign up to the monitoring program (the BFC program) in order to obtain an export license from the Cambodian government. The decision on the overall export quota was, in turn, based on the results of this monitoring exercise. On January 1, 2005, the US–Cambo- dia textile agreement and its quota system ended. However, Cambodian manufacturers, the govern- ment, and trade unions had an interest in the continuation of the monitoring program, and the ILO agreed to extend the program for another three years beyond the end of the CUSBTA. Based on the experience with the BFC program, the monitoring was supplemented by training courses and continues to date.8

The CUSBTA meant a significant novelty in terms of the inclusion of labor provisions in the FTAs signed by Cambodia. Indeed, before 1999, the only reference to labor provisions arising from trade agreements stemmed from the EU GSP. At the time Cambodia was accepted in the GSP in 1996, the only commitment that had to be made was to ensure that no practice of any form of forced labor as defined in ILO Conventions 29 on Forced Labour and 105 on Abolition of Force Labour took place. However, ratification of these conventions was not required.

It is important to bear in mind that, in Cambodia, rapid trade liberalization and related trade policy changes were accompanied by also varying legal and institutional settings on the labor market (see Neak and Robertson, 2009; ILO, 2012). While major legal reforms to improve worker rights did not occur until passage of the 1993 Constitution,9 the Cambodian Labor Law of 1999 and related legislation provided the legal framework for all employer–employee relation- ships established by enterprises of more than eight employees. Likewise, several important ILO Conventions were only ratified in the recent past.10 In the context of this paper, the introduction of a minimum wage in Cambodia’s garment and footwear industry in 1997 deserves special mention. At that time the minimum wage was set at US $40 per month, remained unchanged at US $45 per month from mid-2000 to 2007, increased to US $50 per month in January 2007, and again remain unchanged at US $61 for more than two years between October 2010 and May 2013.11

The previous discussion has detailed the policy changes that Cambodia went through after the CUSBTA was concluded, including: (i) legal, regulative and institutional changes; and (ii) a sys- tem of export incentives linked to the compliance with certain labor standards, which was moni- tored at the factory level. Both factors appear to establish the two necessary links in the causality chain between labor provisions and a reduction of the gender wage gap. We examine in the rest of the paper whether empirical evidence for the existence of such a link can be estab- lished in the case of CUSBTA. In addition, also based on the information detailed above, we distinguish six phases of Cambodian trade liberalization, which are outlined in Table 1. In the following, we use data from the Cambodian Socioeconomic Survey (CSES) from different points in time to evaluate whether and how the gender wage gap has improved under different policy settings. Due to data limitations, we can only carry out our analysis for four periods (not all six phases) whereby the pre-agreement period is our baseline to which the next follow-up periods – CUSBTA without ILO monitoring, CUSBTA with ILO monitoring and post-agreement – are compared.

3 | DATA AND DESCRIPTIVE STATISTICS

The analysis draws on a microeconometric model based on the CSES, conducted by the National Institute of Statistics of the Ministry of Planning in the years 1993/94, 1996, 1997, 1999, 2004, and then annually from 2007 to 2012.12 The time coverage of the dataset corresponds to the period

L�OPEZ MOURELO and SAMAAN | 409

from the years just prior to the CUSBTA came into effect in January 1, 1999 to the most recent year with available data. This paper analyses the change in the gender wage gap over the periods identified in Section 2, which include the following: pre-agreement, agreement without ILO moni- toring, agreement with ILO monitoring, and post-agreement. The pre-agreement data are drawn from the datasets collected in 1996 and 1997; data for the agreement period without and with ILO monitoring correspond to 1999 and 2004, respectively; and, finally, the post-agreement data are from the 2007–2012 surveys.

The CSES is a household survey with questions related to households and their members. It provides information on personal characteristics of each individual in the sample (such as gender, age, marital status and place of residence), as well as information about the composition of his or her household and housing conditions. Moreover, the CSES also contains information related to education, such as literacy level, highest grade level successfully completed and school attendance. Finally, the CSES provides information on individual’s labor characteristics such as employment status, occupation, industry, hours worked and monthly earnings.

Therefore, the CSES allows us to assess the evolution of work (and its characteristics) in the Cambodian textile sector, and examine how it compares with other manufacturing sectors in the country. For the purpose of this study, the analysis focuses on individuals aged 15 or over who worked in the manufacturing sector during the week of reference with available information on average hourly earnings. Thus, the selected sample consists of 6,587 workers in the textile sector and 2,657 workers in other manufacturing sectors; 13 percent of the observations correspond to the pre-agreement period, 9.1 percent to the period of agreement without ILO monitoring, 19.2 percent to the period of agreement with ILO monitoring, and 58.6 percent to the post-agreement period.

As described in Section 2, the textile sector in Cambodia expanded significantly from the 1990s and, in particular, since the CUSBTA was concluded in 1999. Indeed, according to data from the CSES, while the textile sector accounted for only 2 percent of total employment in Cam- bodia before the agreement, this percentage increased to more than 8 percent in 2012. However, and despite this expansion in terms of employment, the composition of the textile labor force by gender remained quite stable over time. Figure 1 shows the evolution in the share of female and male employment in the textile sector between 1996 and 2012. Over this period, the percentage of female textile workers ranged between 77 and 85 percent of total employment in the sector. More- over, as the descriptive statistics will later show, the composition of the textile labor force did not change either in terms of other characteristics that might have an impact on wages such as age and education.

TABLE 1 Phases of Cambodian trade liberalization (TL)

Phase Years Description

1 Pre-1989 Socialist economy

2 1989–1995 TL without labor provisions

3 1996–1998 TL with EU GSP

4 1999–2000 TL with CUSBTA without ILO monitoring

5 2001–2004 TL with CUSBTA with ILO monitoring

6 2005–2012 TL with ILO monitoring without quota increases

410 | L�OPEZ MOURELO and SAMAAN

However, despite this stability in the composition of the labor force, the evolution of real wages in the textile sector differed significantly from what other manufacturing sectors witnessed.13 Fig- ure 2 shows real wages of total, male and female workers in the textile sector and in other manu- facturing sectors over the period 1996–2012. We observe that real wages for garment sector workers increased before and immediately following the agreement in 1999, and roughly stagnated thereafter. Moreover, as the garment sector is composed of only approximately 20 percent men, we can see clearly that the behavior of men’s wages over this time is more erratic due to the smal- ler sample size. By contrast, wages in other manufacturing sectors remained stable over the first decade (i.e. between 1996 and 2009) and increased sharply thereafter, driven mainly by the perfor- mance of men’s wages, as wages of female workers decreased slightly until 2009, and only increased since thereafter.

0

20

40

60

80

100

1996 1997 1999 2004 2007 2008 2009 2010 2011 2012

Women Men

FIGURE 1 Evolution in the composition of employment in the textile sector by gender, 1996–2012 [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

1996 1997 1999 2004 2007 2008 2009 2010 2011 2012

Textile sector

Women Men Total

0

200

400

600

800

1,000

1,200

1,400

1,600

1996 1997 1999 2004 2007 2008 2009 2010 2011 2012

Other manufacturing sectors

Women Men Total

FIGURE 2 Real wages in the textile sector and in other manufacturing sectors, 1996–2012 [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

L�OPEZ MOURELO and SAMAAN | 411

With regard to the rest of the characteristics of the workforce, Tables A1 and A2 in Appendix A depict the means and standard deviations of selected variables for workers in the tex- tile and other manufacturing sectors, respectively, by sex and period of study. The differences in wages across sexes and sectors observed in Figure 2 could arise from dissimilarities in the number of hours worked. In this regard, there are no significant differences in the average number of weekly hours worked by male and female workers in the textile sector. By contrast, male workers in other manufacturing sectors work, on average, 5 hours more per week than their female counter- parts – a gap that remains practically unchanged over time.

Workers in the textile sector are, on average, younger than their counterparts in other manufac- turing sectors. Moreover, while the average age of workers in other manufacturing sectors remains relatively constant, a slight decrease is evident in the textile sector. This could reflect the increas- ing preference of textile employers to hire younger workers.

The average education level of women in the sample is somewhat lower than that of men. In the textile sector, the majority of both women and men have secondary education. However, in other manufacturing sectors, women have largely primary education, whereas more than 50 percent of men have completed secondary education. The number of both male and female workers who have completed tertiary education is very low in all manufacturing sectors. Finally, although the gender gap in education has remained higher in other manufacturing sectors than in the textile industry, it has narrowed substantially over time in both segments. Yet, women’s gains in sec- ondary education completion have not yet matched those of men. With regard to occupation, the vast majority of workers in the textile sector are low-skilled, with virtually no differences between men and women. This percentage remains almost unchanged over time. The rest of the manufac- turing sectors as a whole are also dominated by low-skilled workers, but the shares of medium- and high-skilled workers are slightly higher for both genders.

Figure 3 shows the unadjusted gender wage gap14 in textile and other manufacturing sectors over different periods. Before the textile agreement took effect, the gender wage gap in the textile sector was 31 percent. By the end of 1999, after the first year of agreement, the gender wage gap in the textile sector decreased by 13 percentage points. The fall in the gender wage gap continued during the second phase of the agreement, that is, after the monitoring experiment was imple- mented. Indeed, in 2004, the gender wage gap in the Cambodian apparel industry was 13.5 per- cent, 4.5 percentage points lower than in 1999. Since then, the gender wage gap in this industry has continued to decrease but at a slower pace. During the post-agreement period, the gender wage

0

10

20

30

40

50

Pre-agreement Agreement without ILO monitoring

Agreement with ILO monitoring

Post-agreement

Textile Other manufacturing

FIGURE 3 Unadjusted gender wage gap by sector and period (percent) [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

412 | L�OPEZ MOURELO and SAMAAN

gap was, on average, 11.3 percent, 2.2 percentage points lower than by the end of 2004 when the agreement expired.

It is possible that the unadjusted gender wage gap in the textile sector has narrowed for a vari- ety of reasons, including factors that may be unrelated to the trade agreement. However, the pat- tern is completely opposed to that observed in other manufacturing sectors. The unadjusted gender wage gap in these industries increased continuously over the whole period 1996–2012. Therefore, all these statistics suggest that there might have been compliance with the labor standards on gen- der discrimination since the CUSBTA was concluded and, therefore, women might have reaped benefits from this trade agreement, especially once the monitoring system by the ILO was estab- lished.

The descriptive analysis presented in this section suggests that the change in the gender wage gap observed in the textile sector (and its inconsistency with the trend observed in other manufac- turing sectors) is not simply the result of a change in observable characteristics of the textile work- ers, as the composition of the workforce in this sector remains quite stable over the whole period of study. However, changes in the differences in wages between women and men could also reflect the effect of some unobservable factors. In the following section, we present our empirical strategy to isolate the effect of the CUSBTA on the gender wage gap.

4 | EMPIRICAL STRATEGY

The purpose of this study is to identify the average effect of the CUSBTA on the gender wage gap in the industries affected by this agreement (i.e. the average impact of treatment on the trea- ted). If the existence of such an effect could be demonstrated and its magnitude be quantified, we would be in a position to draw some conclusions about the effectiveness of labor provisions in trade agreements.

More specifically, we are interested in comparing the gender wage gap in the textile sector after the agreement was signed in 1999 to the counterfactual, that is, the gender wage gap in the textile sector over the same period had the CUSBTA not existed. However, the counterfactual is never observed and we have to estimate it, bearing in mind that there might be important differences between the textile sector and the rest of the manufacturing sectors that could be correlated with the gender wage gap. For instance, wages can be, on average, lower in the textile sector than in other manufacturing industries. In this case, the correlation between the CUSBTA and the gender wage gap would be confounded with this wage effect. In our particular case, we need to address the fact that we are not in an experimental setting that would allow us to directly control the effect of the CUSBTA and all confounding factors. However, many of these confounding factors are characteristics that vary across sectors but are fixed over time.

We estimate a difference-in-differences (DID) model to control for these time-invariant charac- teristics that might be correlated with both the CUSBTA (or treatment) and the gender wage gap. The DID estimator allows us to compare changes in the gender wage gap of the textile sector before and after the conclusion of the CUSBTA in 1999 to changes in the gender wage gap of the rest of manufacturing sectors over the same period. As described in Section 3, we use available data from the CSES to estimate the gender wage gap in the Cambodian textile sector covered by the CUSBTA (our treatment group) and in the rest of the manufacturing industries (i.e. our control group, which is not subject to the agreement and, therefore, the gender wage gap in this group should not be affected by the CUSBTA). More specifically, while both groups of industries are affected by general trade liberalization, only the textile sector is subject to the ILO monitoring and

L�OPEZ MOURELO and SAMAAN | 413

the incentive of higher export quotas. We chose years 1996 and 1997 as the baseline period and years 1999, 2004 and post 2007 as the follow-up periods.

The average treatment effect on the treated following the DID method (c) would be then expressed as

c ¼ EðyT1 � yT0 j T ¼ 1Þ � EðyC1 � yC0 j T ¼ 0Þ; (1) where yT0 and y

T 1 are the gender wage gap for the textile sector during the pre-agreement and the

post-agreement period, respectively; and yC0 and y C 1 are the gender wage gap for all other manufac-

turing sectors during the same two periods.15

The representation of the DID estimate in (1) allows for an intuitive interpretation: while the first term is the change in the gender wage gap in the textile sector between the post-treatment and the pre-treatment period, which accounts for the variation due to intertemporal characteristics of the textile sector, the second term is the change in the gender wage gap in the control group, which accounts for time variation that is not due to the effect of the CUSBTA (as it is common to both the textile sector and all other manufacturing industries). Thus, by subtracting this second dif- ference from the first one, we obtain an unbiased estimate of the treatment impact.

The DID estimator denoted by c6 and, therefore, the impact of CUSBTA on the gender wage gap, can be determined through an ordinary least-squares estimation of the following equation:

y ¼ xTb þ c0T þ c1M þ c2P þ c3TM þ c4TP þ c5MP þ c6TMP þ �; (2) where y is the logarithm of real wages; x is a column vector of explanatory variables that vary across individuals and time (constant included); M is a binary variable that takes the value 1 for males, T is a binary variable that takes the value 1 for individuals working in the textile sector; P is a dummy variable taking the value 1 for the post-agreement period; and e is a spherical error term. The coefficient c6 of the interaction term between T, M, and P provides the average effect of the agreement. We choose the pre-agreement years 1996 and 1997 as baseline and we successively compare the effects of changing policy regimes in the following three periods with this baseline. We run the regression for each post period, once without control variables and once including some variables that may affect wages (x).

The DID estimator relies on the assumption that in the absence of treatment the outcomes of treated and non-treated units would have changed in the same manner over time (the so-called “common trend assumption”). This assumption can be tested, requiring trends in wages for textile and other manufacturing workers to be similar during the pre-agreement period. Given that the agreement was concluded in 1999, wages in the textile and in other manufacturing sectors should have followed similar trends during the pre-treatment period (in our dataset, this corresponds to the years 1996 and 1997). In order to test the common trend assumption, we have re-estimated the DID model based on equation (2) comparing the years 1996 and 1997. Table B1 in Appendix B presents the results of this exercise, which show that DID estimates on wages during the pre-treat- ment period are close to zero and not statistically significant. These results suggest that the DID estimates on the impact of the CUSBTA presented in the following section are not determined by incorrect identification assumptions.

In this same regard, it is important to note that we cannot completely rule out the effect that the introduction of the minimum wage in the textile industry in 1997 might have had on the wage gap. However, we do not see any indication of a potential impact of the minimum wage on the gender wage gap. Indeed, the share of textile workers earning less than the minimum wage in 1997 was about 70 percent in both groups, men and women; and, as described in

414 | L�OPEZ MOURELO and SAMAAN

Section 3, the relative composition of the workforce has remained stable over the whole period examined.

5 | ESTIMATIONS OF THE ADJUSTED GENDER WAGE GAP

There exists a vast literature about the gender wage gap and about different methods to account for non-discriminating factors that may explain this gap. We would like to emphasize that the main purpose of the paper is to measure impacts of the CUSBTA on the gender wage gap and, there- fore, we are more concerned with this question and less with an accurate measure of the adjusted gender wage gap.

5.1 | The adjusted gender wage gap This section builds on the descriptive analysis shown in Figure 3 but attempts to account for more non-discriminatory factors that explain the gender wage gap. The differences in male and female wages can partly be explained by differences in observable worker characteristics. In order to remove these differences, a standard Mincerian wage equation is estimated:

lnðwiÞ ¼ xTi b þ lia þ �i (3) where i indexes individuals, w is the real average hourly wage,16 and x is a column vector of con- trol variables capturing personal, educational and professional characteristics of the worker. e is an error term assumed to be independent and identically normally distributed with expected value zero. The vector l contains a set of dummy variables which take the value 1 if the individual belongs to a certain group and 0 otherwise. The coefficient vector a then indicates whether or not a premium (that may or may not be due to discrimination) for certain group membership exists. We run the regression for each period and sector separately so that the only dummy under consid- eration is gender, that is, l reduces to a scalar.

The estimation of the model includes the following explanatory variables organized by cate- gories: personal characteristics of individuals include information on gender, age, age squared, location (urban versus rural), and a dummy variable showing if the individual is married; variables related to human capital endowment are also included, in particular, two dummies for the level of education attained (lower and higher education); finally, the number of weekly hours worked and one dummy variable to indicate if the individual is in a low-skilled occupation are introduced.

With the aim of capturing the difference in the gender wage gap across sectors, the model is estimated by each period of time (i.e. pre-agreement, agreement without ILO monitoring, agree- ment with ILO monitoring, and post-agreement) for individuals working in the textile sector and for workers in other manufacturing sectors, separately. Moreover, in order to explore the robust- ness and sensitivity of the results to different sets of covariates, we run the regressions twice: first without covariates, and a second time including as independent variables the set of regressors described above.17 Table 2 summarizes the results of the ordinary least-squares estimates of equa- tion (3). The full estimations, including results on all variables, are presented in Tables C1 and C2 in Appendix C.

The first specification (i.e. first row for each period) includes only the male dummy and, there- fore, provides the unadjusted gender wage gap. While this raw gender wage gap in the textile sec- tor was significantly below that observed in other manufacturing sectors before the CUSBTA, the

L�OPEZ MOURELO and SAMAAN | 415

decrease since the signature of the CUSBTA was such that, in the post-agreement period, the adjusted gender wage gap in the textile sector was six times below that of other manufacturing sec- tors. More specifically, the gender wage gap in the textile sector decreased steeply from 27 percent during the pre-agreement to 8 percent in the post-agreement period. By contrast, the unadjusted gender wage gap in other manufacturing sectors increased over time – by 9 percentage points between the pre-agreement and the post-agreement period.

Adding the variables that control for workers’ personal and professional characteristics decreases the estimated gender wage gap for both textile and other manufacturing workers. Thus, omitting these variables and the effect they have on wages results in upward bias on the estimated gender wage gap. Clearly the size of the gender wage gap depends on the model used, but it is striking that the reduction of the gender wage gap is even larger when other explanatory factors are accounted for. Thus, there is clear evidence of a falling gender wage gap in the textile sector once the CUSBTA was concluded. However, it is noteworthy that virtually no further reduction of the gender wage gap can be observed for the post-agreement period (i.e. for the years 2007–2012 pooled).

Figure 4 compares the adjusted gender wage gap in the textile sector with that observed in other manufacturing sectors over the different periods. Taken together, with all control variables included, the estimated gender wage gap in the textile sector for the most recent period is 7 per- cent, around 34 percentage points lower than in other manufacturing sectors. However, before the CUSBTA, the adjusted gender wage gap in the textile sector was only 4 percentage points lower than that observed in other manufacturing sectors. Thus, while the adjusted gender wage gap in

TABLE 2 Estimated gender wage gap by sector and period

Textile sector Other manufacturing sectors

Pre-agreement without covariates 0.27*** 0.37***

(0.10) (0.08)

with covariates 0.30*** 0.34***

(0.09) (0.09)

Agreement without ILO monitoring without covariates 0.15* 0.40***

(0.09) (0.12)

with covariates 0.13* 0.48***

(0.08) (0.12)

Agreement with ILO monitoring without covariates 0.11*** 0.56***

(0.04) (0.11)

with covariates 0.08** 0.46***

(0.04) (0.12)

Post-agreement without covariates 0.08*** 0.46***

(0.02) (0.05)

with covariates 0.07*** 0.41***

(0.02) (0.05)

Notes: Table reports the least-squares estimates of equation (3). Standard errors are in parentheses. Significance levels: *significant at 10%; **significant at 5%; *** significant at 1%. The full estimations, including results on all variables, for the textile sector and other manufacturing sectors are respectively presented in Tables C1 and C2. Source: Authors’ calculations based on CSES.

416 | L�OPEZ MOURELO and SAMAAN

the textile sector decreased steeply during the period the agreement was in place, it increased in other manufacturing sectors from 34 percent in the pre-agreement period to 41 percent in the post- agreement phase. These results suggest that the CUSBTA had a significant positive effect on reducing the gender wage gap, something we explore further in Section 6.

5.2 | The Blinder–Oaxaca decomposition Capturing the gender effect on wages with a dummy li, as we did in equation (3), has some draw- backs. The coefficient of the gender dummy provides us with an estimate of the average wage pre- mium for being a man. Note that this regression imposes the same beta vector on both groups, men and women. The returns on the endowments x may, however, be different for men and women as a consequence of discrimination. Blinder (1973) and Oaxaca (1973) proposed decom- posing the average wage difference between the two groups into an explained and an unexplained part (Blinder–Oaxaca decomposition).

The idea is to run the regression in equation (3) separately for men and women, without the vector l, and then determine the average wage difference that is due to gender discrimination. We can follow this approach for each sector and each time period. In order to keep our notation sim- ple, we disregard time and sector indices for the time being and only differentiate group member- ship G between male and female workers by G = {M,F}. Hence, each of the following equations applies to the four different time periods and to the textile and other manufacturing sectors, respec- tively. Let bG denote the estimate of coefficient vector b for group G and let �xG be the sample average. Then for each sector and time period the average wage difference can be written as

lnð�yMÞ � lnð�yFÞ ¼ �xTMbM � �xTFbF: (4) Note that we took the difference between males and females so that in the case of discrimina-

tion toward women we would expect a positive number in equation (4) (a premium for being a man). Adding �xTFbM � �xTFbM on the right-hand side and rearranging terms yields the Blinder–Oax- aca decomposition:

0

10

20

30

40

50

Pre-agreement Agreement without ILO monitoring

Agreement with ILO monitoring

Post-agreement

Textile Other manufacturing

FIGURE 4 Adjusted gender wage gap by sector and by period (percentage) [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

L�OPEZ MOURELO and SAMAAN | 417

lnð�yMÞ � lnð�yFÞ ¼ ð�xM � �xFÞTbM |fflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflffl}

explained

þ�xTFðbM � bFÞ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl}

unexplained

: (5)

The first term on the right-hand side ð�xM � �xFÞTbM represents the component of the wage dif- ference that can be explained through differing “endowments” of the two groups, while the second term �xTFðbM � bFÞ represents differences that are unexplained (differences in the coefficients) and that may stem from discrimination.

The Blinder–Oaxaca decomposition is not unique, and by adding �xTMbF � �xTMbF to the right- hand side of (4) we obtain another version of this decomposition. This time the explained differ- ence is calculated by evaluating the average differences in the endowments with the estimated coefficient of females while the unexplained difference between the coefficient vectors is weighted with the group averages of the male group:

0 .1

.2 .3

1996-1997 1999 2004 2007-2012

The chart shows components of the first version of the Blinder-Oaxaca decomposition (D1)

Blinder-Oaxaca Decomposition Gender Wage Gap - Textile Sector

Total Discrimination

FIGURE 5 Gender wage gap and discrimination in the textile sector (percentage) [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

0 .2

.4 .6

1996-1997 1999 2004 2007-2012

The chart shows components of the first version of the Blinder-Oaxaca decomposition (D1)

Blinder-Oaxaca Decomposition Gender Wage Gap - Other Manufacturing Sector

Total Discrimination

FIGURE 6 Gender wage gap and discrimination in the other manufacturing sector (percentage) [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors’ calculations based on CSES.

418 | L�OPEZ MOURELO and SAMAAN

�lnðyMÞ � �lnðyFÞ ¼ ð�xM � �xFÞTbF |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

explained

þ�xTMðbM � bFÞ |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

unexplained

: (6)

In fact, several more decompositions exist, and there is often no way to determine which one is best. Elder, Goddeeris, and Haider (2010) discuss different methods to determine group differ- ences, including the above decompositions and our method that uses a gender dummy (see also Jann, 2008). By adding both terms, �xTFbM � �xTFbM and �xMTbF � �xMTbF, and the term �xTFbF � �xTFbF to the right-hand side of equation (4) we arrive at a more general decomposition (see Winsborough and Dickinson, 1971). The average difference D is then18

D ¼ �xTMbM � �xTFbF þ �xTFbM � �xTMbM þ �xTMbF � �xTMbF � �xTFbF � �xTFbF (7) or

D ¼ ð�xM � �xFÞTbF |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

endowments

þ�xTFðbM � bFÞ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl}

coefficients

þ ð�xM � �xFÞTðbM � bFÞ |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

interactions

; (8)

D ¼ E þ C þ I: (9) Comparing the more general decomposition in (8) and (9) with the two Blinder–Oaxaca decom-

positions (5) and (6) shows that, in the first variant (5), the endowment effect and the interaction effect form the explained part of the difference,

D1 ¼ E þ I |fflffl{zfflffl}

explained

þ C |{z}

unexplained

(10)

while the second variant considers the endowment effect the “explained” part and the interaction effect and the coefficients effect form the “unexplained” part,

D2 ¼ E|{z} explained

þ I þ C |fflffl{zfflffl}

unexplained

: (11)

Hence, an important difference between the two decompositions is whether the interaction effect is part of the explained or unexplained summand, which has implications for the measure- ment of discrimination.

Next, we present the empirical results of the previously discussed Oaxaca decompositions D1 and D2. We interpret the unexplained difference of the Blinder–Oaxaca decomposition in equa- tions (5) and (6) as the extent of discrimination based on wages, knowing that parts of this dif- ference might also arise due to unobserved factors. Figures 5 and 6 show the discrimination in the textile and other manufacturing sectors according to Blinder–Oaxaca (D1). We can see that the largest share of the difference of the average wages between men and women in both sectors can be attributed to discrimination and not to differences in endowments. Furthermore, discrimi- nation against women in the textile sector declines considerable between the pre-agreement per- iod (1996/1997) and the period with ILO monitoring in place (2004), and increases only slightly in the post-agreement period. In other manufacturing sectors, discrimination remains practically unchanged over the entire time period, with only a minor increase in the period of agreement without ILO monitoring. The alternative decomposition (D2) also shows a reduction in discrimi- nation against women in the textile sector, but with the largest impact in the first period (see

L�OPEZ MOURELO and SAMAAN | 419

Appendix D). Further calculations of D did not alter our results and can be obtained upon request.

The Blinder–Oaxaca decomposition confirms and expands the results of the previous sections, namely a decline of the gender wage gap and discrimination against women in the textile sector after the CUSBTA became effective. In particular, a further decline is observed after the ILO mon- itoring took place, while little change is noticed for the post-agreement period. In addition, practi- cally no change in the gender wage gap or in discrimination can be determined in the control group.

6 | ESTIMATES OF THE IMPACT OF THE CUSBTA ON THE GENDER WAGE GAP

This section presents the DID estimation results of the impact of the CUSBTA on the gender wage gap. Table 3 displays the least-squares estimates of equation (2). As detailed in Section 4, we choose the pre-agreement years 1996 and 1997 as baseline and we successively compare the effects of changing policy regimes in the following three periods with this baseline. Therefore, we estimate equation (2) for each period after the signature of the CUSBTA in 1999. Thus, for each period, the first column reports a basic specification without covariates; the second column intro- duces the set of covariates (x) included in equation (3); and the third column includes a variable on exports as percentage of the gross value added of the sector in order to control for the potential impact of the export incentives created by the CUSBTA.

The DID estimate is the coefficient of the dummy variable that takes the value 1 if the individ- ual is a male worker in the textile sector after the policy change occurred (Textile 9 Male 9 Post interaction). Its interpretation is the average wage increase (decrease) for being a male in the textile sector after the policy intervention, relative to the wage of a female worker in other manufacturing sector before the policy intervention.

The results of this exercise confirm the findings of the previous section. The DID estimate is negative for all estimations, indicating a decrease of the gender wage gap, but significant only for periods of agreement with ILO monitoring and post-agreement. Thus, the impact of the CUSBTA on the gender wage gap appears strong after the monitoring started, as the gender effect on wage decreased by 37 percent during the period the agreement was in place and the BFC program was operating. The effect decreases to 31 percent for the period after the CUSBTA ended but the BFC program carried on. The evidence for an effect of the agreement without monitoring is somewhat mixed. While the DID estimate is also negative, it is not significant at the 10 percent level.

Overall, we conclude that the DID estimations support the results of Section 5 where we com- pared the gender wage gap by sector over time (shown in Figure 4): the CUSBTA agreement has led to a decrease in the gender wage gap in the textile sector whereby the ILO monitoring in the context of the BFC program played an important role. Indeed, the impact is stronger once the BFC program with monitoring is in effect. Importantly, including a variable on exports as percentage of the gross value added of the sector does not change at all the outcomes of our estimations, sug- gesting that the export incentives embedded in the CUSBTA are not affecting our results. There- fore, the effectiveness of the CUSBTA in reducing the gender wage gap in the Cambodian textile industry can be confidently attributed to the labor provisions included in the agreement and, more specifically, to the mechanism of monitoring.

420 | L�OPEZ MOURELO and SAMAAN

T A B L E

3 D if fe re nc es -i n- di ff er en ce s es ti m at es

A gr ee m en t w it h ou

t IL

O m on

it or in g

A gr ee m en t w it h IL

O m on

it or in g

P os t- ag

re em

en t

(i )

(i i)

(i ii)

(i )

(i i)

(i ii)

(i )

(i i)

(i ii)

T ex ti le

0. 00

0. 01

0. 30

0. 00

�0 .0 1

0. 25

0. 00

0. 01

�0 .0 1

(0 .0 8)

(0 .0 7)

(0 .2 3)

(0 .0 7)

(0 .0 7)

(0 .2 2)

(0 .0 6)

(0 .0 5)

(0 .0 6)

M al e

0. 37

** *

0. 35

** *

0. 35

** *

0. 37

** *

0. 37

** *

0. 36

** *

0. 37

** *

0. 37

** *

0. 37

** *

(0 .0 7)

(0 .0 7)

(0 .0 7)

(0 .0 7)

(0 .0 7)

(0 .0 7)

(0 .0 5)

(0 .0 5)

(0 .0 5)

P os t

0. 22

** 0. 25

** 0. 22

** �0

.3 4*

** �0

.1 6*

0. 15

0. 14

** 0. 24

** *

0. 20

** *

(0 .1 1)

(0 .1 1)

(0 .1 1)

(0 .1 0)

(0 .0 9)

(0 .2 7)

(0 .0 6)

(0 .0 6)

(0 .0 6)

T ex ti le

9 M al e

�0 .0 9

�0 .0 8

�0 .0 9

�0 .0 9

�0 .0 7

�0 .0 7

�0 .0 9

�0 .0 6

�0 .0 6

(0 .1 2)

(0 .1 2)

(0 .1 2)

(0 .1 1)

(0 .1 1)

(0 .1 1)

(0 .0 9)

(0 .0 8)

(0 .0 8)

T ex ti le

9 P os t

0. 23

* 0. 33

** *

0. 73

** 0. 97

** *

0. 94

** *

1. 05

** *

0. 43

** *

0. 40

** *

0. 37

** *

(0 .1 3)

(0 .1 2)

(0 .3 2)

(0 .1 1)

(0 .1 0)

(0 .1 4)

(0 .0 7)

(0 .0 6)

(0 .0 7)

M al e 9

P os t

0. 03

0. 02

0. 02

0. 20

* 0. 14

0. 14

0. 09

0. 06

0. 06

(0 .1 3)

(0 .1 3)

(0 .1 3)

(0 .1 1)

(0 .1 1)

(0 .1 1)

(0 .0 7)

(0 .0 6)

(0 .0 6)

T ex ti le

9 M al e 9

P os t

�0 .1 5

�0 .1 9

�0 .1 8

�0 .3 6*

* �0

.3 7*

* �0

.3 7*

* �0

.2 8*

** �0

.3 1*

** �0

.3 1*

**

(0 .1 9)

(0 .1 8)

(0 .1 8)

(0 .1 5)

(0 .1 5)

(0 .1 5)

(0 .1 0)

(0 .1 0)

(0 .1 0)

A ge

0. 03

** *

0. 03

** *

0. 02

** *

0. 02

** *

0. 03

** *

0. 03

** *

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 0)

(0 .0 0)

A ge

sq ua re d

�0 .0 4*

** �0

.0 4*

** �0

.0 3*

** �0

.0 3*

** �0

.0 4*

** �0

.0 4*

**

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 0)

(0 .0 0)

U rb an

0. 29

** *

0. 30

** *

0. 27

** *

0. 28

** *

0. 19

** *

0. 19

** *

(0 .0 4)

(0 .0 4)

(0 .0 3)

(0 .0 3)

(0 .0 2)

(0 .0 2)

B ei ng

m ar ri ed

0. 05

0. 05

0. 02

0. 02

�0 .0 0

�0 .0 0

(0 .0 4)

(0 .0 4)

(0 .0 4)

(0 .0 4)

(0 .0 2)

(0 .0 2)

(C on

ti nu

es )

L�OPEZ MOURELO and SAMAAN | 421

T A B L E

3 (C on

ti nu

ed )

A gr ee m en t w it h ou

t IL

O m on

it or in g

A gr ee m en t w it h IL

O m on

it or in g

P os t- ag

re em

en t

(i )

(i i)

(i ii)

(i )

(i i)

(i ii)

(i )

(i i)

(i ii)

H ig he r ed uc at io n

0. 19

** *

0. 19

** *

0. 11

** *

0. 11

** *

0. 10

** *

0. 09

** *

(0 .0 4)

(0 .0 4)

(0 .0 3)

(0 .0 3)

(0 .0 2)

(0 .0 2)

L ow

-s ki ll ed

0. 16

* 0. 16

* 0. 19

** *

0. 18

** �0

.0 0

�0 .0 0

(0 .0 9)

(0 .0 9)

(0 .0 7)

(0 .0 7)

(0 .0 5)

(0 .0 5)

H ou

rs w or ke d

�0 .0 1*

** �0

.0 1*

** �0

.0 1*

** �0

.0 1*

** �0

.0 1*

** �0

.0 1*

**

(0 .0 0)

(0 .0 0)

(0 .0 0)

(0 .0 0)

(0 .0 0)

(0 .0 0)

E xp

or ts

�1 .5 2

�1 .3 1

0. 13

(1 .1 4)

(1 .0 6)

(0 .1 0)

C on

st an t

6. 29

** *

6. 00

** *

6. 10

** *

6. 29

** *

6. 15

** *

6. 24

** *

6. 29

** *

6. 09

** *

6. 09

** *

(0 .0 6)

(0 .1 7)

(0 .1 9)

(0 .0 6)

(0 .1 5)

(0 .1 6)

(0 .0 5)

(0 .0 9)

(0 .0 9)

O bs er va ti on

s 2, 04

7 2, 04

7 2, 04

7 2, 98

2 2, 98

2 2, 98

2 6, 62

2 6, 62

2 6, 62

2

R -s qu

ar ed

0. 06

0. 16

0. 16

0. 12

0. 18

0. 18

0. 09

0. 15

0. 15

N ot es : T ab le

re po rt s th e le as t- sq ua re s es ti m at es

of eq ua ti on

(2 ). S ta nd ar d er ro rs

ar e in

pa re nt he se s. S ig ni fi ca nc e le ve ls : *s ig ni fi ca nt

at 10 % ; ** si gn if ic an t at

5% ; ** * si gn if ic an t at

1% .

So ur ce : A ut ho rs ’ ca lc ul at io ns

ba se d on

C S E S .

422 | L�OPEZ MOURELO and SAMAAN

7 | CONCLUSIONS

In this paper, we have estimated the effect of the labor provisions embedded in the bilateral textile agreement between the United States and Cambodia on the gender wage gap in the textile sector by comparing changes in this gap with those observed in other manufacturing sectors before, dur- ing, and after the agreement. Using data from the Cambodian Socioeconomic Survey and through different methods, we consistently find a decline of the gender wage gap, including discrimination, in the textile sector after the conclusion of the agreement, while the gap in other manufacturing sectors continuously increased over the same period. Our results suggest that the agreement, during both the phase where only the incentive of higher export quotas was in place and the period where the ILO monitoring through the BFC program was implemented, had a positive effect on the reduction of the gender wage gap until its termination in 2004. In the post-agreement period, dur- ing which the BFC program continued to operate and during which no further export incentives by the United States were provided, it appears that the previously achieved reduction of the gap was maintained but little additional impact can be detected. This holds true for the estimations employ- ing the gender dummy as well as the estimations using the Blinder–Oaxaca decomposition.

In order to identify the causal impact of the agreement on the gender wage gap, we perform a difference-in-difference estimation using the other manufacturing sectors as control group. Our results point in the same direction as before, and therefore we find that the CUSBTA led to a decrease in the gender wage gap in the textile sector, which is only statistically significant for the last two periods (during the period when export incentives were combined with ILO monitoring at the factory level, and during the post-agreement period). Moreover, when comparing changes in the gender wage gap in these two last phases, we observe that the impact is higher during the per- iod of agreement with ILO monitoring and the effect decreases slightly during the post-agreement period. Finally, we control for possible differences in exports between the two sectors and find that exports themselves cannot explain reductions in the gender wage gap.

In summary, we find evidence that the CUSBTA with its specific implementation mechanism had a significant effect on the reduction of the gender wage gap and on the reduction of wage dis- crimination in the Cambodian textile industry. A comparison between periods with and without monitoring substantiates the importance of monitoring. Some ambiguity about the exact causes of the effects remains. For example, it is possible that some effects only occur with a time lag. Insti- tutional changes and raised awareness of gender discrimination, or companies learning how to comply effectively with labor regulations, may take a few years before a measurable effect on wages can be determined. In consequence, we may have dedicated them to monitoring, whose start and ending period can be empirically observed. Similarly, the anticipation of future monitoring could also be a reason for a reduction of the gender wage gap even before the monitoring starts.

For these reasons, it remains challenging to make a clear conclusion as to the extent to which the actual monitoring, learning effects, raised awareness, or anticipation of monitoring reduced the gender wage. The exact mechanism through which the gender wage gap was reduced in the Cam- bodian textile sector after the agreement remains a question for future research. We can only deter- mine that the CUSBTA as a whole, including the monitoring dimension, had an important impact on reducing the gender wage gap in textile factories. Another open question is how far our results also apply to dimensions other than gender, for example child labor or freedom of association, that are also specifically covered by the agreement. Finally, we can only speculate as to how far our results can be generalized to other countries, and other sectors with comparable setups.

From a policy perspective, we consider the CUSBTA and the connected BFC program effective instruments to link trade to better working conditions, and possibly to better ensure that the

L�OPEZ MOURELO and SAMAAN | 423

benefits from trade are shared more equally. The use of similar agreements in other countries can therefore be a viable policy option. Nevertheless, the administrative cost and sustainability, espe- cially with regard to the monitoring mechanism, must not be underestimated. Cambodia is a rela- tive small country in which the textile industry is geographically concentrated in one region. The textile sector is a light, labor-intensive manufacturing industry where workers are also typically geographically concentrated in the same location. In countries and in sectors where similar condi- tions can be found, agreements like CUSBTA can be effective, provided such policies are sup- ported by all stakeholders.

ACKNOWLEDGMENTS

We would like to thank Uma Rani Amara, Colin Fenwick, David Kucera, Sandra Polaski (all ILO), and Anil Verma (University of Toronto), and all members of the advisory commitee of the joint Swiss-Canadian-ILO project “Improving worker rights in globalizing economies” for helpful comments. We would also like to thank the editor and two anonymous reviewers for their com- ments and suggestions. Neither the donors of the project nor the ILO had any role in the study design, or the collection, analysis or interpretation of data, or the the writing of this article, or the decision to submit the article for publication. Any view expressed or conclusions drawn represent the views of the authors and do not necessarily represent ILO views or ILO policy. The views expressed herein should be attributed to the authors and not to the ILO, its management or its con- stituents.

ENDNOTES

1 Labor provisions in trade agreements can be broadly defined as “(i) any labour standard which establishes mini- mum working conditions, terms of employment or worker rights, (ii) any norm on the protection provided to work- ers under national labour law and its enforcement, as well as (iii) any framework for cooperation in or monitoring of these issues” (ILO, 2013).

2 See, for example, Elliot and Freeman (2003) for an overview of the arguments. 3 One could also argue that such legal and institutional changes in response to the agreement are another form of effectiveness.

4 For a detailed description of the BFC and the Better Work Program in general see, for example, Kotikula, Pournik, and Robertson (2015).

5 http://databank.worldbank.org. 6 Unless otherwise stated, we do not specifically distinguish between apparel, garments and textiles. In our empirical analysis we combine all related industries and refer to them as the “textile sector.”

7 See Polaski (2009) for a more detailed description of this trade agreement. 8 More information on BFC operations can be found at http://betterfactories.org/. 9 This law recognizes human rights, including, among others, the following: the right to choose employment; the right to equal pay for equal work; the right to form and join unions; the right to strike and to hold non-violent demonstrations; and the abolition of all forms of discrimination against women.

10 In August 1999, Cambodia ratified six of the eight ILO Fundamental Conventions, namely: Convention No. 105 (Abolition of Forced Labour), No. 87 (Freedom of Association and Protection of the Right to Organise), No. 98 (Right to Organise and Collective Bargaining), No. 100 (Equal Remuneration), No. 111 (Discrimination), and No. 138 (Minimum Age). Fundamental Convention No. 29 on Forced Labour had already been ratified in 1969; and Fundamental Convention No. 182 on Worst Forms of Child Labour was ratified in March 2006.

11 Since then, the minimum wage in Cambodia’s garment and footwear industry has increased continuously to reach US $140 per month in January 2016.

424 | L�OPEZ MOURELO and SAMAAN

12 We currently lack the metadata for the 1993/94 survey and cannot use these data in the analysis. 13 The wage data refer to average monthly wages in riels. Wages were deflated using Consumer Price Index pub- lished by the IMF. In some years, wages were reported as a single quantity; in other years, they were broken down by primary and secondary job. In order to create comparable wage data across years, primary and secondary wages and earnings (in both cash and kind) were summed to arrive at a total income figure. Because secondary wages were small relative to primary wages, total wages were grouped according to the nature of the primary job only.

14 The gender wage gap is defined as the difference between the gross average earnings of male and female employ- ees expressed as percentage of gross average hourly earnings of male employees.

15 Note that we could not use the sample averages of y in equation (1) as the adjusted gender wage gap is already itself an average (difference). By applying DID, we are effectively calculating the double differences of this aver- age pay difference in the treatment group and the control group before and after the treatment took place. Hence, we are actually dealing with wages in six groups: men versus women, textile sector versus other manufacturing, and pre-treatment-period versus post-treatment-period. We are interested in the female wage in the textile sector after the CUSBTA relative to the wage in the other five groups. So technically, we are already in the case of triple differences.

16 We use the real average hourly wage as dependent variable to account for the impact on wages of potential differ- ences between women and men in the number of hours worked. Some empirical evidence shows that the gender wage gap might be partly driven by the fact that men are more prone to work overtime hours and the rising hourly wage returns to work (Cha and Weeden, 2014). We also add a covariate on hours worked in all regressions to control for this potential overwork effect. Our estimation results show that an increase in the number of hours worked has a negative and statistically significant effect on average wages.

17 In order to control for a potential selectivity bias in wage equations, we have run all our regressions using the Heckman selection model, which allowed us to net out factors affecting the probability of having a wage, namely gender, age, marital status, education, location, and the number of children in the household. We could carry out this procedure to estimate the adjusted gender wage gap in the textile sector for all periods and in the control group for the post-agreement period, that is, when the sector–period combination allows for a sufficient number of observations. However, when the Heckman selection model was run for other manufacturing sectors in the pre- agreement and agreement periods, the maximum likelihood estimates could not converge due to the small sample. Our results did not change significantly in those cases where we could actually control for selectivity bias (includ- ing the DID regression), which suggests that our estimations are not biased by sample selection. These estimations are available upon request.

18 Note that an alternative would be to weight the differences using the male coefficients bm and means �xM, which would then lead to D ¼ ð�xM � �xMÞTbF þ �xTFðbM � bFÞ � ð�xM � �xFÞTðbM � bFÞ.

REFERENCES

Aissi, J., Peels, R., & Samaan, D. (2017). Evaluating the effectiveness of labour provisions in trade agreements – an analytical framework. International Labour Review. DOI: 10.1111/ilr. 12066.

Asian Development Bank (2013). Gender equality in the labor market in Cambodia. Manila: Asian Development Bank.

Baldwin, R. (2012). Are free trade agreements contagious?. Journal of International Economics, 88, 1–16. Blinder, A. (1973). Wage discrimination: Reduced form and structural estimates. Journal of Human Resources, 8,

436–455. Cha, Y., & Weeden, K. A. (2014). Overwork and the slow convergence in the gender gap in wages. American Soci-

ological Review, 79, 457–484. Egger, P., & Larch, M. (2008). Interdependent preferential trade agreement memberships: An empirical analysis.

Journal of International Economics, 76, 384–399. Elder, T., Goddeeris J., & Haider S. (2010). Unexplained gaps and Oaxaca-Blinder decompositions. Labour Eco-

nomics, 17, 284–290. Elliot, K. A., & Freeman, R. (2003). Can labor standards improve under globalization?. Washington, DC: Institute

for International Economics.

L�OPEZ MOURELO and SAMAAN | 425

Fenwick, C. (2016). The ILO and national labour law reform: Six case studies. In S. Marshall & C. Fenwick (Eds.), Labour regulation and development (pp. 235–287). Geneva: International Labour Office.

ILO (2012). Decent work country profile Cambodia. Geneva: International Labour Organization. ILO (2013). Social dimensions of free trade agreements. Geneva: International Labour Organization. ILO (2016). Assessment of labour provisions in trade and investment agreements. Geneva: International Labour

Organization. ILO (2017). What explains strong and weak employment figures in the Cambodian garment sector?. Cambodian

Garment and Footwear Sector Bulletin, 6, International Labour Organization. IMF (2013). World economic outlook: Transitions and tensions. Washington, DC: International Monetary Fund. Jann, B. (2008). The Blinder-Oaxaca decomposition for linear regression models. Stata Journal, 8, 453–479. Kohl, T., Brakman, S., Garretsen, H. (2016). Do trade agreements stimulate international trade differently? Evidence

from 296 trade agreements. World Economy, 39, 97–131. Kotikula, A., Pournik, M., & Robertson, R. (2015). Interwoven: How the Better Work Program improves job and

life quality in the apparel sector. Washington, DC: World Bank. Mejija-Mantilla, C., & Woldemichael, M. T. (2017). To sew or not to sew? Assessing the welfare effects of the gar-

ment industry in Cambodia. Policy Research Working Paper No. 8061, Washington, DC: World Bank. Neak, S., & Robertson, R. (2009). Globalization and working conditions: Evidence from Cambodia. In R. Robert-

son, D. Brown, G. Pierre & M. L. Sanchez-Puerta (Eds.), Globalization, wages and the quality of jobs (pp. 97– 119). Washington, DC: World Bank.

Oaxaca, R. (1973). Male-female wage differentials in urban labor markets. International Economic Review, 14, 693– 709.

Polaski, S. (2006). Combining global and local forces: The case of labour rights in Cambodia. World Development, 34, 919–932.

Polaski, S. (2009). Harnessing global forces to create decent work in Cambodia. Geneva: International Labour Organization.

Robertson, R. (2011). Apparel wages before and after Better Factories Cambodia. Geneva: International Labour Organization.

Savchenko, Y., & Acevedo, G. L. (2012). Female wages in the apparel industry Post-MFA. Policy Research Work- ing Paper No. 6061, Washington, DC: World Bank.

Winsborough, H., & Dickinson, P. (1971). Components of Negro-white income differences. In Proceedings of the Social Statistics Section (pp. 6–8). Washington, DC: American Statistical Association.

World Economic Forum (2016). The global gender gap report. Geneva: World Economic Forum.

How to cite this article: L�opez Mourelo E, Samaan D. Can labor provisions in trade agreements promote gender equality? Empirical evidence from Cambodia. Rev Dev Econ. 2018;22:404–433. https://doi.org/10.1111/rode.12347

426 | L�OPEZ MOURELO and SAMAAN

A P P E N D IX

A : S U M

M A R Y

S T A T IS

T IC

S

T A B L E

A 1

S am

pl e ch ar ac te ri st ic s fo r ob

se rv at io ns

in th e te xt il e se ct or

P re -a gr ee m en t 19

96 – 19

97 A gr ee m en t w it h ou

t IL

O m on

it or in g 19

99 A gr ee m en t w it h IL

O m on

i- to ri n g 20

04 P os t- ag

re em

en t 20

07 – 20

12

W om

en M en

W om

en M en

W om

en M en

W om

en M en

n = 40

0 n = 99

n = 47

1 n = 98

n = 11

24 n = 27

2 n = 34

28 n = 71

1

A ve ra ge

m on

th ly

w ag es

13 7, 17

0. 7

19 8, 88

0. 5

21 8, 52

8. 9

26 6, 73

6. 3

25 8, 00

1. 3

29 8, 15

5. 6

24 2, 03

9. 0

27 2, 90

9. 9

(1 50

,6 67

.4 )

(3 35

,5 77

.6 )

(1 24

,0 31

.9 )

(1 63

,4 30

.0 )

(1 17

,8 73

.3 )

(2 08

,1 69

.9 )

(1 41

,5 09

.9 )

(1 77

,1 95

.7 )

A ve ra ge

w ee kl y ho

ur s w or ke d

44 .6 0

41 .0 5

52 .4 9

53 .3 0

54 .3 0

53 .3 7

54 .6 7

55 .1 4

(1 5. 38

) (1 6. 98

) (1 0. 45

) (1 2. 93

) (9 .9 7)

(1 0. 16

) (9 .6 9)

(9 .1 6)

A ge

33 .1 7

38 .6 7

27 .1 2

29 .9 3

23 .7 3

25 .6 1

25 .2 8

25 .2 6

(1 3. 53

) (1 7. 77

) (1 0. 99

) (1 3. 95

) (7 .5 0)

(8 .8 6)

(8 .1 1)

(7 .6 9)

A ge

sq ua re d

12 .8 3

18 .0 8

8. 56

10 .8 8

6. 19

7. 34

7. 05

6. 97

(1 0. 72

) (1 6. 61

) (7 .7 9)

(1 1. 88

) (5 .0 2)

(6 .5 9)

(5 .5 0)

(5 .3 9)

L oc at io n in

ur ba n ar ea s

0. 42

0. 32

0. 38

0. 46

0. 19

0. 22

0. 34

0. 36

(0 .4 9)

(0 .4 7)

(0 .4 9)

(0 .5 0)

(0 .3 9)

(0 .4 2)

(0 .4 8)

(0 .4 9)

M ar ri ed

0. 38

0. 65

0. 31

0. 43

0. 22

0. 33

0. 29

0. 36

(0 .4 8)

(0 .4 8)

(0 .4 6)

(0 .5 0)

(0 .4 1)

(0 .4 7)

(0 .4 5)

(0 .4 8)

P ri m ar y ed uc at io n

0. 41

0. 37

0. 34

0. 17

0. 39

0. 25

0. 32

0. 21

(0 .4 9)

(0 .4 9)

(0 .4 8)

(0 .3 8)

(0 .4 9)

(0 .4 3)

(0 .4 7)

(0 .4 1)

S ec on

da ry

ed uc at io n

0. 40

0. 40

0. 52

0. 74

0. 53

0. 69

0. 63

0. 75

(0 .4 9)

(0 .4 9)

(0 .5 0)

(0 .4 4)

(0 .5 0)

(0 .4 6)

(0 .4 8)

(0 .4 4)

T er ti ar y ed uc at io n

0. 01

0. 00

0. 00

0. 01

0. 00

0. 02

0. 00

0. 02

(0 .0 7)

(0 .0 0)

(0 .0 0)

(0 .1 0)

(0 .0 3)

(0 .1 3)

(0 .0 5)

(0 .1 3)

L ow

-s ki ll ed

oc cu pa ti on

0. 97

0. 98

0. 99

0. 96

0. 99

0. 98

0. 99

0. 99

(0 .1 8)

(0 .1 4)

(0 .0 8)

(0 .2 0)

(0 .0 9)

(0 .1 3)

(0 .0 8)

(0 .1 1)

So ur ce : A ut ho rs ’ ca lc ul at io ns

ba se d on

C S E S .

L�OPEZ MOURELO and SAMAAN | 427

T A B L E

A 2

S am

pl e ch ar ac te ri st ic s fo r ob

se rv at io ns

in ot he r m an uf ac tu ri ng

se ct or s

P re -a gr ee m en t 19

96 – 19

97 A gr ee m en t w it h ou

t IL

O m on

it or in g 19

99 A gr ee m en t w it h IL

O m on

i- to ri n g 20

04 P os t- ag

re em

en t 20

07 – 20

12

n = 17

6 n = 53

4 n = 87

n = 18

7 n = 11

0 n = 27

2 n = 31

4 n = 98

4

A ve ra ge

m on

th ly

w ag es

16 1, 97

0. 2

20 8, 65

5. 5

18 4, 96

8. 5

26 3, 49

4. 6

13 2, 42

3. 8

21 6, 35

1. 4

15 9, 65

3. 7

28 2, 15

2. 2

(2 46

,4 97

.2 )

(1 72

,5 22

.4 )

(1 62

,1 94

.6 )

(1 70

,5 86

.7 )

(1 36

,9 26

.1 )

(2 11

,2 81

.0 )

(1 03

,1 74

.9 )

(2 11

,2 09

.9 )

A ve ra ge

w ee kl y ho

ur s w or ke d

41 .7 2

47 .4 1

45 .3 0

50 .0 4

51 .7 1

52 .4 9

48 .7 9

54 .1 4

(1 6. 91

) (1 4. 13

) (1 3. 73

) (1 2. 88

) (1 2. 74

) (1 4. 21

) (1 4. 18

) (1 1. 94

)

A ge

35 .9 7

36 .6 0

38 .4 4

36 .4 2

28 .7 5

31 .9 2

33 .4 4

31 .8 1

(1 3. 66

) (1 2. 42

) (1 4. 44

) (1 3. 63

) (1 2. 14

) (1 2. 41

) (1 4. 81

) (1 1. 99

)

A ge

sq ua re d

14 .7 9

14 .9 3

16 .8 3

15 .1 1

9. 72

11 .7 3

13 .3 7

11 .5 6

(1 1. 17

) (1 0. 25

) (1 1. 69

) (1 1. 60

) (9 .1 2)

(9 .4 6)

(1 1. 75

) (9 .2 3)

L oc at io n in

ur ba n ar ea s

0. 39

0. 44

0. 45

0. 48

0. 07

0. 22

0. 25

0. 33

(0 .4 9)

(0 .5 0)

(0 .5 0)

(0 .5 0)

(0 .2 6)

(0 .4 2)

(0 .4 3)

(0 .4 7)

M ar ri ed

0. 45

0. 78

0. 64

0. 74

0. 41

0. 61

0. 38

0. 62

(0 .5 0)

(0 .4 2)

(0 .4 8)

(0 .4 4)

(0 .4 9)

(0 .4 9)

(0 .4 9)

(0 .4 9)

P ri m ar y ed uc at io n

0. 29

0. 37

0. 38

0. 28

0. 58

0. 40

0. 40

0. 32

(0 .4 5)

(0 .4 8)

(0 .4 9)

(0 .4 5)

(0 .5 0)

(0 .4 9)

(0 .4 9)

(0 .4 7)

S ec on

da ry

ed uc at io n

0. 31

0. 49

0. 23

0. 58

0. 23

0. 43

0. 33

0. 56

(0 .4 6)

(0 .5 0)

(0 .4 2)

(0 .4 9)

(0 .4 2)

(0 .5 0)

(0 .4 7)

(0 .5 0)

T er ti ar y ed uc at io n

0. 01

0. 01

0. 00

0. 02

0. 00

0. 01

0. 01

0. 01

(0 .0 8)

(0 .1 1)

(0 .0 0)

(0 .1 5)

(0 .0 0)

(0 .1 2)

(0 .0 8)

(0 .1 1)

L ow

-s ki ll ed

oc cu pa ti on

0. 83

0. 92

0. 95

0. 98

0. 97

0. 93

0. 96

0. 95

(0 .3 8)

(0 .2 7)

(0 .2 1)

(0 .1 3)

(0 .1 6)

(0 .2 6)

(0 .2 0)

(0 .2 1)

So ur ce : A ut ho rs ’ ca lc ul at io ns

ba se d on

C S E S .

428 | L�OPEZ MOURELO and SAMAAN

APPENDIX B: COMMON TREND ASSUMPTION

TABLE B1 DID estimation of the gender wage gap during the pre-agreement period

Logarithm of real hourly wages

without covariates with covariates

Textile �0.13 �0.13 (0.12) (0.11)

Male 0.31*** 0.26**

(0.11) (0.11)

Follow-up period �0.15 �0.24* (0.14) (0.13)

Textile 9 Male �0.07 �0.03 (0.17) (0.17)

Textile 9 Follow-up 0.25 0.25

(0.16) (0.16)

Male 9 Follow-up 0.10 0.12

(0.16) (0.15)

Textile 9 Male 9 Follow-up 0.02 �0.02 (0.26) (0.25)

Age 0.03***

(0.01)

Age squared �0.04*** (0.01)

Living in urban areas 0.34***

(0.05)

Being married 0.08

(0.06)

Higher education 0.18***

(0.05)

Low-skilled occupation 0.20**

(0.10)

Hours worked �0.01*** (0.00)

Constant 6.38*** 5.97***

(0.10) (0.24)

Observations 1,204 1,204

R-squared 0.04 0.13

Notes: Table reports the least square estimates of equation (2). Standard errors are in parentheses. Significance levels: *significant at 10%; **significant at 5%; *** significant at 1%. Source: Authors’ calculations based on CSES.

L�OPEZ MOURELO and SAMAAN | 429

A P P E N D IX

C : M

IN C E R

R E G R E S S IO

N S : D E T A IL

E D

R E S U L T S

T A B L E

C 1

M in ce r re gr es si on

s fo r th e te xt il e se ct or

P re -a gr ee m en t

A gr ee m en t w it h ou

t IL

O m on

i- to ri n g

A gr ee m en t w it h IL

O m on

it or in g

P os t- ag

re em

en t

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

M al e

0. 27

** *

0. 30

** *

0. 15

* 0. 13

* 0. 11

** *

0. 08

** 0. 08

** *

0. 07

** *

(0 .1 0)

(0 .0 9)

(0 .0 9)

(0 .0 8)

(0 .0 4)

(0 .0 4)

(0 .0 2)

(0 .0 2)

A ge

0. 03

** *

0. 04

** *

0. 05

** *

0. 04

** *

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 0)

A ge

sq ua re d

�0 .0 5*

** �0

.0 6*

** �0

.0 8*

** �0

.0 6*

**

(0 .0 1)

(0 .0 1)

(0 .0 1)

(0 .0 1)

L iv in g in

ur ba n ar ea s

0. 30

** *

0. 22

** *

0. 10

** *

0. 13

** *

(0 .0 7)

(0 .0 6)

(0 .0 4)

(0 .0 1)

B ei ng

m ar ri ed

0. 09

0. 05

�0 .0 0

�0 .0 2

(0 .0 8)

(0 .0 7)

(0 .0 4)

(0 .0 2)

H ig he r ed uc at io n

0. 19

** 0. 17

** *

0. 04

0. 06

** *

(0 .0 8)

(0 .0 6)

(0 .0 3)

(0 .0 1)

L ow

-s ki ll ed

oc cu pa ti on

0. 15

�0 .4 4*

�0 .2 1

�0 .1 3*

(0 .2 1)

(0 .2 5)

(0 .1 5)

(0 .0 7)

H ou

rs w or ke d

�0 .0 1*

** �0

.0 2*

** �0

.0 2*

** �0

.0 1*

**

(0 .0 0)

(0 .0 0)

(0 .0 0)

(0 .0 0)

(C on

ti nu

es )

430 | L�OPEZ MOURELO and SAMAAN

T A B L E

C 1

(C on

ti nu

ed )

P re -a gr ee m en t

A gr ee m en t w it h ou

t IL

O m on

i- to ri n g

A gr ee m en t w it h IL

O m on

it or in g

P os t- ag

re em

en t

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

C on

st an t

6. 30

** *

5. 90

** *

6. 74

** *

7. 36

** *

6. 92

** *

7. 33

** *

6. 86

** *

6. 78

** *

(0 .0 4)

(0 .3 4)

(0 .0 3)

(0 .3 5)

(0 .0 2)

(0 .2 1)

(0 .0 1)

(0 .1 0)

O bs er va ti on

s 49

8 49

8 56

9 56

9 1, 39

6 1, 39

6 4, 12

3 4, 12

3

R -s qu

ar ed

0. 02

0. 14

0. 01

0. 16

0. 01

0. 10

0. 00

0. 09

N ot es : T ab le

re po rt s th e le as t sq ua re

es ti m at es

of eq ua ti on

(3 ). S ta nd ar d er ro rs

ar e in

pa re nt he se s. S ig ni fi ca nc e le ve ls : *s ig ni fi ca nt

at 10 % ; ** si gn if ic an t at

5% ; ** * si gn if ic an t at

1% .

So ur ce : A ut ho rs ’ ca lc ul at io ns

ba se d on

C S E S .

L�OPEZ MOURELO and SAMAAN | 431

T A B L E

C 2

M in ce r re gr es si on

s fo r ot he r m an uf ac tu ri ng

se ct or s

P re -a gr ee m en t

A gr ee m en t w it h ou

t IL

O m on

i- to ri n g

A gr ee m en t w it h IL

O m on

it or in g

P os t- ag

re em

en t

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

w it h ou

t co va

ri -

at es

w it h co va

ri -

at es

M al e

0. 37

** *

0. 34

** *

0. 40

** *

0. 48

** *

0. 56

** *

0. 46

** *

0. 46

** *

0. 41

** *

(0 .0 8)

(0 .0 9)

(0 .1 2)

(0 .1 2)

(0 .1 1)

(0 .1 2)

(0 .0 5)

(0 .0 5)

A ge

0. 03

0. 07

** *

0. 02

0. 05

** *

(0 .0 2)

(0 .0 2)

(0 .0 2)

(0 .0 1)

A ge

sq ua re d

�0 .0 2

�0 .0 7*

** �0

.0 3

�0 .0 6*

**

(0 .0 2)

(0 .0 2)

(0 .0 3)

(0 .0 1)

L iv in g in

ur ba n ar ea s

0. 34

** *

0. 36

** *

0. 66

** *

0. 32

** *

(0 .0 7)

(0 .1 0)

(0 .1 3)

(0 .0 4)

B ei ng

m ar ri ed

0. 05

�0 .0 7

0. 08

0. 03

(0 .0 9)

(0 .1 4)

(0 .1 3)

(0 .0 5)

H ig he r ed uc at io n

0. 17

** 0. 18

0. 12

0. 11

** *

(0 .0 7)

(0 .1 1)

(0 .1 1)

(0 .0 4)

L ow

-s ki ll ed

oc cu pa ti on

0. 24

** �0

.3 7

0. 39

* �0

.1 7*

(0 .1 2)

(0 .3 3)

(0 .2 1)

(0 .0 9)

H ou

rs w or ke d

�0 .0 1*

** �0

.0 3*

** �0

.0 1*

* �0

.0 1*

**

(0 .0 0)

(0 .0 0)

(0 .0 0)

(0 .0 0)

C on

st an t

6. 29

** *

5. 85

** *

6. 52

** *

6. 54

** *

5. 95

** *

5. 72

** *

6. 43

** *

6. 20

** *

(0 .0 7)

(0 .3 3)

(0 .1 0)

(0 .5 2)

(0 .1 0)

(0 .4 8)

(0 .0 4)

(0 .2 0)

O bs er va ti on

s 70

6 70

6 27

4 27

4 38

2 38

2 1, 29

5 1, 29

5

R -s qu

ar ed

0. 03

0. 11

0. 04

0. 26

0. 06

0. 14

0. 07

0. 16

N ot es : T ab le

re po rt s th e le as t sq ua re

es ti m at es

of eq ua ti on

(3 ). S ta nd ar d er ro rs

ar e in

pa re nt he se s. S ig ni fi ca nc e le ve ls : *s ig ni fi ca nt

at 10 % ; ** si gn if ic an t at

5% ; ** * si gn if ic an t at

1% .

So ur ce : A ut ho rs ’ ca lc ul at io ns

ba se d on

C S E S .

432 | L�OPEZ MOURELO and SAMAAN

APPENDIX D: BLINDER–OAXACA DECOMPOSITION (D2)

0 .1

.2 .3

1996-1997 1999 2004 2007-2012

The chart shows components of the second version of the Blinder-Oaxaca decomposition (D2)

Blinder-Oaxaca Decomposition Gender Wage Gap - Textile Sector

Total Discrimination

0 .2

.4 .6

1996-1997 1999 2004 2007-2012

The chart shows components of the second version of the Blinder-Oaxaca decomposition (D2)

Blinder-Oaxaca Decomposition Gender Wage Gap - Other Manufacturing Sector

Total Discrimination

L�OPEZ MOURELO and SAMAAN | 433

Copyright of Review of Development Economics is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Millennials and the Gender Wage Gap in the U.S.: A Cross-Cohort Comparison of Young Workers Born in the 1960s and the 1980s

Kristen Roche1

Published online: 10 July 2017 # International Atlantic Economic Society 2017

Abstract Using two cohorts of young workers born in the early 1960s and early 1980s, this paper analyzes the temporal change in the U.S. gender wage gap and its determinants, which persists for both explained and unexplained reasons. Results suggest that the gender wage gap closed four (seven) percentage points at the mean (median) between cohorts. It finds cross-cohort evidence that young females’ increas- ing returns to marriage and a changing occupational wage structure contributed to a narrowing of the gap. Nonetheless, the majority of this convergence remains unex- plained due to relative improvements in unobservable institutional factors or heteroge- neity for females. Compared to the previous generation, millennials likely entered a more progressive, female-friendly labor market. It is also possible that female millen- nials are more ambitious and competitive in their early years of work experience relative to females born in the 1960s.

Keywords U.S. gender wage gap . Millennials

JEL J01 . J16 . J31

Introduction

Research on the gender wage gap in the U.S. finds that despite substantial gains in women’s earnings since the late 1970s, convergence slowed in the 1990s and early 2000s and continues to persist today. However, data from the 2011 American Community Survey indicate that women are increasingly becoming the sole or primary

Atl Econ J (2017) 45:333–350 DOI 10.1007/s11293-017-9546-6

* Kristen Roche rochek@mtmary.edu

1 School of Business , Mount Mary University, 2900 North Menomonee River Parkway, Milwaukee, WI 53222, USA

earner in American households and nearly a quarter of married women now earn more than their husbands, compared to 6% of married women in the 1960s (Wang et al. 2013). Overall, this shift is likely due to several factors including gender differences in employment during and after the Great Recession, changing family dynamics, and a rise in single-mother households.

How does the economic literature on the gender wage gap reconcile with these post- Great Recession trends? The majority of current research investigates the gender wage gap up until the 1990s and we know little of the early millennial generation experience, that is, young workers born in the 1980s. As social and labor market norms progress toward gender equality, we might expect female millennials to be different from young females of past generations. These differences, both measurable and unmeasurable, have likely impacted the female wage penalty. On average, millennials experienced different childhoods as it was more common for them to grow up in households with working mothers as well as fathers contributing relatively more to child-rearing and household production. Female millennials are relatively more educated, were exposed to a greater variety of career opportunities, and likely entered careers with more progressive norms toward working women (Wang et al. 2013). On average, they are also more likely to delay marriage and fertility in their early years of work experience (Taylor et al. 2011).

This paper compares young workers in two single-cohort longitudinal surveys, the NLSY79 and NLSY97, to investigate gender inequality among millennials in the current labor market. The earlier cohort is comprised of individuals born in the early 1960s, and the later cohort includes individuals born in the early 1980s. Given that labor market entrants have historically driven much of the gender wage gap conver- gence (Blau and Kahn 2007), the current study focuses on young workers rather than a representative sample of workers of all ages.

Unadjusted descriptive statistics depict moderate cross-cohort improvement in gen- der equality. For example, the gender log wage differential closes 2.4 log points at the mean and 3.5 log points at the median. In addition, the mean female percentile in the male wage distribution moved up four percentile points between cohorts. Indeed, 58% of male millennials out-earned female millennials, compared to 62% of young males 20 years prior.

This paper estimates temporal changes in the gender wage gap and its determinants to explore how and why the wage gap may have shifted across cohorts. The findings indicate that in 20 years, when controlling for the standard gender wage gap specification variables, the gap among young workers closed four percentage points at the mean and seven percentage points at the median. Estimates from quantile regression suggest that the female penalty increases across the wage distribution, and the shape of the distribution is nearly identical between cohorts. By comparing the determinants of wages by cohort and esti- mating cross-cohort temporal change, it suggests there are notable differences in family and job characteristics between the cohorts. Female millennials earn marriage premiums that are similar to the male experience of both cohorts. In terms of job characteristics, female millennials make a considerable improve- ment in the return to self-employment and are subject to a more favorable occupational wage structure. Moreover, cross-cohort changes in human capital contribute very little to convergence.

334 Roche K.

A Juhn et al. (1993) decomposition confirms a similar story of moderate conver- gence among young workers, although only 17% of it can be explained by changes in measured characteristics and prices. Institutional factors, such as a less discriminating labor market, or unobserved heterogeneity, such as attributes correlated with higher wages, are assumedly more favorable for female millennials relative to young females of the baby boomer generation.

Literature Review

Research on wage inequality continues to be an important topic studied by economists and other social scientists. One of the contributing factors of this inequality is the gender wage gap. Studies find that the after decades of unequal pay and a gender wage gap around 40%, the gap dramatically improved in the late 1970s through 1980s, slowed in the 1990s, and started to pick up again in the early 2000s (Blau and Kahn 2007). In 2011, the Bureau of Labor Statistics reported that the gap reached a low of 16.5% (Bureau of Labor Statistics 2011). While the overall trend is positive with women progressing toward equal pay, the gap continues to persist.

Explaining why the gender wage gap persists is complex. We know that several reasons are at play, some of which researchers can explain, and some of which remain unexplained speculation. Three types of reasons are summarized below: measurable factors, institutional factors, and unobserved heterogeneity.

First, it is known that measurable differences in human capital and job characteristics explain a portion of the gap. Although women now have relatively more years of education and their returns to higher education are rising faster relative to men’s returns (DiPrete and Buchmann 2006), they still have fewer or disrupted years of experience. These temporary leaves both decrease human capital and delay training and promotions in workers’ early careers (Mincer and Polachek 1974; Barron et al. 1993; O’Neill and Polachek 1993; Blau and Kahn 2006). Occupational choice is another important factor, whereby women are more likely to select into lower-paying, but mother-friendly jobs (Lowen and Sicilian 2009; Solberg and Laughlin 1995). While occupational segrega- tion still plays a considerable role in explaining the wage gap, women began to narrow this gap in the 1980s by becoming more educated and therefore having a greater range of career choice (Blau and Khan 2006).

Secondly, given that a portion of the wage gap cannot be explained, one can speculate that institutional factors such as discrimination and social norms may also play a role in gender pay disparity. Of course, this is difficult to quantify. Experimental studies such as Neumark et al. (1996) and Goldin and Rouse (2000), find evidence of gender discrimination. Empirical studies that directly test for discrimination find evidence that sexism lowers labor market outcomes (Charles et al. 2009) and that productivity data in the U.S. manufacturing industry support the presence of gender discrimination (Burnette 2012). However, other studies support Becker’s (2010) pre- diction that market forces reduce or eliminate discrimination in the long run (Black and Strahan 2001; Hellerstein et al. 2002).

Finally, the gender wage gap may be a result of unobserved heterogeneity. That is, the unexplained part of the gender wage gap may not be attributable to discrimination, but rather unobserved variable bias. Researchers cannot measure every attribute of a

Millennials and the gender wage gap in the U.S. 335

worker’s personality, effort, or preference for time spent working versus household production or leisure. The gender wage gap could be determined, in part, by women’s lack of competitiveness (Gneezy et al. 2003), negotiation skills (Babcock and Laschever 2003), or negotiation initiation (Rigdon 2012; Leibbrandt and List 2012); however, other research argues that gender differences in competitiveness and the returns to psychological attributes have little effect on the gender wage gap (Manning and Saidi 2010; Manning and Swaffield 2005).

Research indicates that young women’s wages are more equal to young men’s wages and younger generations have historically driven much of the gender wage gap convergence (Blau and Khan 2007). This motivates the question of why younger generations have a more equal wage ratio, yet few papers focus on young workers, particularly in the current labor market. Manning and Swaffield (2005) find that although a gender wage gap does not exist in labor market entry, the gap is nearly 25% after ten years of experience, half of which can be explained by differences in human capital. Fortin (2008) finds evidence that gender differences in young worker’s preferences for higher earnings versus family factors plays a small role in explaining the gender wage gap. Other papers that analyze the gender wage gap in young MBAs and lawyers find that women earn less than their male counterparts due to career interruptions and time spent child rearing (Bertrand et al. 2010; Wood et al. 1993).

While most papers in the gender wage gap research analyze the penalty among older generations of women with traditional attitudes about family, education, and career choice, this line of research focuses on younger workers. Yet, it is still unclear how young workers today differ from the young workers in the baby boomer generation. Analyzing time-series data to measure the change in the gender wage gap over time or cross-sectional data on young labor market entrants to measure the determinants of the wage gap can only tell us part of the story. This paper seeks to extend this line of research by analyzing the cross-cohort differences of young workers between two generations.

Data and Descriptive Statistics

This paper draws on two birth cohorts from the 1979 and 1997 National Longitudinal Survey of Youth (NLSY79 and NLSY97, respectively [Bureau of Labor Statistics 1979–2012, 1997–2014]). Each cohort is restricted to an analogous sample in order to attain age comparability between cohorts. The NLSY79 is a nationally representative sample of individuals aged 14 to 21 in 1979 who are interviewed from 1979 to 2010. In this cohort, which this paper refers to as the 1960s birth cohort, 13 waves from 1979 to 1991 are pooled, and the sample is restricted to individuals born between 1960 and 1962. In 1979, they are 17 to 19 years old and by the last measured wave, 1991, they are 29 to 31 years old.

Similarly, the NLSY97 is a nationally representative sample of individuals aged 12 to 16 in 1997 who are interviewed from 1997 to 2011. This cohort, referred to as the 1980s birth cohort, includes 13 waves from 1999 to 2011. It is restricted to include individuals born between 1980 and 1982, so that they are in the same age range across the 11 waves as the 1960s birth cohort. In summary, the birth cohorts represent two groups of young workers: one group born in the 1960s and working in the 1980s, i.e.,

336 Roche K.

the baby boomer generation, and the other group born in the 1980s and working in the 2000s, i.e., the millennial generation.1

Both cohorts are limited to individuals who work at least part-time for at least two waves. Given that this is a requirement of the fixed effects model, the sample is adjusted to be the same in the pooled ordinary least squares (OLS) models to allow for comparable samples. To further reduce heterogeneity and omit outliers, person- years are excluded if the young worker is enrolled in school or earning less than $1 or more than $200 per hour in 2000 dollars. These sample restrictions, along with the size of the NLS panels, limit the analysis to 4031 workers in the 1960s birth cohort and 4556 workers in the 1980s birth cohort.

Table 1 displays sample sizes and descriptive statistics of selected variables by birth cohort and gender.2 Comparing young workers across cohorts, it is evident that females have made relatively greater advances in unadjusted wages, yet still earn about two dollars less than the average male. Both the average and the median female are better off in the later birth cohort, while male wages only improve at the mean, indicating increasing variation in the male wage distribution.

Several other cross-cohort differences are worth mentioning. First, there are large family differences as millennials seem to delay this stage relative to baby boomers.3

Although females are still more likely to be married and have more children than males, millennials are less likely to be married and have fewer children in their early years of working. Young workers of both genders have slightly more years of work experience and education, on average, in the later birth cohort; however, females exceed males in their average years of education by 0.59 years in the earlier cohort and 0.64 years in the later cohort. Job characteristics generally follow the same pattern by gender, whereby across cohorts there is evidence for trends in deunionization, privatization, and an increase in the self-employment rate. Finally, work effort variables show that on average, young workers of both genders work more weeks, although fewer work full-time, in the later birth cohort. The decrease in full-time status may be voluntary or involuntary. On one hand, millennials may be choosing part-time work as a way to balance work and family demands versus not working at all. On the other hand, these young workers may be involuntarily working part-time hours due to economic reasons, i.e., the Great Recession.

Methodology

To start, the gender wage gap is analyzed using several indicators for wage inequality. These indicators, including the mean and median gender log wage differential, the implied female to male pay ratio, and the mean female percentile

1 Retention is critical to the validity of longitudinal data sets. Until 1991, the NLSY79 retention rate was 90.9%, and up to 2011, the NLSY97 retention rate was 84.1%. While attrition likely reduces the precision of this paper’s results, it could also bias the results if attrition is non-random. 2 The National Longitudinal Surveys suggest that researchers do not use sample weights when implementing regression analysis on longitudinal data, and thus descriptive statistics and results are constructed using unweighted data. 3 Of course, it is also possible that a larger percentage of this cohort will choose not to marry or not have children, but this statistic cannot be accurately measured at this early point in the individual’s lifecycle.

Millennials and the gender wage gap in the U.S. 337

in the male wage distribution, are common measures of wage inequality in the gender wage gap literature.

Next, the gender wage gap is estimated in the two birth cohorts using pooled to OLS and quantile regression at the .10, .25, median, .75, and .90 quantiles. This methodol- ogy compares workers to themselves at other points in time, as well as other workers. The dependent variable is the log of hourly wages in the worker’s current job, where wages have been adjusted to 2000 dollars using the Consumer Price Index-All Urban Consumers (2017). Log wages are regressed on a female dummy variable and other variables controlling for demographics (age, age squared, race), family characteristics (number of children, marital status), human capital (years of schooling, years of experience, years of experience squared), job characteristics (occupation, self-employ- ment, union status, public sector), and work effort (full-time, annual weeks worked). To explicitly test whether the gender wage gap has changed between cohorts and if the change is statistically significant, a third estimation using OLS and quantile regression includes both cohorts in a combined model with the addition of a 1980s birth cohort dummy and interactions of all independent variables with this cohort dummy.4 Given the clustered nature of the data, the standard errors are corrected by clustering individuals in all pooled OLS regressions.

4 A similar methodology is used by Avellar and Smock (2003) to compare the motherhood wage penalty across two birth cohorts.

Table 1 Descriptive statistics of selected variables by birth cohort and gender

Males Females Change

1960s 1980s 1960s 1980s Males Females

Sample size

Person-year units 17,385 14,779 15,568 13,598

Persons 2171 2308 2130 2248

Hourly wage per hour

Mean 11.96 13.11 9.77 11.16 1.14 1.40

Median 10.26 10.24 8.38 9.08 −0.02 0.70 Married 36% 24% 46% 31% −12% −16% Number of children 0.65 0.39 0.83 0.79 −0.26 −0.03 Years of education 11.9 12.7 12.5 13.3 0.74 0.78

Years of experience 1.2 1.3 1.1 1.2 0.06 0.05

Union 14% 11% 11% 9% −0.02 −0.02 Self-employed 6% 7% 4% 5% 2% 1%

Public sector 9% 2% 13% 4% −6% −9% Full-time 90% 81% 76% 70% −9% −7% Annual weeks worked 44.2 45.8 42.2 45.2 1.6 3.0

Source: Data for the 1960s individuals include 4031 workers born between 1960 and 1962 from the 1979 NLSY, waves 1979 to 1991 (Bureau of Labor Statistics 1979–2012). Data for the 1980s individuals include 4556 workers born between 1980 and 1982 from the 1997 NLSY, waves 1999 to 2011 (Bureau of Labor Statistics 1997–2014).

338 Roche K.

Subsequently, temporal changes in the determinants of the wage gap are analyzed using pooled OLS and fixed effects models. While pooled OLS models are useful in fully exploiting the longitudinal data by incorporating time-invariant characteristics of individuals, they likely contain greater omitted-variable bias and underestimate the effect of discrimination in gender wage gap studies (Choudhury 1993). Thus, fixed effects models are estimated in order to control for unobserved heterogeneity. This methodology compares workers to themselves at other points in time and consequently only estimates coefficients for time-variant variables. While this does not permit estimation of the gender wage gap as gender is time-invariant, it does allow for the estimation of the determinants of the wage gap by splitting the cohorts by gender.

Pooled OLS and fixed effects models are estimated for four groups: men in the 1960s birth cohort, men in the 1980s birth cohort, women in the 1960s birth cohort, and women in the 1980s birth cohort.5 A combined model, one with both male cohorts and the other with both female cohorts, is estimated using pooled OLS and fixed effects to test the signs and statistical significance of the cross-cohort changes.

While multivariate regression analysis is helpful in exploring cross-cohort changes, it is limited in its ability to identify whether the determinants are shifting due to changes in individuals’ quantities of measured characteristics, prices of these characteristics, or simply the effect of the residual. Subsequently, a decomposition of the gender wage gap between the birth cohorts is implemented using the Juhn-Murphy-Pierce (JMP) de- composition method (Juhn et al. 1993) and applied in the gender wage gap literature by Blau and Khan (1997, 2006).6 The JMP decomposition identifies four components that determine the change in the wage gap between two periods, in this analysis, the 1960s birth cohort and the 1980s birth cohort. These four components are: changes in the measured labor market characteristics of females compared to males, changes in the prices of measured labor market characteristics, changes in the unmeasured labor market characteristics of females compared to males (commonly referred to as the gap effect, which may reflect female’s relative improvement in unmeasured character- istics and/or reduced gender discrimination in the labor market), and changes in the prices of unmeasured labor market characteristics.

By limiting the data to individuals 31 years of age and younger, it is acknowledged that the analysis is limited to estimating the effects of variables for young workers only. 7 It does not investigate wage differences for a representative sample. The intent of this paper is to assess and decompose the temporal change among young workers born in the 1960s versus the 1980s. Given that both cohort samples are limited to young workers, the estimated models and decomposition will likely overestimate or underestimate the results in the same direction, and this will have a negligible effect on the assessment of temporal change.

5 For all fixed effects models, the Hausman test indicates a need for a fixed effects model versus a random effects model. 6 Although the JMP decomposition method is widely used in the wage inequality literature, it is not without shortcomings. These papers, along with Datta Gupta et al. (2006) and Lemieux (2006), describe some of the issues surrounding the technique. 7 Previous research finds that the motherhood penalty decreases with delayed fertility (Buckles 2008; Miller 2011). Thus, the coefficient on the number of children in the female-only models is likely overestimated compared to other analyses that measure the motherhood penalty using a sample of women who have reached the end of their child-bearing age.

Millennials and the gender wage gap in the U.S. 339

Results

Indicators of Wage Inequality

Table 2 presents wage inequality indicators for the two birth cohorts, as well as the change in indicators between cohorts. To start, as evident by the increasing standard deviations of log wages, wage inequality rose between cohorts, and relatively more among men (0.053 log points compared to 0.044 among women). The gender differ- ential in log wages fell 0.024 log points at the mean and 0.035 log points at the median. Similarly, the implied female to male pay ratio among young workers rose from 82% in the 1960s birth cohort to nearly 87% in the 1980s birth cohort. Another indicator to consider is the mean female percentile in the male wage distribution, which controls for changes in the wage structure, e.g., rising skill prices. On average, in the 1960s birth cohort, young women out-earn 38% of young men, and in the 1980s birth cohort, young women out-earn 42% of young men. Collectively, these trends indicate further improvement in gender equality for young millennials, and although the 20-year changes in the indicators are relatively small in economic size compared to the trends measured in the 1970s and 80s, they are generally consistent with the slower conver- gence measured in the 1990s (Blau and Khan 2006).

Pooled OLS Results

Results summarizing the female coefficient from pooled OLS and quantile regression models are found in Table 3. Using OLS (median) regression, an 18.6 (17.1) percentage point penalty for women is estimated in the earlier birth cohort, and a 14.4 (10.3) percentage point penalty for women is estimated in the later birth cohort. While one could infer a four to seven percentage point convergence in the gender wage gap by

Table 2 Wage inequality indicators

1960s 1980s Change

Standard deviation

Males 0.498 0.551 0.053

Females 0.503 0.547 0.044

Gender log wage differential (mean) 0.085 0.061 −0.024 Gender log wage differential (median) 0.087 0.052 −0.035 Implied female/male pay ratio a 0.819 0.865 0.046

Mean female percentile in the male wage distribution b 37.96 41.95 3.99

Source: Data for the 1960s individuals include 4031 workers born between 1960 and 1962 from the 1979 NLSY, waves 1979 to 1991 (Bureau of Labor Statistics 1979–2012). Data for the 1980s individuals include 4556 workers born between 1980 and 1982 from the 1997 NLSY, waves 1999 to 2011 (Bureau of Labor Statistics 1997–2014). a Computed as exp.(ln wf) / exp.(ln wm), where ln wf is the average log female wage and ln wm is the average log male wage. b Computed by assigning each woman a percentile ranking in the indicated year’s male wage distribution and calculating the female mean and median of these percentiles.

340 Roche K.

comparing these two models, the size and statistical significance of this change is explicitly tested by estimating a combined model with a cohort dummy and interactions of the cohort dummy with the independent variables. Accordingly, the results find evidence of a statistically significant 4.0 (6.6) percentage point decrease in the female wage penalty.

Quantile regression results suggest that OLS overestimates the penalty, although this difference is quite larger in the 1980s birth cohort. This results in a larger temporal change in the median gender wage gap, which is closer to 7 percentage points. Consistent with previous literature, the female wage penalty increases across the wage distribution, meaning high-earning women are subject to the highest penalties (Garcia et al. 2001). The shape of the distribution is generally the same in both cohorts, where the wage penalty for females in the top 10% of the wage distribution is about 8 percentage points higher than the wage penalty for females in the bottom 10% of the wage distribution.

Table 4 displays the determinants of wages by gender and cohort estimated by pooled OLS. A few differences between cohorts are significant enough to mention. First, female millennials now earn an 11 percentage point marriage premium that is similar to the male experience in both male birth cohorts. Similarly, the combined model estimates a statistically significant 6 percentage point temporal change between cohorts, although this premium is reduced one percentage point if the married female also has children. This finding is consistent with the trend that the marriage premium is increasing over time for women (Avellar and Smock 2003) and suggests a shift in the role that female millennials play as wives and primary or dual-income earners. Second, returns to human capital remain mostly unchanged between cohorts. Females of both cohorts benefit from slightly higher returns to education and millennial males benefit from a small increase in their return to experience; however, the cross-cohort changes in the combined model are statistically insignificant. Third, female millennials make large

Table 3 Female coefficient from pooled OLS and quantile regression models by birth cohort

OLS Quantile

0.1 0.25 0.5 0.75 0.9

1960s birth cohort −0.186 −0.134 −0.153 −0.171 −0.189 −0.215 (0.010) (0.008) (0.005) (0.006) (0.007) (0.008)

1980s birth cohort −0.144 −0.084 −0.091 −0.103 −0.147 −0.166 (0.012) (0.008) (0.006) (0.005) (0.007) (0.010)

Combined cohorts (temporal change) 0.040 0.053 0.059 0.066 0.045 0.051

(0.015) (0.011) (0.008) (0.008) (0.010) (0.013)

Source: Data for the 1960s individuals include 4031 workers born between 1960 and 1962 from the 1979 NLSY, waves 1979 to 1991 (Bureau of Labor Statistics 1979–2012). Data for the 1980s individuals include 4556 workers born between 1980 and 1982 from the 1997 NLSY, waves 1999 to 2011 (Bureau of Labor Statistics 1997–2014).

Notes: Results are from a log hourly earnings regression. The regressions control for age, race, marital status, number of children, education, experience, union status, self-employment, public sector, full-time status, annual weeks worked, occupation, region, and urban location. Combined cohort models control for cohort. Standard errors are in parentheses and clustered by individual. All coefficients are statistically significant at the 1% level.

Millennials and the gender wage gap in the U.S. 341

T ab

le 4

D et er m in an ts of

w ag es

fr om

p oo le d O L S m od el s by

b ir th

co ho rt an d ge nd er

C at eg o ry

V ar ia b le

M al es

F em

al es

C o m bi ne d co ho rt s (t em

p or al ch an g e)

1 96 0 s

19 80 s

1 96 0 s

19 80 s

M al es

F em

al es

F am

il y

M ar ri ed

0 .1 0 7* * *

0. 11 0* **

0 .0 3 9* * *

0. 10 8 ** *

−0 .0 03

0 .0 5 7* * *

ch ar ac te ri st ic s

(0 .0 14 )

(0 .0 1 8)

(0 .0 12 )

(0 .0 16 )

(0 .0 2 3)

(0 .0 21 )

N o.

of ch il d re n

0 .0 2 5* * *

0. 02 7 ** *

0 .0 0 7

0. 00 7

0. 0 06

0 .0 0 9

(0 .0 08 )

(0 .0 11 )

(0 .0 08 )

(0 .0 07 )

(0 .0 1 3)

(0 .0 11 )

M ar ri ed *c hi ld re n

−0 .0 10

0. 00 2

−0 .0 15 *

−0 .0 34 ** *

0. 0 10

−0 .0 24 *

(0 .0 10 )

(0 .0 1 6)

(0 .0 10 )

(0 .0 11 )

(0 .0 1 9)

(0 .0 15 )

H um

an E du ca ti o n

0 .0 4 8* * *

0. 04 4 ** *

0 .0 5 6* * *

0. 05 7 ** *

−0 .0 06

0 .0 0 2

ca pi ta l

(0 .0 03 )

(0 .0 0 3)

(0 .0 03 )

(0 .0 03 )

(0 .0 0 5)

(0 .0 05 )

E xp er ie nc e

0 .1 13 ** *

0. 12 1 ** *

0 .1 12 ** *

0. 11 4* **

0. 0 09

0 .0 0 2

(0 .0 09 )

(0 .0 0 9)

(0 .0 09 )

(0 .0 11 )

(0 .0 1 3)

(0 .0 15 )

Jo b

U ni on

0 .0 2 7

0. 20 1 ** *

−0 .0 10

0. 16 0 ** *

0. 1 76 * **

0 .1 7 7* * *

ch ar ac te ri st ic s

(0 .0 18 )

(0 .0 1 9)

(0 .0 20 )

(0 .0 24 )

(0 .0 2 6)

(0 .0 31 )

S el f- em

p lo y ed

0 .0 5 2* * *

0. 43 6 ** *

−0 .2 71 * **

0. 33 5 ** *

0. 3 84 * **

0 .6 0 3* * *

(0 .0 34 )

(0 .0 4 0)

(0 .0 51 )

(0 .0 59 )

(0 .0 5 3)

(0 .0 78 )

P ub li c se ct o r

0 .0 1 3* * *

0. 04 8 *

0 .0 0 9

0. 02 2

0. 0 37

0 .0 2 1

(0 .0 18 )

(0 .0 2 8)

(0 .0 15 )

(0 .0 20 )

(0 .0 3 3)

(0 .0 25 )

M an ag em

en t

0 .2 6 1* * *

0. 38 4 ** *

0 .1 8 4* * *

0. 47 7 ** *

0. 11 8*

0 .2 8 9* *

(0 .0 41 )

(0 .0 6 0)

(0 .0 45 )

(0 .1 11 )

(0 .0 7 2)

(0 .1 21 )

P ro fe ss io na l

0 .3 7 6* * *

0. 34 9 ** *

0 .2 5 1* * *

0. 40 4 ** *

−0 .0 31

0 .1 5 0

(0 .0 44 )

(0 .0 6 0)

(0 .0 45 )

(0 .1 1 0)

(0 .0 7 5)

(0 .1 20 )

S er vi ce s

0 .0 7 7*

0. 00 5

−0 .0 48

−0 .0 57

−0 .0 71

−0 .0 06

(0 .0 39 )

(0 .0 5 8)

(0 .0 43 )

(0 .1 1 0)

(0 .0 7 0)

(0 .1 20 )

342 Roche K.

T ab

le 4

(c on ti nu ed )

C at eg o ry

V ar ia b le

M al es

F em

al es

C o m bi ne d co ho rt s (t em

p or al ch an g e)

1 96 0 s

19 80 s

1 96 0 s

19 80 s

M al es

F em

al es

S al es

0 .2 5 7* * *

0. 14 6 ** *

0 .0 7 0

0. 19 1 *

−0 .1 1 5

0 .1 2 1

(0 .0 49 )

(0 .0 5 8)

(0 .0 48 )

(0 .1 09 )

(0 .0 7 6)

(0 .1 21 )

A dm

in is tr at iv e

0 .1 6 6* * *

0. 14 1 ** *

0 .0 7 8*

0. 32 1 ** *

−0 .0 21

0 .2 4 4* *

(0 .0 40 )

(0 .0 5 8)

(0 .0 42 )

(0 .1 09 )

(0 .0 7 0)

(0 .1 18 )

L ab or er

0 .2 8 5* * *

0. 34 7 ** *

0 .1 3 3* * *

0. 51 8 ** *

0. 0 57

0 .3 8 1* * *

(0 .0 39 )

(0 .0 5 7)

(0 .0 46 )

(0 .1 24 )

(0 .0 6 8)

(0 .1 34 )

P ro du ct io n

0 .2 5 7* * *

0. 20 8 ** *

0 .0 8 2*

0. 25 3 ** *

−0 .0 52

0 .1 7 5

(0 .0 40 )

(0 .0 5 7)

(0 .0 43 )

(0 .1 11 )

(0 .0 6 9)

(0 .1 21 )

T ra ns p or ta ti on

0 .2 3 5* * *

0. 17 7 ** *

0 .1 3 0*

0. 26 8 ** *

−0 .0 58

0 .1 3 3

(0 .0 41 )

(0 .0 5 7)

(0 .0 69 )

(0 .1 1 9)

(0 .0 7 0)

(0 .1 38 )

W o rk

ef fo rt

F ul l- ti m e

0 .0 3 8*

0. 08 0 ** *

0 .0 4 6* * *

0. 08 3 ** *

0. 0 41

0 .0 3 9* *

(0 .0 19 )

(0 .0 1 8)

(0 .0 12 )

(0 .0 14 )

(0 .0 2 6)

(0 .0 18 )

A nn ua l w ee k s

0 .4 4 9* * *

0. 18 0 ** *

0 .4 3 6

0. 13 4 ** *

−0 .2 79 * **

−0 .3 13 * **

(0 .0 37 )

(0 .0 4 0)

(0 .0 33 )

(0 .0 42 )

(0 .0 5 4)

(0 .0 53 )

S ou rc e: D at a fo r th e 1 96 0 s in d iv id ua ls in cl ud e 40 31

w or ke rs bo rn

b et w ee n 1 96 0 an d 1 96 2 fr o m th e 1 97 9 N L S Y ,w

av es

19 7 9 to 19 9 1 (B ur ea u of

L ab or

S ta ti st ic s 19 7 9– 2 01 2 ). D at a fo r

th e 1 98 0 s in di vi du al s in cl ud e 4 55 6 w or ke rs bo rn

b et w ee n 19 80

an d 19 82

fr om

th e 19 9 7 N L S Y , w av es

19 9 9 to

20 11

(B ur ea u of

L ab or

S ta ti st ic s 19 97 – 20 1 4) .

N ot es : R es u lt s ar e fr o m

a lo g ho u rl y w ag e re gr es si on . T he

re gr es si o ns

al so

co nt ro l fo r ag e, ag e sq ua re d, ra ce , re gi o n, ex p er ie n ce

sq ua re d, an d an

ur ba n lo ca ti on

du m m y. T h e ex cl ud ed

oc cu pa ti on

ca te go ry

is fa rm

in g an d fo re st ry .C

om b in ed

co h or tm

od el s co nt ro lf or

co ho rt .S ta nd ar d er ro rs ar e in pa re nt he se s an d cl u st er ed

b y in di vi du al .* ,* *, an d ** * in di ca te st at is ti ca l

si g ni fi ca n ce

at th e 10 % , 5%

, an d 1%

le ve l, re sp ec ti v el y.

Millennials and the gender wage gap in the U.S. 343

improvements in their occupational wage structure. Males, on the other hand, experi- ence an occupational wage structure that is detrimental to the later cohort. In addition to differences in occupations, the return to self-employment increases for both genders, where self-employed women (men) experience a sizable 60 (38) percentage point increase across cohorts. Fourth, there is a similar change in the return to work effort for both genders. The wage premium for full-time work increases 4 percentage points and the premium for annual weeks worked falls.

Fixed Effects Results

Thus far, it is possible that unmeasured factors affect both wages and its determinants, leading to omitted variable bias which is likely greater in the pooled OLS estimates. Furthermore, it is possible that the type of individual with particular characteristics, such as a parent or an individual with a certain occupation or level of education, shifts over time in measured and unmeasured ways. For those reasons, fixed effects models are employed to control for stable, unobserved heterogeneity. The results from the fixed effects models are summarized in Table 5. A brief comparison of the fixed effects and pooled OLS results follow.

First, relative to the pooled OLS results, a smaller yet still positive and statistically significant 3 percentage point increase in the marriage premium is evident for millen- nial females. Second, whereas the OLS estimates little change and statistical signifi- cance in the return to human capital, fixed effects result in lower returns to education and experience for both male and female millennials. Specifically, the temporal change in the return to education for females (males) is a 1.5 (1.1) percentage point reduction, and the temporal change in the return to experience is a 3.0 (1.4) percentage point reduction. Third, like the OLS model suggests, female millennials are considerably better off in their returns to occupations. To a large extent, the fixed effects model estimates even larger, more statistically significant temporal changes in the occupa- tional wage structure relative to the OLS model. Fourth, unlike the OLS results, female millennials are better off in their returns to work effort compared to male millennials. Across cohorts, the wage premium for full-time work increases a statistically significant 5.4 percentage points for females, but the smaller 2.2 percentage point increase for males is statistically insignificant.

JMP Decomposition Results

Table 6 summarizes the JMP decomposition of changes in the gender wage gap between the 1960s birth cohort and the 1980s birth cohort. The results indicate that the gap closed .049 log points, or approximately 4.9 percentage points, between generations. Measured characteristics and their corresponding prices explain 17% of this convergence, while 83% of this change remains unexplained. Decomposition results from the first two components, explained characteristics and explained prices, are grouped to reflect the four categories used above: family characteristics, human capital, job characteristics, and work effort.

Historically, human capital improvements among females have played a large role in gender wage convergence (Blau and Khan 2006; Datta Gupta et al. 2006). However, this is not the case with millennial females, as changes in education and experience had

344 Roche K.

T ab

le 5

D et er m in an ts of

w ag es

fr om

fi xe d ef fe ct s m od el s by

b ir th

co ho rt an d ge nd er

M al es

F em

al es

C o m bi n ed

C oh o rt (t em

po ra l ch an ge )

1 96 0 s

1 98 0 s

19 60 s

19 8 0s

M al es

F em

al es

P er so n- y ea rs

1 7, 3 85

1 4, 7 79

15 ,5 6 8

13 ,5 98

32 ,1 64

29 ,1 66

P er so ns

2 17 1

2 30 8

21 30

22 4 8

40 5 8

39 7 5

A vg . p er so n -y ea rs

8 .0

6 .4

7. 3

6. 0

7. 9

7. 3

C at eg or y

V ar ia bl e

F am

il y

M ar ri ed

0 .1 0 3* * *

0 .1 2 4* * *

0. 07 4 ** *

0. 1 05 * **

0. 0 13

0. 0 29 *

ch ar ac te ri st ic s

(0 .0 10 )

(0 .0 13 )

(0 .0 1 0)

(0 .0 1 3)

(0 .0 16 )

(0 .0 16 )

N o . of

ch il d re n

0 .0 5 9* * *

0 .0 5 1* * *

0. 04 5 ** *

0. 0 61 * **

0. 0 14

0. 0 19 * *

(0 .0 07 )

(0 .0 09 )

(0 .0 0 8)

(0 .0 0 7)

(0 .0 11 )

(0 .0 10 )

M ar ri ed *c hi ld re n

−0 .0 1 9* **

0 .0 0 2

−0 .0 33 ** *

−0 .0 32 ** *

0. 0 03

−0 .0 07

(0 .0 07 )

(0 .0 12 )

(0 .0 0 8)

(0 .0 0 9)

(0 .0 14 )

(0 .0 12 )

H um

an E d uc at io n

0 .0 5 5* * *

0 .0 2 4* * *

0. 07 0 ** *

0. 0 26 * **

−0 .0 11 **

−0 .0 15 * **

ca pi ta l

(0 .0 07 )

(0 .0 04 )

(0 .0 0 7)

(0 .0 0 4)

(0 .0 05 )

(0 .0 05 )

E x pe ri en ce

0 .0 9 9* * *

0 .0 8 6* * *

0. 11 1* * *

0. 0 79 * **

−0 .0 14

−0 .0 30 * **

(0 .0 06 )

(0 .0 08 )

(0 .0 0 8)

(0 .0 0 9)

(0 .0 10 )

(0 .0 11 )

Jo b

U n io n

0 .0 7 0*

0 .1 2 4* * *

0. 00 7

0. 0 95 * **

0. 0 61 * *

0. 1 20 * **

ch ar ac te ri st ic s

(0 .0 40 )

(0 .0 14 )

(0 .0 4 5)

(0 .0 1 5)

(0 .0 28 )

(0 .0 33 )

S el f- em

pl oy ed

0 .0 8 6* * *

0 .3 9 6* * *

−0 .2 56 ** *

0. 2 26 * **

0. 3 13 * **

0. 5 10 * **

(0 .0 15 )

(0 .0 17 )

(0 .0 2 0)

(0 .0 2 0)

(0 .0 23 )

(0 .0 28 )

P ub li c se ct o r

0 .0 4 2* * *

0 .0 6 7* * *

0. 07 6 ** *

0. 0 37 * *

0. 0 44 *

−0 .0 32

(0 .0 14 )

(0 .0 24 )

(0 .0 1 3)

(0 .0 1 9)

(0 .0 27 )

(0 .0 23 )

M an ag em

en t

0 .0 7 1* * *

0 .2 4 2* * *

0. 11 7* **

0. 4 09 * **

0. 1 74 * **

0. 3 18 * **

(0 .0 23 )

(0 .0 41 )

(0 .0 4 3)

(0 .0 5 7)

(0 .0 45 )

(0 .0 72 )

Millennials and the gender wage gap in the U.S. 345

T ab

le 5

(c on ti nu ed )

M al es

F em

al es

C o m bi n ed

C oh o rt (t em

po ra l ch an ge )

1 96 0 s

1 98 0 s

19 60 s

19 8 0s

M al es

F em

al es

P ro fe ss io na l

0 .1 5 7* * *

0 .2 3 8* * *

0. 11 4* **

0. 3 55 * **

0. 1 00 * *

0. 2 77 * **

(0 .0 25 )

(0 .0 41 )

(0 .0 4 2)

(0 .0 5 6)

(0 .0 46 )

(0 .0 71 )

S er v ic es

−0 .0 1 9

−0 .0 3 3

−0 .0 41

0. 0 00

−0 .0 40

0. 0 51

(0 .0 22 )

(0 .0 39 )

(0 .0 4 1)

(0 .0 5 6)

(0 .0 43 )

(0 .0 70 )

S al es

0 .0 4 0

0 .0 8 1* *

0. 03 1

0. 1 95 * **

0. 0 12

0. 1 74 * *

(0 .0 26 )

(0 .0 40 )

(0 .0 4 4)

(0 .0 5 6)

(0 .0 46 )

(0 .0 72 )

A d m in is tr at iv e

0 .0 0 8

0 .0 7 8* *

0. 02 4

0. 2 72 * **

0. 0 46

0. 2 67 * **

(0 .0 23 )

(0 .0 40 )

(0 .0 4 1)

(0 .0 5 6)

(0 .0 44 )

(0 .0 70 )

L ab or er

0 .1 0 4* * *

0 .2 3 5* * *

0. 08 7 **

0. 5 10 * **

0. 1 04 * **

0. 4 31 * **

(0 .0 20 )

(0 .0 39 )

(0 .0 4 3)

(0 .0 7 5)

(0 .0 42 )

(0 .0 87 )

P ro du ct io n

0 .1 12 ** *

0 .1 6 0* * *

0. 11 7* **

0. 2 73 * **

0. 0 22

0. 1 73 * *

(0 .0 21 )

(0 .0 40 )

(0 .0 4 2)

(0 .0 5 9)

(0 .0 43 )

(0 .0 73 )

T ra ns po rt at io n

0 .0 9 1* * *

0 .1 10 ** *

0. 09 2 *

0. 2 59 * **

−0 .0 07

0. 1 78 * *

(0 .0 23 )

(0 .0 39 )

(0 .0 5 7)

(0 .0 6 1)

(0 .0 43 )

(0 .0 85 )

W o rk

F ul l- ti m e

−0 .0 1 4

0 .0 0 3

−0 .0 05

0. 0 47 * **

0. 0 22

0. 0 54 * **

ef fo rt

(0 .0 10 )

(0 .0 10 )

(0 .0 0 8)

(0 .0 0 9)

(0 .0 15 )

(0 .0 12 )

A n nu al w ee k s

0 .3 0 1* * *

0 .1 6 3* * *

0. 28 5 ** *

0. 1 61 * **

−0 .1 90 * **

−0 .1 54 * **

(0 .0 27 )

(0 .0 35 )

(0 .0 2 7)

(0 .0 3 4)

(0 .0 45 )

(0 .0 44 )

S ou rc e: D at a fo r th e 1 96 0 s in d iv id ua ls in cl ud e 40 31

w or ke rs bo rn

b et w ee n 1 96 0 an d 1 96 2 fr o m th e 1 97 9 N L S Y ,w

av es

19 7 9 to 19 9 1 (B ur ea u of

L ab or

S ta ti st ic s 19 7 9– 2 01 2 ). D at a fo r

th e 1 98 0 s in di vi du al s in cl ud e 4 55 6 w or ke rs bo rn

b et w ee n 19 80

an d 19 82

fr om

th e 19 9 7 N L S Y , w av es

19 9 9 to

20 11

(B ur ea u of

L ab or

S ta ti st ic s 19 97 – 20 1 4) .

N ot es : R es u lt s ar e fr o m

a lo g ho u rl y w ag e re gr es si on . T he

re gr es si o ns

al so

co nt ro l fo r ag e, ag e sq ua re d, ra ce , re gi o n, ex p er ie n ce

sq ua re d, an d an

ur ba n lo ca ti on

du m m y. T h e ex cl ud ed

oc cu pa ti on

ca te g or y is fa rm

in g an d fo re st ry . C o m bi ne d co h or t m od el s co nt ro l fo r co h or t. S ta n da rd

er ro rs ar e in

pa re nt he se s. *, ** , an d * **

in di ca te st at is ti ca l si gn if ic an ce

at th e 10 % ,

5% , an d 1%

le ve l, re sp ec ti ve ly .

346 Roche K.

little to do with convergence across cohorts. Instead, among observable characteristics, convergence is attributable to females upgrading occupations, improving their relative levels of work effort, and having fewer children in their early years of work experience. Collectively, the job characteristics category, which mostly accounts for changes in occupations between the cohorts, accounted for over 80% of the fall in the gender wage gap due to observable characteristics.

Similar to the earlier finding that the mean female percentile in the male wage distribution increased four percentiles between the cohorts, the gap effect indicates that young women moved up the male residual wage distribution by .045 log points, or approximately 4.5 percentage points. This finding is likely due to a combination of reasons. Compared to young females in the earlier birth cohort, female millennials working in the 2000s may have more favorable unmeasurable skills, e.g., competitiveness, ambition, or negotiation, or they may have entered the labor market with less gender discrimination and/or more favorable supply and demand conditions for females. This finding is consistent with Weinberger and

Table 6 Decomposition of changes in the gender pay gap, 1960s birth cohort to 1980s birth cohort

Decomposition Component Effect Std. Error

Change in differential −0.049 Explained −0.008 Unexplained −0.041

Observed characteristics (x’s)

All x’s −0.045** 0.005 Family (married, number of children, married*children) −0.003 Human capital (years of schooling, years of experience) 0.001

Job characteristics (occupation, self-emp., union status, public sector) −0.039 Work effort (full-time, annual weeks worked) −0.007

Observed prices (beta’s)

All beta’s 0.045** 0.006

Family (married, number of children, married*children) −0.002 Human capital (years of schooling, years of experience) 0.003

Job characteristics (occupation, self-emp., union status, public sector) 0.041

Work effort (full-time, annual weeks worked) 0.000

Unexplained differential −0.041 Gap effect −0.045** 0.015 Unobserved prices 0.005** 0.005

Source: Data for the 1960s individuals include 4031 workers born between 1960 and 1962 from the 1979 NLSY, waves 1979 to 1991 (Bureau of Labor Statistics 1979–2012). Data for the 1980s individuals include 4556 workers born between 1980 and 1982 from the 1997 NLSY, waves 1999 to 2011 (Bureau of Labor Statistics 1997–2014).

Notes: Specification controls for age, age squared, race, region, experience squared, and an urban location dummy. Standard errors of the all x’s effect and all prices effect, and approximate standard errors of the gap effect and unmeasured prices effect, are derived using the method by the appendix to Datta Gupta et al. (2006).

**Statistically significant at the .05 level.

Millennials and the gender wage gap in the U.S. 347

Kuhn’s (2010) conclusion that the majority of the gender wage gap convergence from 1959 to 1999 was attributable to unobservable labor market factors.

The results suggest that changes in prices, on the other hand, proved unfavorable to females in the later birth cohort. Both measured prices and unmeasured prices increase the gender wage gap, although changes in measured prices account for the majority of this increase. Indeed, the gap increases .045 log points due to detrimental changes in measured prices, which is largely driven by changes in the returns to occupations. Thus, although young females in the labor market today have upgraded their occupations, the prices attached to these occupations have declined. The only factor price to close the gap is the price of family characteristics, whereby marriage and children are less costly for young women working in the 2000s.

Conclusion

This paper analyzes the gender wage gap in the millennial generation, an age group that has seen little attention in this line of economic literature. It uses two cohorts of young workers born in the early 1960s and the early 1980s to examine temporal change in the gender wage gap and its determinants and decompose the changes into explainable and unexplainable components.

Descriptive statistics and multivariate regression methods depict moderate and statistically significant cross-cohort convergence, although decomposing the female- male differential suggests that much of this convergence remains unexplained. Pooled OLS and fixed effects results provide evidence of a favorable shift in the marriage premium for female millennials so that marriage has a comparable effect on wages independent of gender. This finding is consistent with the anecdotal trend of females entering marriage as dual income earners as opposed to supporting wives and supple- mentary income earners. These models also provide considerable cross-cohort evidence for a changing occupational wage structure that favors millennial females and disfavors millennial males. Additionally, both genders benefit from increasing returns to self- employment. The implication of cross-cohort changes in work effort is uncertain. Between cohorts, the return to full-time work increases for both genders, though fixed effects imply this improvement is twice as large for females. On average, millennials of both genders work relatively more annual weeks per year compared to the baby boomer generation, yet the return on this effect falls.

In decomposing the gap, it is concluded that that while these observable factors may have played a small role in the convergence toward gender equality among millennials, unmeasurable factors played a larger role. A gap effect of the same magnitude is measured as the total change in the wage differential. Interpretation of this gap effect remains ambiguous, but one might expect it to be a combination of unobservable institutional factors and heterogeneity. Compared to baby boomers, millennials may have entered a more progressive, gender-equal labor market, or females in partic- ular may have more favorable labor market supply and demand conditions. Moreover, the unmeasurable characteristics between males and females may have narrowed over time. Female millennials may be more ambitious, more competi- tive, or have a greater willingness and ability to negotiate wages in their early years of work experience.

348 Roche K.

Younger generations have historically driven much of the gender wage gap conver- gence (Blau and Khan 2007), and this paper finds that the gender wage gap continues to converge for young millennials. Of course, the long-term effect of this progress is unknown. Policy implications arise from the determinants of the gender wage gap, which are both internal and external. Policies aimed at internal barriers should encour- age women to select into higher-paying occupations through female-friendly programs that focus on science, technology, engineering, and math (STEM) education.

Policies aimed at external barriers, such as discrimination and work environments, are also necessary. Government and employer policies that encourage the integration of work, household, and family responsibilities, as well as progressive work environ- ments, are critical in the long path to gender wage equality. Increasing the availability of policies designed for both women and men enables a shift in work and family responsibilities. For example, on-site childcare makes it convenient for parents to combine work and family, and family leave for new fathers allows men to take on more household responsibilities. Another policy implication arises from the issue of women self-selecting in lower-paying jobs.

This paper contributes a modest and initial step in exploring gender inequality among the current labor market’s youngest workers. Data limitations are undoubtedly a factor in its ability to measure these worker’s attributes as we do not yet have access to information covering their full life-cycle. As the initial group of millennials born in the 1980s age pass childbearing years, future research can analyze their labor market behavior in the longer term. Furthermore, future work on this topic could investigate a full compensation gap that considers not only the wage gap, but the gap in benefits coverage between men and women. It would be interesting to find if accounting for fringe benefits, such as retirement benefits, flexible working arrangements, or paid vacation days, narrows or widens the existing gender wage gap.

References

Avellar, S., & Smock, P. J. (2003). Has the price of motherhood declined over time? A cross-cohort comparison of the motherhood wage penalty. Journal of Marriage and Family, 65(3), 597–607.

Babcock, L., & Laschever, S. (2003). Women Don't ask: Negotiation and the gender divide. Princeton: Princeton University Press.

Barron, J. M., Black, D. A., & Loewenstein, M. A. (1993). Gender differences in training, capital, and wages. Journal of Human Resources, 28(2), 343–364.

Becker, G. (2010). The economics of discrimination (2nd ed.). Chicago: University of Chicago Press. Bertrand, M., Goldin, C., & Katz, L. F. (2010). Dynamics of the gender gap for young professionals in the

financial and corporate sectors. American Economic Journal: Applied Economics, 2(3), 228–255. Black, S. E., & Strahan, P. E. (2001). The division of spoils: Rent-sharing and discrimination in a regulated

industry. American Economic Review, 91(4), 814–831. Blau, F. D., & Kahn, L. M. (1997). Swimming upstream: Trends in the gender wage differential in the 1980s.

Journal of Labor Economics, 15(1), 1–42. Blau, F. D., & Kahn, L. M. (2006). The US gender pay gap in the 1990s: Slowing convergence. Industrial &

Labor Relations Review, 60(1), 45–66. Blau, F. D., & Kahn, L. M. (2007). The gender pay gap: Have women gone as far as they can? The Academy

of Management Perspectives, 21(1), 7–23. Buckles, K. (2008). Understanding the returns to delayed childbearing for working women. The American

Economic Review, 98(2), 403–407. Bureau of Labor Statistics, U.S. Department of Labor. (2011). Usual weekly earnings of wage and salary

workers – second quarter 2011.

Millennials and the gender wage gap in the U.S. 349

Bureau of Labor Statistics, U.S. Department of Labor (2017). CPI-All Urban Consumers (Current Series. Retrieved from http://www.usinflationcalculator.com/inflation/consumer-price-index-and-annual-percent- changes-from-1913-to-2008/.

Bureau of Labor Statistics, U.S. Department of Labor. National Longitudinal Survey of Youth 1997 cohort. (1997–2014). Retrieved from https://www.bls.gov/nls/. Date last accessed: October 2016.

Bureau of Labor Statistics, U.S. Department of Labor. National Longitudinal Survey of Youth 1979 cohort. (1979–2012). Retrieved from https://www.bls.gov/nls/. Date last accessed: October 2016.

Burnette, J. (2012). Testing for wage discrimination in U.S. manufacturing. US Census Bureau Center for economic studies paper no. CES-WP- 12-23. Available at SSRN: http://ssrn.com/abstract=2205114 or doi:10.2139/ssrn.2205114

Charles, K. K., Guryan, J., & Pan, J. (2009). Sexism and women’s labor market outcomes. Working Paper. Choudhury, S. (1993). Reassessing the male-female wage differential: A fixed effects approach. Southern

Economic Journal, 60(2), 327–340. Datta Gupta, N., Oaxaca, R. L., & Smith, N. (2006). Swimming upstream, floating downstream: Comparing

women’s relative wage progress in the United States and Denmark. Industrial and Labor Relations Review, 59(2), 243–266.

DiPrete, T. A., & Buchmann, C. (2006). Gender-specific trends in the value of education and the emerging gender gap in college completion. Demography, 43(1), 1–24.

Fortin, N. M. (2008). The gender wage gap among young adults in the United States: The importance of money versus people. Journal of Human Resources, 43(4), 884–918.

Garcia, J., Hernández, P. J., & Lopez-Nicolas, A. (2001). How wide is the gap? An investigation of gender wage differences using quantile regression. Empirical Economics, 26(1), 149–167.

Gneezy, U., Niederle, M., & Rustichini, A. (2003). Performance in competitive environments: Gender differences. The Quarterly Journal of Economics, 118(3), 1049–1074.

Goldin, C., & Rouse, C. (2000). Orchestrating impartiality: The impact of ‘blind’ auditions on female musicians. American Economic Review, 90, 715–741.

Hellerstein, J. K., Neumark, D., & Troske, K. R. (2002). Market forces and sex discrimination. Journal of Human Resources, 37, 353–380.

Juhn, C., Murphy, K. M., & Pierce, B. (1993). Wage inequality and the rise in returns to skill. Journal of Political Economy, 101(3), 410–442.

Leibbrandt, A., & List, J. A. (2012). Do women avoid salary negotiations? Evidence from a large scale natural field experiment. NBER Working Paper No. 18511.

Lemieux, T. (2006). Increasing residual wage inequality: Composition effects, noisy data, or rising demand for skill? American Economic Review, 96(3), 461–498.

Lowen, A., & Sicilian, P. (2009). Family-friendly fringe benefits and the gender wage gap. Journal of Labor Research, 30(2), 101–119.

Manning, A., & Saidi, F. (2010). Understanding the gender pay gap: What's competition got to do with it? Industrial and Labor Relations Review, 63(4), 681–698.

Manning, A., & Swaffield, J. (2005). The gender gap in early-career wage growth, CEP discussion Paper No. 700.

Miller, A. R. (2011). The effects of motherhood timing on career path. Journal of Population Economics, 24(3), 1071–1100.

Mincer, J., & Polachek, S. (1974). Family investments in human capital: Earnings of women. Journal of Political Economy, 82, S76–S108.

Neumark, D., Bank, R. J, & Van Nort, K. D. (1996). Sex discrimination in restaurant hiring: An audit study. The Quarterly Journal of Economics, 111(3), 915–941.

O’Neill, J., & Polachek, S. (1993). Why the gender gap in wages narrowed in the 1980s. Journal of Labor Economics, 11(1), 205–228.

Rigdon, M. (2012). An experimental investigation of gender differences in wage negotiations. Working paper. Available at doi:10.2139/ssrn.2165253.

Solberg, E., & Laughlin, T. (1995). The gender pay gap, fringe benefits, and occupational crowding. Industrial and Labor Relations Review, 48(4), 692–708.

Taylor, P., Passel, J., Wang, W., Velasco, G. (2011). For millennials, parenthood trumps marriage. Pew Research Center.

Wang, W., Parker, K., & Taylor, P. (2013). Breadwinner moms. Pew Research Center. Weinberger, C. J., & Kuhn, P. J. (2010). Changing levels or changing slopes? The narrowing of the gender

earnings gap 1959-1999. Industrial and Labor Relations Review, 63(3), 384–406. Wood, R. G., Corcoran, M. E., & Courant, P. N. (1993). Pay differences among the highly paid: The male-

female earnings gap in lawyers' salaries. Journal of Labor Economics, 11(3), 417–441.

350 Roche K.

Atlantic Economic Journal is a copyright of Springer, 2017. All Rights Reserved.

m a y / j u n e 2 0 1 7 w o r k f o r c e . c o m | Workƒorce 17

FOR YOUR BENEFIT

Closing the wage gap for women and minorities is a goal that many employers support, but ac- complishing it in their workplace is a daunting task that requires sifting through complex HR data. Starting this year, they will be required to dive in and address discrepancies.

Private employers with more than 100 employ- ees will be required to provide pay data and hours worked by March 2018 under new Equal Employ- ment Opportunity Commission reporting require- ments issued in September. Employers currently provide demographic data, including gender, race and ethnicity through an EEO-1 report but will now be required to submit pay information that will be analyzed to determine pay inequities.

For HR this offers an opportunity to play a strategic role in tackling a problem that is both administrative and a critical business need that af- fects diversity, recruitment and retention.

In 2015, African Americans earned just 75 per- cent as much as whites in median hourly pay and women earned 83 percent as much as men, ac- cording to a July 2016 report by the Pew Re- search Center. In fact, women and workers of all races and ethnicities combined — with the ex- ception of Asian males — lag behind white males in hourly earnings, the report found.

“In the past, the thinking was that the market- place establishes the value of a job, but today there is a concern that the marketplace may not accu- rately reflect that, that it may be based on histori- cal biases,” said employment attorney William Martucci, an instructor at Georgetown Universi- ty’s School of Continuing Studies. He said that a growing number of university HR management programs like Georgetown’s are teaching students how to identify and address pay disparities.

“It’s HR’s function to determine whether their pay practices are consistent with providing the company with a competitive advantage,” he said.

“The role of HR was once administrative, but to- day it’s all strategic.”

To better understand pay practices, vendors like ADP are developing tools to sort through the data.

In March, ADP launched Pay Equity Explorer to examine potential wage gaps according to race, gender, locations and job description.

Don Weinstein, ADP’s chief strategy officer, said that while the tool’s development was prompted by compliance changes, clients are also focused on serv- ing employees and their company’s bottom line.

Employee assistance programs offer valuable resources to employees going through hard times, like substance abuse or mental health issues. But almost half of the workforce will be made up of millennials by 2020,

according to behavioral health care company Magellan Health’s “Work- force 2020” report. And many EAPs are still stuck in the one-size-fits-all approach, which doesn’t account for how the workforce of today differs from the workforce of 30 years ago.

“You have to shift and stay current and relevant as time progresses,” said Tina Thompson, senior vice president of health and performance solutions for behavioral health company Beacon Health Options.

“If you stand still and provide an outdated process or service, you run the risk of becoming a dinosaur.”

Younger workers in general are comfortable do- ing their research before they contact an EAP, said Thompson, who has been in the industry for 30 years. In the past, EAPs may not have expected that people would do their homework. They would have relied on the EAP itself to educate them.

“They put things out there on social media, where it is open to many sources. They might be pushing things out to people they’ve never met or don’t know, because it’s just how communication has evolved from years past,” she said.

Although being proactive toward one’s behavioral health is a positive development, there are a few caveats. People may be overwhelmed by hundreds if not thousands of options for resources on the internet, and not all of this information has necessarily been vetted, said Thompson. They may end up with unreliable information about the state of their health.

As a connected group of people, younger workers may also have a sense of urgency and impatience when it comes to getting information and accessing services. Older generations waited for things.

It’s important not to generalize though, Thompson noted. Some of the stereotypes that hold true — older people being bad with technology or younger people being uncomfortable communicating via telephone — are becoming less true.

“You do need to understand the generation you’re speaking to and what their expectations are,” she said. “And then on top of that, you need to get an idea for where that individual is in the process.”

That’s where offering options and steering clear of a one-size-fits-all ap- proach comes in. An EAP needs stay relevant to everyone who uses the ser- vice, she said. That means offering a variety of access points, allowing com-

munication electronically, as well as speaking to a person.

Talking about certain topics around behavioral health can be stressful for employees using an EAP, and the goal of the service is to make sure people get their needs met in the right way and in the right time, she added.

EAPs Shift Appeal to Court Generational Differences The changing makeup of the multigenerational workforce means that EAPs have to reconsider what kind of assistance they offer and how. By Andie Burjek

HR Addresses Pay Wage Gap By Rita Pyrillis

Tina Thompson

For more on EAPs, see the Sector Report

on p. 52

Copyright of Workforce is the property of MediaTec Publishing, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

16 | LJ | OCTOBER 15, 2018

What’s old is new again in this year’s job market for newly credentialed librarians. Our snapshot shows place- ments are resurging in traditional library settings, as well as continuing to gain strength in nontraditional areas that benefit from classic LIS skill sets.

The annual LJ Placements & Salaries survey reveals a healthy job market for graduates of American Library Association (ALA)–accredited LIS master’s degree programs. The results suggest that core skills such as cataloging, refer- ence, and collection development are essential in traditional and nontraditional settings alike. Results also reinforce the strength and value of the LIS master’s degree, both in pre- paring graduates to excel in a wide range of placements and in signaling the skills and talents of candidates to potential employers. While there are some hints that salary levels are f lattening, there are glimmers of improvement in gender pay inequities and interesting shifts across employment sectors.

Forty-two of the 52 ALA-accredited schools participated in this year’s survey. They reported producing a total of 4,389 graduates during calendar year 2017. Thirty-one per- cent of these completed questionnaires about their job search status and experiences.

Graduates predominantly self-reported as female (80%),

while 18% were male, 1% nonbinary or not sure, and about 1% declined to answer. The 2017 graduates identified them- selves as white/non-Hispanic (76%), Asian/Pacific Islander (8%), Hispanic/Latino (5%), black/African American (4%), or more than a single race (4%). The gender and ethnic composition of this class was very similar to last year’s. For the third year in a row, the typical graduate was relatively young. The majority were 35 or younger (71%). Eleven per- cent were 46 or older, and the average age was 33.

FULL-TIME, SATISFACTION GROWING An increasing number of graduates found full-time posi- tions (85%). This is a slight uptick from 2016 and marks the fourth year in a row that full-time employment has exceeded 80%. Nine out of ten of these full-time positions are permanent. Only 15% of employed 2017 graduates took part-time positions, continuing the declining trend noted

last year and down by half from the per- centage of 2015 graduates who reported part-time status. The majority of this year’s part-timers hold only one position, with 40% reported holding two, for an average of 1.5 positions, which is similar to last year. Six percent of graduates reported that they were unemployed compared to 7% least year. That’s higher than the overall U.S. un- employment rate for the comparable period: about 3.95%, according to the U.S. Bureau of Labor Statistics. It’s nearly three times higher than the overall 2017 unemployment for holders of a master’s degree: 2.2%.

More than three in four graduates are satisfied with their full-time job placement (78%). The highest levels of satisfac- tion were expressed by full-time employees of public librar- ies (84%), school libraries (83%), special libraries (82%) and private industry (82%). Government library employees were the least satisfied with their positions (63%), a marked decline from last year’s 76%. Many graduates commented that the source of their satisfaction was achieving the position they aspired to have (“my dream job”) and demonstrating their learning from their LIS program. Others noted that they were now in a good “stepping stone” position that would provide experience needed to advance into preferred positions later. Other job characteristics that fueled satisfaction were good pay and benefits, schedule flexibility, camaraderie with

FOUNDATIONS AND FUTURES

Full-time employment and traditional library settings are up; gender pay gap narrows

By Suzie Allard

Suzie Allard (sallard@utk.edu) is Professor of Information Sciences and Associate Dean of Research, University of Tennessee College of Communication & Information, Knoxville. She is Principal Investigator (PI) or co-PI on grants funded by IMLS, NSF, and other foundations. She is a member of the DataONE Leadership Team and the Networked Digital Library of Theses and Dissertations Board of Directors and winner of the 2013 LJ Teaching Award

PL ACEMENTS & SAL ARIES 2018

TABLE 1 STATUS OF 2017 GRADUATES*

CURRENTLY UNEMPLOYED NUMBER OF NUMBER OF EMPLOYED EMPLOYED OR % SCHOOL SCHOOLS GRADUATES IN OUTSIDE CONTINUING TOTAL EMPLOYED REGION REPORTING RESPONDING LIS FIELD OF LIS EDUCATION ANSWERING FULL-TIME Northeast 10 296 256 25 7 288 84% Midwest 10 384 213 38 21 272 87% Southeast 9 231 224 17 11 224 84% South Central 9 267 264 38 17 264 90% West 4 189 186 19 12 186 79% TOTAL/AVG. 42 1,367 1,029 137 68 1,234 85%

TABLE BASED ON SURVEY RESPONSES FROM SCHOOLS AND INDIVIDUAL GRADUATES. FIGURES WILL NOT NECESSARILY BE FULLY CONSISTENT WITH SOME OF THE OTHER DATA REPORTED. TABLES DO NOT ALWAYS ADD UP, INDIVIDUALLY OR COLLECTIVELY, SINCE BOTH SCHOOLS AND INDIVIDUALS OMITTED DATA IN SOME CASES.

coworkers, understanding and capable managers and men- tors, and a comfortable setting. Some graduates emphasized interesting and fulfilling work, with challenges and room for growth, but also circumstances that allowed them to be successful and make a positive difference. Several expressed a deep fondness for the users or communities they serve.

Graduates in 2017 who said they were dissatisfied with their job expressed frustration about having to settle for part-time or temporary work, working multiple positions to support themselves, missing benefits, or feeling trapped in a nonprofessional library position or a position outside of LIS. Several noted that they were frustrated by unsuccess- ful attempts to land a professional LIS position. Many feel underpaid and underemployed and long to put their master’s degree to work in a fulfilling environment. Some dissatis- fied graduates were employed in professional LIS positions but not in their primary area of interest or expertise. Others had issues with management, coworkers, or insufficient re- sources. A few dissatisfied individuals mentioned their plans to improve their work life by switching to a more fulfilling position or relocating.

BUILDING ON A PROFESSIONAL PAST A majority of graduates reported that LIS is their first career (57%). About half indicated that they were already working in a library prior to starting an LIS program. Echoing past results, this year’s survey found that 43% of the 2017 gradu- ates are now career-changers with professional experience in a different domain.

The most common starter field for these graduates was education (34.7%), including those who started in K–12, higher education, adult education, and specialties such as music or math instruction. Business was the next most fre- quent previous profession (11.6%) and included subfields like sales, economics, finance, insurance, and human resources. Communication-related careers were the next largest group (9.4%), bundling publishing, writing, journalism, advertis- ing, and video production. Law careers (7.4%) were the next most common prelude to LIS studies; this category included attorneys, paralegals, and law enforcement. Those with backgrounds in entertainment and the arts (5.7%) formed the next group, including TV, music, theater arts, visual arts, video production, and museums.

SALARIES RISE SLIGHTLY The average salary for 2017 graduates employed full-time is $52,152. That’s only about 1% higher than last year’s aver- age, but it does continue the positive trend that began in 2013, and it has risen 17% from 2011. The average hourly wage rate held steady at $19.02, representing an annual full- time salary of just under $40,000.

Regional variation in average salary level conformed to the 2016 pattern. The Pacific region produced the highest average regional salary ($67,712), while the Southeast gen- erated the lowest ($47,428), a differential of over $20,000. However, the variance among the averages for all regions other than the Pacific is substantially smaller, ranging from $4,947 to $1,612. Surprisingly, the highest individual salary reported in the

2017 survey did not come from the Pacific region but from the Southeast! (This analysis does not account for regional variations in cost of living.)

GENDER GAP NARROWS This year brought some improvement in salary disparity by gender. Although the overall average salary for male gradu- ates was 12.6% higher than that for females, this differential is about six percentage points lower than in 2016. The contrast in salary level by gender is illustrated most starkly by the range of the data: the lowest full-time salary reported by a woman ($17,500) is $5,500 less than the lowest salary earned by a male graduate. The highest reported salary earned by a man was $145,000, $12,500 above the highest wage paid to a female graduate. Because LJ received so few nonbinary re- sponses to the gender question, the sample is too small to yield statistically significant results when compared to placements and salaries of other genders. Therefore, all gender compari- sons shown throughout the feature are male to female only. Nonbinary responses are included in the “all” category.

Average salaries for male graduates were higher than for women in all but one region. The most pronounced difference was for the Pacific region, where male salaries were 21.6% higher than for female graduates, though this is less than half the size of the largest differential in 2016. The Northeast region exhibited only a small gender salary differ- ential of 3.5% but still favored men. The lone exception was the Mountain region, in which male average salaries were 9.2% lower than for females.

Variations in average salaries by work settings show many of the expected gender discrepancies but with some move- ment toward parity. Last year’s female graduates working in special libraries are actually making 4.7% more than their male counterparts on average. In three other settings, men are earning only slightly more than women (government librar- ies, 0.6%, archives, 1.3%; and private industry, 2.3%). Echo- ing last year, the largest pay discrepancy by gender occurred in nonprofits, where male graduates are paid 17.2% more

than women on average. Men who work in school libraries are earning an average of 12.4% more than women. The pay bias favoring men was also exhib- ited in academic libraries (10.2%

OCTOBER 15, 2018 | LJ | 17 WWW.LIBRARYJOURNAL.COM REVIEWS, NEWS, AND MORE

TABLE 2 PLACEMENTS & FULL-TIME SALARIES OF 2017 GRADUATES BY REGION DIFFERENCE NUMBER OF AVERAGE SALARY IN AVERAGE REGION PLACEMENTS Women Men Nonbinary* All† M/F SALARY Northeast 269 $51,867 $53,688 $58,500 $52,375 3.5% Southeast 249 46,081 52,400 45,998 47,428 13.7% South Central 196 47,676 54,008 73,500 49,040 13.3% Midwest 243 49,831 55,997 50,125 51,205 12.4% Mountain 52 49,820 45,213 53,000 49,532 -9.2% Pacific 169 64,747 78,764 60,000 67,712 21.6% Canada/Intl. 21 46,050 52,575 70,000 49,900 14.2% TOTAL 1,199 50,797 57,220 57,333 52,152 12.6%

THIS TABLE REPRESENTS ONLY SALARIES REPORTED AS FULL-TIME. SOME DATA WERE REPORTED AS AGGREGATE WITHOUT BREAKDOWN BY GENDER OR REGION. COMPARISON WITH OTHER TABLES MAY SHOW DIFFERENT NUMBER OF PLACEMENTS. *INCLUDES NONBINARY, UNSURE, AND DECLINED TO ANSWER GENDER. †THE NONBINARY SAMPLE IS TOO SMALL TO YIELD STATISTICALLY SIGNIFICANT RESULTS WHEN COMPARED TO PLACEMENTS AND SALARIES OF OTHER GENDERS. THEREFORE, ALL GENDER COMPARISONS SHOWN ARE MALE TO FEMALE ONLY.

MORE ONLINE For the full charts, plus graduates’ firsthand feedback on degree requirements, job search tips, and much more, see www.libraryjournal.com/?detailStory=placements2018

higher) and public libraries (9.2%) in this survey. It is notable that the size of the male-biased pay differential in almost all settings was substantially smaller than for 2016 graduates.

SALARIES BY LIBRARY TYPE Salaries earned by 2017 graduates varied depending on work setting. The largest financial rewards came from private industry jobs, with an average annual salary of more than $78,000. This was 41% above the average salary level for the next most lucrative setting, government libraries. Non- profit organizations provided the third-highest average sal- ary ($51,590). Among the more traditional work environ- ments, school libraries yielded the highest average salaries ($51,472), while salaries in special, academic, and public libraries registered in the upper $40,000 range. The average salary for archives/special collections work was the lowest

($43,428) and had dropped about 6% from the prior year’s survey. This runs counter to the past three surveys, in which public library salaries were the lowest on average.

PUBLIC LIBRARIES were the largest career destination at some 32% of 2017 graduates. The current average salary for public library positions was $45,061. This continues the trend of a modest increase over the past two years. This year’s starting full-time salaries in public libraries ranged widely ($19,656–$118,000). The public library average salary was lowest in the Southeast region and 17% below that region’s overall average. By contrast, the Pacific public library salary was substantially higher than other U.S. regions but com- pared the least favorably with other organization types in its own area (32% lower than the overall Pacific salary average).

On average, male graduates working in public libraries earned 9.2% more than females in the class of 2017. This disparity widened slightly since last year, despite women ac- counting for 80% of this year’s public library placements and having the highest individual public library salary reported in this survey ($118,000).

COLLEGE AND UNIVERSITY LIBRARIES claimed 23% of 2017 graduates, equaling the prior year’s figure. Their overall average salary was $48,930, an increase of 4.8% over 2016. The salary range for academic library positions was unusually wide ($19,000–$145,000). Both the highest and lowest salaries were in the Southeast region. With the exception of the Southeast and Mountain regions, academic library salaries were lower than the overall averages for most sectors. This gap was largest for the Pacific (29.5% lower) and South Central (11.9%) regions.

Gender-based salary differentials for academic libraries were similar to the findings for public libraries. Although women comprised 81% of their hires, male graduates’ start- ing salaries were 10.2% higher than women’s on average, a larger differential than for 2016 graduates. This effect may have been amplified somewhat by the year’s overall top sal- ary being paid to a male graduate in this setting.

SCHOOL LIBRARIES hired 10% of the 2017 graduates, down from 13.9% in the prior survey. The average full-time salary for this setting was $51,472, 3.3% lower than 2016 but within 1% of this year’s overall average salary. School library salaries ranged from $17,500 (in the Southeast) to $92,000 (in the Pacific region). In the Northeast, school librarians’ average salary was 14.6% higher than the overall average for that region. However, in all other regions, school librarians’ average salary was lower than the overall regional average.

Male graduates filled only 8.4% of this year’s school li- brary placements, but their salaries were 12.4% higher on average than those of female graduates. This is the second- highest gender-based salary differential among all the work settings and the most disparate for a traditional library type. On the positive side, this is less than half the school library gender pay differential for 2016, and this year’s top indi- vidual salary in this setting was earned by a female graduate.

ARCHIVES AND SPECIAL COLLECTIONS employed 5% of the 2017 graduates, bouncing back from last year’s de- cline to reach a similar level to 2015. Conversely, the av- erage salary for this organizational type was $43,428 this year, dropping 4.9% from 2016. This library type offered the lowest average salary and was 16.7% below the overall national average. The salary range for this setting varied

18 | LJ | OCTOBER 15, 2018

PL ACEMENTS & SAL ARIES 2018

TABLE 3 2017 TOTAL GRADUATES AND PLACEMENTS BY SCHOOL*

EMPLOYED GRADUATES FULL-TIME RESPONSE SCHOOLS ALL** ALL** NO. REC’D RATE

Alabama 69 16 25 36.2% Albany 35 7 8 22.9% Arizona 44 8 12 27.3% Buffalo 63 10 12 19.0% Catholic* 33 4 6 18.2% Clarion 120 14 23 19.2% East Carolina 71 5 7 9.9% Florida State 88 6 9 10.2% Hawaii Manoa 28 15 22 78.6% Illinois Urbana-Champaign* 242 28 41 16.9% Indiana-Bloomington 87 26 30 34.5% Indiana-Purdue 82 6 8 9.8% Iowa 37 16 21 56.8% Kentucky 85 23 31 36.5% Long Island 104 8 11 10.6% Louisiana State 47 23 26 55.3% Maryland 76 29 38 50.0% Michigan* 126 93 109 86.5% Missouri 43 13 15 34.9% NC Chapel Hill* 66 – 65 98.5% NC Greensboro 178 18 27 15.2% North Texas 364 36 44 12.1% Oklahoma 53 16 18 34.0% Pratt 63 14 18 28.6% Queens 65 28 33 50.8% Rutgers 92 39 49 53.3% San José 505 85 117 23.2% Simmons 269 95 113 42.0% South Carolina 116 30 32 27.6% South Florida 81 12 16 19.8% Southern Mississippi 40 20 22 55.0% St. Catherine 56 23 30 53.6% St. John’s 28 12 16 57.1% Syracuse 68 9 13 19.1% Tennessee 55 18 18 32.7% Texas-Austin* 92 43 51 55.4% Texas Woman’s 148 27 32 21.6% Valdosta State 68 26 31 45.6% Washington 127 28 38 29.9% Wayne State 153 30 34 22.2% Wisconsin-Madison* 93 41 51 54.8% Wisconsin-Milwaukee 129 32 45 34.9% TOTAL/AVERAGE 4,389 1,032 1,367 31.1%

TABLES DO NOT ALWAYS ADD UP, INDIVIDUALLY OR COLLECTIVELY, OWING TO OMITTED DATA FROM SCHOOLS AND/OR INDIVIDUALS. *SOME SCHOOLS CONDUCTED THEIR OWN SURVEY AND PROVIDED RAW DATA. COMPARISON WITH OTHER TABLES MAY SHOW DIFFERENT NUMBERS OF PLACEMENTS. **INCLUDES NONBINARY, UNSURE, AND DECLINED TO ANSWER GENDER.

from only $23,800 for a single Cana- dian/International position to a high of $61,000 from the Northeast. Average salaries for archivists were highest in the Northeast and Midwest, although even in those regions, they were still well below their overall regional aver- ages (9.3% and 8.3%, respectively). It is noteworthy that this work situation involves a relatively high proportion of temporary employment; 46% of 2017 graduates working full-time in this set- ting are temporary.

Only three male graduates were hired into archival positions in 2017, so gender salary comparison is of limited value. The average salary for these male graduates was only 1.3% higher than the average salary for females working in this setting, and men were recipients of both the highest and lowest indi- vidual salaries paid in this setting.

GOVERNMENT LIBRARY positions were chosen by 3% of 2017 job seek- ers, who were rewarded with an aver- age salary of $55,285, 6.0% above the overall average salary level. Almost half of government placements were in the Southeast. The range of government salaries this year was both narrower and lower ($19,636–$75,000) than for 2016 graduates. However, government li- brary compensation outperformed the overall regional averages in five sectors: Northeast (8.8% higher), Southeast (21.2%), Mountain (16.1%), Canada/ International (4.2%), and South Central (20.3%).

The government librar y setting came the closest to achieving parity in compensation by gender. The four men hired for government library jobs this year were paid an average salary only $338 higher than the average salary earned by their 20 female counterparts. This is a substantial change from last year’s 21.8% gender pay differential.

Thirteen percent of this year’s graduates accepted full- time positions in PRIVATE INDUSTRY and were rewarded with an average annual salary of $78,094, up 6.6% from the already generous 2016 level. Private industry was again the most lucrative option for 2017’s job seekers, with an aver- age salary 50% higher than the overall average. Regional average salaries for private industry were far larger than the overall averages for all regions except Canada/International. The most impressive regional salaries for this work setting were for the Pacific (38.5% above the average salary for the region), Mountain (56.5% higher), South Central (73.3%), Southeast (34.3%), and the Northeast (32.2%). The sal- ary range for private industry this year was typically broad ($25,000–$132,500), ref lecting the wide variety of positions open to LIS graduates in nontraditional environments. The

lowest salary level was in the South Central region, while the top salary was present in both the Northeast and the Pacific. Private industry provided one of the better levels of salary parity by gender, with males earning only 2.3% more than women on average.

SPECIAL LIBRARIES hired 3% of the 2017 graduates, at an average salary level of $49,546, a modest 2.5% increase over the 2016 salary average for this setting. This was the only work context for which a gender salary disparity fa- vored female graduates; women working in this area earned salaries 4.5% higher on average than men received. The salaries for special library hires varied from $27,000 in the South Central region to $70,000 for a single position in the Canada/International category. The special library regional salaries for three regions outperformed their regions’ overall average salary levels: Canada/International (40.3% higher), Mountain (13%), and the Northeast (4.1%). No Pacific re-

OCTOBER 15, 2018 | LJ | 19 WWW.LIBRARYJOURNAL.COM REVIEWS, NEWS, AND MORE

TABLE 4 PLACEMENTS BY AVERAGE FULL-TIME SAL ARY OF REPORTING 2017 GRADUATES

AVERAGE SALARY PLACEMENTS TOTAL SCHOOLS Women Men Nonbinary** All Women Men Nonbinary** PLACEMENTS

Alabama $43,950 $39,500 – $43,208 10 2 – 12 Albany 57,250 – $60,000 58,167 2 – 1 3 Arizona 34,125 55,675 – 41,308 4 2 – 6 Buffalo 52,114 145,000 – 67,595 5 1 – 6 Catholic* 69,000 62,000 – 66,667 2 1 – 3 Clarion 49,188 55,000 – 49,833 8 1 – 9 East Carolina 31,720 48,167 – 44,055 1 3 – 4 Florida State 41,630 – – 41,630 4 – – 4 Hawaii Manoa 48,332 56,559 – 50,800 7 3 – 10 Illinois Urbana- 50,148 41,028 53,000 49,034 18 3 1 22 Champaign Indiana-Bloomington 50,844 48,333 – 50,447 16 3 – 19 Indiana-Purdue 41,750 – – 41,750 4 – – 4 Iowa 46,857 43,000 35,995 46,818 7 2 1 11 Kentucky 47,783 47,082 – 47,513 8 5 – 13 Long Island 58,900 90,000 – 64,083 5 1 – 6 Louisiana State 43,511 41,647 105,000 47,121 13 2 1 16 Maryland 51,687 59,522 56,000 53,740 18 6 1 25 Michigan* 70,156 82,500 66,250 73,853 45 21 2 68 Missouri 49,000 48,750 – 48,900 6 4 – 10 NC Greensboro 42,263 41,588 – 42,070 10 4 – 14 North Texas 44,779 54,575 – 46,085 26 4 – 30 Oklahoma 44,181 41,000 – 43,828 8 1 – 9 Pratt 52,164 52,500 – 52,192 11 1 – 12 Queens 53,655 50,000 47,500 52,754 15 3 1 19 Rutgers 52,148 55,610 48,000 52,379 23 3 1 27 San José 50,422 63,851 60,000 52,750 47 9 1 58 Simmons 50,141 48,493 66,000 50,567 60 10 3 73 South Carolina* 40,096 36,667 – 39,741 26 3 – 29 South Florida 33,948 39,500 – 35,058 8 2 – 10 Southern Mississippi 38,153 45,200 – 38,982 15 2 – 17 St. Catherine 48,833 44,763 35,000 46,820 12 6 1 19 St. John’s 53,232 46,750 – 51,792 7 2 – 9 Syracuse 38,000 50,000 – 40,000 5 1 – 6 Tennessee 48,954 – 42,000 48,458 13 – 1 14 Texas-Austin* 67,797 63,750 – 66,833 32 10 – 42 Texas Woman’s 43,593 55,000 – 44,112 21 1 – 22 Valdosta State 44,070 51,750 – 46,758 13 7 – 20 Washington 57,559 80,000 – 59,013 22 1 – 24 Wayne State 47,954 57,750 – 49,226 17 4 – 22 Wisconsin-Madison* 49,391 50,250 44,500 49,041 19 2 2 23 Wisconsin-Milwaukee 52,537 46,125 70,000 52,082 13 4 1 18 TOTAL/AVERAGE 50,797 57,220 57,333 52,152 606 140 18 768

THIS TABLE REPRESENTS PLACEMENTS AND SALARIES REPORTED AS FULL-TIME. SOME INDIVIDUALS OR SCHOOLS OMITTED INFORMATION, RENDERING INFORMATION UNUSABLE. *SOME SCHOOLS CONDUCTED THEIR OWN SURVEY AND PROVIDED RAW DATA. **INCLUDES NONBINARY, UNSURE, AND DECLINED TO ANSWER GENDER.

gion hires for this workplace type were made this year. The special libraries category bundles a wide variety of libraries (medical, art, historical, industrial/corporate, nonacademic museum, and others), so regional differences may be over- shadowed by other factors.

NONPROFIT ORGANIZATIONS employed 3% of this year’s graduates. The average salary for this sector was $51,590, nearly equivalent to the overall average salary earned by the class of 2017. It was, however, a substantial drop of 11.7% from last year’s average for nonprofit positions and slightly below what the 2015 graduates received. The range of salaries this year was less broad than usual, with a low of $25,000 and a peak of $90,000. Three regions delivered sal- aries for this work sector that exceeded their overall regional average salary levels: Southeast (26.5% higher), South Cen- tral (16.2%), and Mountain (5%). There were no nonprofit placements in the Canada/International region. Following the pattern of 2016, nonprofit organizations displayed the highest level of gender pay disparity; male graduates’ average

salary was 17.2% higher than the average for female hires. However, the size of the gender pay differential this year for nonprofits was down substantially from 29.6% for 2016 graduates.

The remaining graduates with full-time employment re- ported that they work for other types of organizations (4%) or a vendor (1%). Graduates who work for other kinds of organi-

zations earn an average salary of $50,677, which is 2.8% lower than the overall average salary, and down 5.2% from last year’s average for this employer category. The salary range is from $27,000 (in the Midwest) to $90,000 (present in both the Northeast and Pacific regions). This catch-all category was only the second (with special libraries) in this year’s sur- vey to turn the tables on gender pay dis- parity. The average salary for women graduates who work for other kinds of organizations was 4.2% above their male counterparts’.

RESPONSIBILITIES Two standard survey questions explored the range of job assignments for newly hired information professionals. In the first, graduates could select any appli- cable items from a list of 37 duties. The results confirm that their positions are often multidimensional. Each item was selected by at least 3% of graduates. The most-cited assignments were reference and information services (53%), collec- tion development and acquisitions (42%), outreach (37%), patron programming (33%), circulation (32%), readers’ advi- sory (30%), and training, teaching, and instruction (30%).

Graduates were also asked to identify their single primary duty. The top four were reference and information services (13%), children’s services (10%), school librarian/school library media special- ist (7%), and archival and preservation (6%). The responses on these two mea- sures are consistent with 2016 results.

Working graduates also provided their full job titles and their assess- ment of whether their position is in an

20 | LJ | OCTOBER 15, 2018

PL ACEMENTS & SAL ARIES 2018

TABLE 6 FULL-TIME SALARIES OF REPORTING GRADUATES BY PRIMARY JOB ASSIGNMENT

NO. % OF LOW HIGH AVERAGE MEDIAN ASSIGNMENT RECEIVED TOTAL SALARY SALARY SALARY SALARY

Access services 13 2.1% $25,000 $56,000 $44,040 $44,000 Administration 32 5.2% 21,000 145,000 49,440 45,000 Adult services 22 3.6% 29,000 61,000 46,822 49,925 Archival & preservation 27 4.4% 23,800 75,400 45,214 45,600 Assessment 4 0.7% 33,000 60,000 51,750 57,000 Budgeting/finance 2 0.3% 51,000 60,000 55,500 55,500 Children’s services 58 9.5% 19,656 60,000 43,266 42,550 Circulation 17 2.8% 19,000 51,360 38,498 40,000 Collection development/ 14 2.3% 23,000 71,000 47,748 50,000 acquisitions Communications, PR, 2 0.3% 46,000 54,000 50,000 50,000 & social media Data analytics 13 2.1% 45,000 86,000 59,077 55,000 Data curation & management 6 1.0% 40,000 63,276 49,213 47,500 Digital content management 24 3.9% 35,000 80,000 51,950 50,000 Emerging technologies 6 1.0% 40,000 60,000 48,833 48,000 Government documents 4 0.7% 38,000 62,000 49,308 48,617 Information technology 11 1.8% 32,000 75,000 52,142 54,000 Knowledge management 8 1.3% 25,000 105,000 62,500 63,500 Market intelligence/business 2 0.3% 55,000 60,000 57,500 57,500 research Metadata, cataloging, & 22 3.6% 19,636 95,000 46,499 48,000 taxonomy Outreach 14 2.3% 33,000 75,000 52,439 54,500 Patron programming 5 0.8% 47,500 68,000 57,350 58,000 Public services 13 2.1% 30,000 90,000 46,633 42,000 Records management 9 1.5% 32,000 85,300 55,171 56,000 Reference/information 75 12.2% 23,000 67,000 48,199 48,000 services School librarian/school 54 8.8% 19,500 90,000 51,536 51,416 library media specialist Solo librarian 6 1.0% 30,000 60,000 42,883 43,150 Systems technology 8 1.3% 40,000 57,900 51,108 50,000 Teacher librarian 25 4.1% 35,000 92,000 53,778 50,000 Technical services 11 1.8% 22,500 65,000 45,473 43,000 Training, teaching, & 29 4.7% 37,400 71,400 52,003 52,000 instruction User experience/usability 4 0.7% 46,000 79,000 62,500 62,500 analysis Website design 2 0.3% 50,300 75,000 62,650 62,650 Young adult/teen services 32 5.2% 32,000 64,000 44,674 42,400 Other 39 6.4% 32,000 118,000 55,803 49,000 TOTAL/AVERAGE 613 19,000 145,000 49,248 48,000

THIS TABLE REPRESENTS FULL-TIME PLACEMENTS REPORTED BY PRIMARY JOB ASSIGNMENT. SOME INDIVIDUALS OMITTED PLACEMENT INFORMATION, THEREFORE COMPARISON WITH OTHER TABLES MAY SHOW DIFFERENT NUMBERS OF PLACEMENTS AND AVERAGE AND MEDIAN SALARIES.

TABLE 5 AVERAGE SALARY FOR STARTING LIBRARY POSITIONS, 2011–2017

Library Schools Avg. Starting Difference Percentage YEAR Represented Salary in Avg. Salary Change 2011 41 $44,565 $2,009 4.72% 2012 41 $44,503 ($62) -0.14% 2013 40 $45,650 $1,147 2.58% 2014 39 $46,987 $1,337 2.93% 2015 39 $48,371 $1,384 2.95% 2016 40 $51,798 $3,427 7.08% 2017 41 $52,152 $354 0.68%

emerging area of LIS. The most unique titles were Cloud Consultant, Discovery Librarian, Creative Technologies Librarian, User Experience and Digital Scholarship Librar- ian, Data Indexer, Digital Preservation Librarian, Digital Forensics Lab Assistant, Analyst Collection Workf low Con- sultant, Digital Asset Librarian, Digital Asset Management Fellow, e-Content Analyst, Open Education Librarian, Data Analytics & Visualization Librarian, Intellectual Property Manager, and Tween Librarian.

Relatively few graduates (14%) believed that their job is in an emerging area of LIS practice. Areas mentioned include management and curation of all kinds of digital content, assets, and collections, including data and data sets, and customized database configuration and data transforma- tion. Several graduates mentioned identifying, managing, and providing instruction about new technologies. Many noted scholarly communication activities, including acces- sioning publications and managing institutional repositories. Other activities included managing technology for user experience testing and artificial intelligence; researcher sup- port through data analysis and visualization/GIS; support of open education resources and access; transitions from physical digital media to cloud-based media and streaming services; metadata and machine auto-indexing; using digital forensics tools for archiving and preservation of digital assets; and automated data risk classification for storage and access.

Some graduates occupy positions that have a traditional title, but their job duties include emergent areas in the field. Some examples are a medical librarian in charge of 3-D printing, a reference assistant who is responsible for digital curation of local history collections, a teen and technology librarian who manages and develops programming for a Maker/innovation space, an outreach librarian who performs digital outreach through social media, and a reference and access services associate who does social media curation.

THE JOB SEARCH Some 56% of respondents indicated that, upon graduation, they stayed with their employer or in the position they held prior to or while attending the LIS mas- ter’s program. Of graduates who stayed with the employer, 31% indicated that they received a raise after obtaining the degree. Others enjoyed a change in sta- tus, being promoted (22%) or moving from support to professional staff (21%). Gaining tenure eligibility (2%) affected only a few. Other circumstances graduates referred to were the ability to apply for better positions, or preparation for later transfers to other organizations. Forty-four percent reported no change in status after getting the degree.

Graduates who were looking for a position with a new employer shared their experiences when conducting their job search. Most began their search four to six months before graduation (an average of 4.7 months). Only 17% started looking after graduating. Among graduates seeking new employers, 40% were hired in their new professional po- sition prior to graduation. Only 21% relocated for their

placement, and there was very little difference in the aver- age salary earned by relocators ($49,040) versus those who took positions close to home ($48,884). On average, it took graduates about four months to find their new job, an im- provement over the prior year.

Graduates said the most helpful job-seeking resource was Indeed.com (33%). Government job websites at the city, state, and regional levels (30%) and the ALA online job list (29%) were also cited. To a lesser extent, campus job boards and Listservs (15%) and the INALJ website (12%) helped to inform their searches. Some graduates performed searches directly on the websites of places they might like to work (13%). Results for this measure have changed somewhat from last year, with an increase in the relative importance of general job search resources like Indeed and the absence of LIS professional organization sites other than ALA.

LIS SCHOOLS & JOB PLACEMENTS LIS schools connect students with vital information about available openings. A Listserv was the most commonly used channel for disseminating position announcements (87%). Many schools also used social media accounts such as Face- book and Twitter to circulate job information (58%). Some also spread the word through student organizations or activi- ties (50%), or by posting paper announcements in communal areas (45%). Only about a third have formal job placement centers or services (32%). And only six schools reported that they have a formal mentoring program for their graduates, offering formal links among alumni, students, and employers; mentoring from library staff in corresponding areas of interest; individual career counseling; faculty advisors serving as men- tors; and courses about job searching.

Using some combination of these communication chan- nels, 67% of LIS schools reported that they shared between 100 and 499 job announcements with their students in the last year. On average, each school made 534 announcements available to students. Some electronic channels such as social media and Listservs may also allow job announcements to reach alumni and other stakeholders. The schools reported that 80% of the available positions in libraries in 2017 were full-time. The schools were asked about the relative propor- tions of announcements for traditional vs. nontraditional placements in 2017; among the schools that addressed this, 39% said that the proportion was unchanged from 2016. ■

OCTOBER 15, 2018 | LJ | 21 WWW.LIBRARYJOURNAL.COM REVIEWS, NEWS, AND MORE

TABLE 7 FULL-TIME SALARIES BY TYPE OF ORGANIZATION AND GENDER

TOTAL PLACEMENTS AVERAGE SALARY ORGANIZATION Women Men Nonbinary** All Women Men Nonbinary** All Public Libraries 173 38 3 215 44,324 48,388 48,167 45,061 College/University Libraries 160 32 3 197 48,081 53,004 43,333 48,930 School Libraries 86 8 1 95 50,971 57,300 48,000 51,472 Government Libraries 20 4 2 26 55,002 55,340 58,000 55,285 Private Industry 64 31 4 100 77,676 79,458 76,250 78,094 Special Libraries 14 3 2 19 47,813 45,667 67,500 49,546 Archives/Special Collections 19 3 1 23 43,686 44,267 35,995 43,428 Nonprofit Organizations 15 4 1 20 49,487 58,000 57,500 51,590 Other Organizations 48 16 1 65 51,049 48,977 60,000 50,677

THIS TABLE REPRESENTS ONLY FULL-TIME SALARIES AND ALL PLACEMENTS REPORTED BY TYPE. SOME INDIVIDUALS OMITTED PLACEMENT INFORMATION, RENDERING SOME INFORMATION UNUSABLE. *INCLUDES NONBINARY, UNSURE, AND DECLINED TO ANSWER GENDER.

Copyright of Library Journal is the property of Media Source, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Isabelle BensiDoun,* Danièle tranCart°

Career Choices and the Gender Pay Gap: The Role of Work Preferences and Attitudes

Progress towards gender equality is held back by numerous obstacles, both in the private sphere, with the unequal division of domestic tasks, and in the occupational sphere, with the disparities between men’s and women’s wages and the glass ceiling over women’s careers. There are multiple reasons – linked to women’s education, working hours, employment sector and family constraints – why women tend to hold jobs that are less qualified, less valued and less well paid than those of men. Yet even when men and women share the same characteristics, a wage difference of around 12% is still observed. Based on this observation, Isabelle BensiDoun and Danièle tranCart examine the various components of the gender wage gap, focusing especially on work preferences and attitudes. Applying a wage gap decomposition model to data from the “Génération 1998 à 10 ans” survey, they show that women’s lower wages are explained in part by differences in these preferences and attitudes.

Women’s status has changed substantially since the time when Schopenhauer laughed at the very idea of women holding a position of power.(1) Women are now more educated than men, and go out to work not only, as in the past, to provide a second source of income for their family, but also, in many cases, to achieve personal fulfilment and a satisfying career. Despite their education and labour market investment, women’s wages are still lower, on average, than those of men. There are several explanations for this: women more often work part-time; they are also more qualified, but the subjects they study prepare them for careers in less well paid lines of work. Yet even after taking these factors into account, a non-negligible share of the gender wage gap remains unexplained. This means either that women face discrimination on the labour

(1) “The mere idea of seeing women sitting on the judges’ bench raises a smile.” The Basis of Morality, Chapter VI.

* Centre d’études prospectives et d’information internationale, CEPII.

° Centre d’études de l’emploi et du travail, CEET.

Correspondence: Isabelle Bensidoun, Centre d’études prospectives et d’information internationale, 20 avenue de Ségur, 75007 Paris, email : isabelle.bensidoun@cepii.fr



Population-E, 73 (1), 2018, 035-060 DOI: 10.3917/pope.1801.0035

market, or that other less obvious or less measurable factors are at play. For example, gender differences in the priority given to work, in personality, values or attitudes may play a role. In the final summary on gender questions in the Handbook of Labor Economics, Bertrand (2010) suggested exploring this angle, after describing the results of laboratory experiments that revealed gender differences in negotiating skills and in attitudes to risk and competition.(2) Psychological research has also identified gender disparities in personality traits and preferences. In her theory of preferences, Hakim (2004) highlights the importance of values and attitudes in employment decisions and career choices, but also in pay levels, when personal goals and preferences are involved rather than general moral stances or opinions. Hence, while women’s place on the labour market, and in society more generally, has been transformed since Schopenhauer’s time, the social norms which shape our preferences still bear the stigmata of long-standing past beliefs. The purpose of this article is to assess the influence of these gender differences in preferences and attitudes on the French labour market, and on the gender pay gap in particular. It follows on from work by Filer (1983), Mueller and Plug (2006), Fortin (2008), Grove et al. (2011), Cobb-Clark and Tan (2011), and Nyhus and Pons (2012) on the role of these factors qualified as “non-cognitive” in the international economic literature.

Most studies, excepting that of Cobb-Clark and Tan (2011), consider the direct effect of non-cognitive variables (preferences and personality traits) on wage gaps – i.e. their effect on individual productivity – and measure the contribution of these variables using a traditional decomposition method (Blinder, 1973; Oaxaca, 1973). However, these variables may also determine individuals’ career choices(3) and employers’ recruitment decisions (Chantreuil and Epiphane, 2013), thus explaining, in part, the occupational segregation between men and women observed on the labour market. Indeed, this is the conclusion drawn by Filer (1986), Ham et al. (2009), Falter and Wendelspiess Chávez Juárez (2012), John and Thomsen (2012): alongside more traditional explanatory variables (education, work experience), non-cognitive aspects are factors of heterogeneity between individuals which influence occupational choices, notably via their effect on preferences. To take account of this indirect mechanism whereby preferences and attitudes may influence wages, but also of the potentially discriminatory nature of occupational segregation, wage gaps are decomposed using the method proposed by Brown, Moon and Zoloth (1980). The wage gap is thus decomposed into an inter-occupational component (linked to the differences between male and female distributions across different occupations) and an intra-occupational component (linked to wage differences

(2) See Bertrand (2010) and Eswaran (2014) for a summary of this research.

(3) The meaning of the word “choice” in this article does not rule out the notion of constraint; likewise for preferences.

I. BensIdoun, d. TrancarT

36

within occupations), each component being split into an explained gap and an unexplained gap by the gender differences in characteristics.

By using this decomposition method, we apply an approach similar to that of Cobb-Clark and Tan (2011). But as well as a different country of observation and different non-traditional variables, our study adopts a different method to capture the influence of non-cognitive factors on wage gaps. While Cobb-Clark and Tan (2011) assess the contribution of preferences and attitudes to the wage gap by comparing estimates with and without non-cognitive variables, we propose to make a detailed decomposition. While certain technical precautions are necessary, as we shall see below (Section II), this method provides an accurate measure of the share of the wage gap attributable to these factors.

Ours is the first study to explore how gender differences in preferences and attitudes are liable to influence the gender wage gap in France. The CEREQ(4) survey used here, “Génération 1998 à 10 ans” (The 1998 cohort, 10 years on), includes subjective questions that provide insights into the potential effect of the level of priority given to one’s career, attitudes to risk and optimism about future career prospects on wage differences between young men and young women.

I. Literature review

Over the last 10 to 15 years, a growing body of research has explored the effects of non-traditional factors on labour market behaviours. After considering the potential impact of education, experience, cognitive skills (which mobilize memory, language, reasoning or problem-solving) on employment decisions or individual pay levels, attention is now turning to the role of non-cognitive capacities, notably personality traits, but also social preferences or norms. Psychologists and sociologists have long understood the key importance of these factors in decision-making, and they now form part of economists’ standard “toolbox”. For sociologists, there is nothing new in the idea that the gendered roles socially assigned to men and women shape their preferences and personality traits, and that their occupational choices and career aspirations are influenced accordingly. In economics, research on the contribution of non- cognitive variables to the gender wage gap is more recent. We will examine the results obtained by Filer (1983), Mueller and Plug (2006), Fortin (2008), Grove et al. (2011), Cobb-Clark and Tan (2011) and Nyhus and Pons (2012). Their studies cover different populations (samples), use different non-cognitive variables, and apply different decomposition methods to the results obtained (Table 1).

(4) Centre d’études et de recherches sur les qualifications (Centre for research on qualifications), Marseille.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

37

Ta b

le 1

. Li

te ra

tu re

r e

vi e

w o

f th

e i

n fl

u e

n ce

o f

n o

n -c

o g

n it

iv e v

a ri

a b

le s

o n

t h

e g

e n

d e

r w

a g

e g

a p

Fi le

r (1

98 3)

M u

el le

r an

d P

lu g

( 20

06 )

Fo rt

in (

20 08

) G

ro ve

e t

al . (

20 11

) C

o b

b -C

la rk

N

yh u

s an

d P

o n

s (2

01 2)

Sa m

p le

N o n -r

ep re

se n ta

ti ve

sa

m p le

( so

u th

-e as

t U

S,

ab o ve

-a ve

ra g e

ed u ca

te d )

U SA

, 1 9 7 2

N =

3 ,5

4 4 N

= 5

,0 2 5

W is

co n

si n

1 9

9 2

, se

co n d

ar y

sc h

o o

l le

av er

s, 1

9 5

7

N =

5 ,0

2 5

U S

1 9

8 6

3

0 y

ea r-

o ld

s

N  =

 6 ,5

2 2

M B

A s

tu d

en ts

, U

S,

1 9

9 8

( m

o re

t h

an 3

5 h

)

N  =

 9 3

3

A u

st ra

lia 2

0 0

1 -2

0 0 6

R ep

re se

n ta

ti ve

s am

p le

N =

5 ,3

9 7

N et

h er

la n d s

2 0 0 5

A g es

1 6 -6

4

N  =

 6 2 2

W ag

e M

o n th

ly H

o u

rl y

H o

u rl y

A n

n u

al H

o u

rl y

H o u rl y

N o n -c

o g n it iv

e va

ri ab

le s

(N C

V )

Pr ef

er en

ce s

Pe rs

o n

al it y

tr ai

ts

Pe rs

o n

al it y

tr ai

ts

+ p

re fe

re n

ce s

Pe rs

o n

al it y

tr ai

ts

+ p

re fe

re n

ce s

Pe rs

o n

al it y

tr ai

ts

Pe rs

o n al

it y

tr ai

ts

+ t

im e

p re

fe re

n ce

T im

e w

h en

v ar

ia b le

s w

er e

m ea

su re

d A

t ti m

e o f

su rv

ey A

t ti m

e o

f su

rv ey

A t

en d

o f

se co

n d

ar y

ed u

ca ti o

n 8

y ea

rs b

ef o

re la

b o

u r

m ar

ke t

en tr

y

H yp

o th

es is

o f

ad u lt

in va

ri an

ce o

f p

er so

n al

it y

tr ai

ts

A ft

er la

b o u r

m ar

ke t

en tr

y, c

o n tr

o lli

n g

fo r

ag e

In tr

a- o cc

u p at

io n s

(w it h in

o cc

u p at

io n s)

9 6

.6 %

Ex p la

in ed

2 1

.7 %

U n ex

p la

in ed

7 4

.9 %

In te

r- o cc

u p at

io n s

(a cr

o ss

o cc

u p at

io n s)

3 .4

%

Ex p la

in ed

–0 .4

%

U n ex

p la

in ed

3 .8

%

To ta

l d if fe

re n ce

(d if fe

re n ce

o f

lo g *

1 0 0

if u

n sp

ec ifi

ed )

W ag

e d if fe

re n ce

U

SD 2

3 7 .4

0 5 8 .7

2 2

.9 1

5 .5

1 4

.3 2 4 .6

To ta

l e xp

la in

ed 6 8

.7 %

2 4

.8 %

6 0

.0 %

2 1

.3 %

3 7 .0

%

o /w

N C

V 4 .0

% 7

.3  

8 .4

% 1

7 .4

% N

eg at

iv e

an d

v er

y lo

w 1 1 .5

%

To ta

l u n ex

p la

in ed

6 7 .0

% 3 1 .3

% 7

5 .2

% 4

0 .0

% 7

8 .7

% 6 3 .0

%

o /w

N C

V 1 0 .0

% –4

.5 %

1 3

.0 %

N o

t d

et er

m in

ed N

o t

d et

er m

in ed

0 .4

%

I. BensIdoun, d. TrancarT

38

In most of the listed studied (four out of six), the unexplained component of the wage gap is large, representing between 63% (Nyhus et Pons, 2012) and more than three-quarters of the total gap (Fortin, 2008; Cobb-Clark and Tan, 2011). The contribution stemming from gender differences in non-cognitive variables is very low and negative in Cobb-Clark and Tan’s study, and positive, at around 4%, in Filer’s study, 7.3% in Mueller and Plug, 8.4% in Fortin, 11.5% in Nyhus and Pons and as much as 17.4% in Grove et al. Only two studies – those of Fortin and of Grove et al. – give the significance levels of the various components of the explained wage gap. In the first, the contribution of the non-cognitive variables is lowered from 8.4% to 7.4% as a consequence, and in the second from 17.4% to 8.2%. All in all, the range extends “at best” from a small negative quantity to 8.2%.

The differences in market returns to these variables (“o/w NCV” line in the unexplained total of Table 1), account for 13% of the wage gap in Fortin and 10% in Filer. Their contribution is very low (0.4%) in Nyhus and Pons, and negative (–4.5%) in Mueller and Plug. However, the various authors do not estimate the significance of this component.

II. Method

After briefly describing the selected wage gap decomposition, we will present our methodological improvements with respect to existing studies, notably that of Cobb-Clark and Tan (2011).

The wage gap decomposition proposed by Brown, Moon and Zoloth (1980) has several advantages. It allows us to consider that the gender-based occupational segregation observed on the labour market is the result of individual preferences, but also of discriminatory behaviours (unlike the more widely used Oaxaca- Blinder decomposition method). It also allows us to determine the way in which preferences and attitudes can directly influence wage gaps, i.e. their effect on individual productivity, but also their impact on individuals’ choices of occupational category (OC) and employers’ recruitment decisions, and hence their effect on occupational segregation. This decomposition is expressed as follows:

j j j j

−−ˆ ˆ− ˆ R lnW

j m _p

j m – p

j wi + R lnW

j m _p

j w – p

j wi

\ \ \ \

\\

ˆ ˆ (1)R p

j w b

j m _X

j m – X

j wi + R

p

j w X

j w _b

j m – b

j wi +

lnWm – lnW w =

Explained component

Unexplained component

Explained component

Unexplained component

Intra-OC wage gap (within each OC)

Inter-OC wage gap (between OCs)

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

39

Where lnW m and

lnW w are the mean of the log of men’s and women’s wages,

and lnW

j m is the mean of the log of men’s wages in the OCj .

The first component of the equation represents the intra-OC wage gap which is explained by average differences in male characteristics X

j m− , and

female characteristics X j w− , (the fact that they do not have the same educational

level, work experience, working hours or preference, for example), while the second measures the unexplained part, stemming from differences in the returns to wages of these characteristics for women b

j w ˆ , and men b

j m ˆ , in other

words, the gender differences in the contributions of the chosen characteristics to wage levels.

In this wage gap decomposition, we see that the “justified” (explained) wage gaps are those stemming from gender differences in productivity; for example, men’s higher pay levels are justified by their greater average work experience. The unexplained differences, for their part, are linked to differences in returns to these characteristics – the fact that men with a given qualification are paid more than women for example, a situation that is totally unjustified.

Likewise, the inter-OC wage gap is decomposed into two components, the first of which represents the explained part, i.e. the difference between the observed distribution of men by OC, p

j m , and the counterfactual distribution

of women p j w ˆ , i.e. that which would exist if women with equivalent characteristics

had the same access as men to the different occupational categories. The second component measures the difference between this counterfactual distribution of women and their observed distribution p

j w , thereby capturing the unexplained

part of the inter-OC wage differences, those attributable to the differential access of men and women to the various occupational categories.

To make this decomposition, the returns of men’s and women’s characteristics must be estimated, along with the counterfactual distribution of women in the various OCs.

The equations of wages by OC for men and women have the following standard form:

It is assumed that the choice of OCs is determined by the interaction of supply factors (individual skills and preferences for an occupation compatible with family constraints) and demand factors (employers’ decisions to hire an individual based in his/her productive characteristics). These interactions are summarized in reduced form as follows:

lnW j m = b

j m X

j m + f

j m, j = 1, 2, ... , J (2)

lnW j w = b

j w X

j w + f

j w, j = 1, 2, ... , J (3)

p_y = j | Xoi = p ij =

exp_c j X

i oi

1 + R j k exp_c

j X

i oi– 1

= 1

(4)

I. BensIdoun, d. TrancarT

40

Where pij represents the probability that individual i is employed in the OC j determined by the variables Xo and the estimated coefficients cj.

These OC choices are modelled by a multinomial logit for men to evaluate the counterfactual situation for women in terms of job distribution (p

j w ˆ ).

1. Detailed decomposition

Previous studies that applied this decomposition (Chamkhi and Toutlemonde, 2015; Cobb-Clark and Tan, 2011; Meng and Meurs, 2001; Reilly, 1991) simply measured the four overall components, i.e. the explained and unexplained components of wage gaps between and within OCs. Yet measures of the characteristics (explained parts) or their returns (unexplained parts) that contribute to these various components of the gender wage gap provide key information not only to guide policy-makers – they identify the factors to be acted upon in order to reduce wage gaps – but also to assess the relative contribution of attitudes and preferences to this gap.

There are several reasons why these detailed decompositions were not made. First, overall BMZ decompositions, distinguishing between wage gap components within OCs and between OCs, already enable us to determine whether wage inequalities are due to unequal pay for equal work, or to unequal work despite equal qualifications.

Second, when the factors explaining the gender wage gap include qualitative variables, the results of estimations made using a reference group for these variables cannot be used directly to detail the unexplained parts, given that these parts are dependant on the reference groups used in the estimations (Oaxaca and Ransom, 1999). Hence, to obtain decompositions that are invariant to the choice of reference groups, Yun (2005) suggested transforming the estimated coefficients by expressing them as a difference with respect to the average and adding the coefficient of the reference group (Bensidoun and Trancart, 2015). This is the approach applied here.

Another difficulty arises, linked to the use of a non-linear model to estimate the OC choices. To circumvent this problem and obtain a detailed decomposition of inter-OC wage gaps, and hence be able to assess the influence of preferences and attitudes on overall gender wage gaps, a linear model was used to estimate OC choices.(5)

In this case, the equation

is replaced by

(5) See Bensidoun and Trancart (2015) for a discussion of the advantages and drawbacks of the linear probability model in this case.

p_y = j | Xoi = p ij =

exp_c j X

i oi

1 + R j k exp_c

j X

i oi– 1

= 1

_y = j | Xoi = p ij = c

j X

i o, j = 1, 2, ... , J

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

41

III. Description of data

The “Génération” survey used here was conducted by CÉREQ and includes subjective questions to determine career preferences, attitudes to risk and perceptions of future career prospects. Its aim was to analyse the first years of working life of a cohort of young people leaving the education system at the same time, whatever their age, educational level or skillset. The Génération 1998 survey concerns young people who left the education system in 1998, and who were interviewed in 2001 2003, 2005 and 2008. The survey weightings are always fitted to the 1998 cohort of school leavers.

The “Génération 1998 à dix ans” survey (Generation 1998, ten years on), conducted in 2008, is used here for all the variables except the preferences and attitudes variables which, as will be explained below, are based on the first Génération 1998 interview in 2001. The analysis concerns individuals in employment in 2008, excluding the self-employed,(6) who answered the question on working hours(7) (full time versus part-time). The sample comprises 9,422 individuals, of which 4,625 men and 4,797 women. As actual working hours are not available in the survey, monthly wages (including bonuses and 13th month, if applicable) are modelled and information on working hours by job category is used as a control variable.

The two modelled variables, employment and wages, are shown in Table 2 by OC, while the independent variables are presented at overall level (Table 3) to show the average differences observed between men and women.

1. Individual, familial and occupational characteristics

Table 2 shows that men and women are distributed differently across the ten selected occupational categories:(8) women are significantly over- represented in intermediate occupations in the social and health sectors, and among sales and clerical workers, while men are over-represented among engineers, supervisors, and above all among manual workers. The differences are especially pronounced for the clerical/sales worker and manual worker categories. This is consistent with the results of Brinbaum and Trancart (2015) and Meron et al. (2006) who found substantial gender segregation in employment when educational levels are low. However, beyond educational level itself, it is the specific skillset (Table 3) that doubtless contributes to this gender segregation between clerical and manual workers: women tend

(6) The sample includes 443 artisans and traders of whom more than 80% are self-employed.

(7) For this reason, 125 individuals were excluded.

(8) A detailed list comprising 10 categories was established on the basis of the INSEE list of 24 categories, ensuring that sufficient numbers were included in each category (2% in the distributions by sex). The “unskilled clerical worker” category was constructed using the scale developed by Chardon (2001).

I. BensIdoun, d. TrancarT

42

to train for jobs in the service sector (65%) and men for jobs in the industrial sector (67%).(9)

Ten years after completing their education, the mean monthly wage of young men (€1,963 in 2008) is 27.6% higher (0.24 log points) than that of young women (€1,538 in 2008, Table 2).(10) At the bottom of the wage hierarchy (manual and clerical workers), the gender gap is very wide, at between 25% and 45%, while at the other extreme, only administrative and commercial

(9) The study options in the last “class” attended were recoded in accordance with the specialities in INSEE’s NSF list. Codes 100 and 136 cover general specialities, 200 to 255 industrial specialities and 300 to 346 service specialities.

(10) This corresponds, for young women, to a wage 21.6% lower than that of young men, i.e. slightly less than the difference observed for all wage-earners in France (24%).

Table 2. Distribution of jobs and wages by occupational category

Jobs in % Mean log of wages M/F wage gap in %Men Women

M/F wage gap

Men Women M/F wage

gap

Administrative and commercial professional (31, 37: CAC)

5.3 4.1 1.2** 8.05 7.81 0.25*** 24.8***

Civil service professional / scientific, artistic and cultural professional (33, 34, 35: CFP)

5.3 7.4 –2.1*** 7.65 7.54 0.12*** 9.1*

Engineer / technical professional (38: ING)

9.0 2.2 6.8*** 7.96 7.84 0.12*** 18.6**

Civil service associate professional (42, 44, 45: PI_FP)

2.1 5.3 –3.2*** 7.38 7.33 0.05 7.3*

Health and social work associate professional (43: PI_SS)

3.3 14.6 –11.3*** 7.43 7.29 0.14*** 15.9***

Administrative and commercial associate professional, technician and supervisor (46, 47, 48: Aut_PI)

21.4 15.7 5.8** 7.52 7.38 0.14*** 11.8**

Unskilled clerical worker (Chardon scale: ENQ)

4.1 13.7 –9.6*** 7.18 6.91 0.27*** 28.7***

Skilled clerical worker (Chardon scale: EQ)

9.5 28.2 –18.7*** 7.36 7.14 0.22*** 24.7***

Skilled manual worker (62, 63, 64, 65: OQ)

26.0 3.9 22.2*** 7.34 7.08 0.26*** 25.6***

Unskilled manual worker (67, 68, 69: ONQ)

13.8 5.0 8.9*** 7.29 6.92 0.37*** 44.8***

Total 100 100 7.48 7.24 0.24*** 27.6***

The abbreviations and codes used in the INSEE classification are given in parentheses. Statistical significance: * p < 0.10; ** p < 0.05; *** p < 0.01. Source: Authors’ calculations based on CÉREQ data, Génération 1998 à 10 ans.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

43

Table 3. Individual, family and employment characteristics in 2008 and preferences in 2001(a)

Men Women M/F difference

Age 31.3 31.8 –0.50***

Experience (months) 111.4 107.6 3.8***

Qualification

Level of education

No qualifications, lower secondary 40.0 26.0 14.0***

Upper secondary (baccalauréat) 22.8 25.9 –3.1***

2 years post-secondary 18.3 23.7 –5.4***

Postgraduate diploma 18.9 24.4 –5.5***

Skillset

General 15.3 24.8 –9.5***

Industrial 57.0 10.1 46.9***

Services 27.7 65.1 –37.4***

School year retaken by secondary school entry 25.2 17.8 7.4***

Residence in Paris region 18.6 17.5 1.1

Working hours

Full time 96.9 72.5 24.4***

80 % 1.4 18.0 –16.6***

60 % 0.4 2.9 –2.5***

Part time 0.9 4.8 –3.9***

Less than half time 0.4 1.8 –1.4***

Family characteristics

Has a partner 65.6 74.2 –8.6***

Has children 46.9 63.8 –16.9***

Occupational characteristics. Sector of activity

Industry 31.2 9.8 21.4***

Administration, education, health, social 19.5 49.0 –29.5***

Other services 33.8 29.6 4.2***

Non-response and farming(b) 15.5 11.6 3.9***

Non-standard working hours 13.3 11.7 1.6*

Public-sector employment 20.4 36.9 –16.5***

Open-ended contract 88.7 85.6 3.1***

Supervisory responsibility

No employees 64.7 78.9 –14.2***

1 employee 6.9 4.2 2.7***

2-5 employees 15.4 9.6 5.8***

6+ employees 13.0 7.2 5 .8***

Company payroll

Less than 10 employees 17.3 17.8 –0.5

10-49 employees 21.8 16.0 5.8***

50-499 employees 26.7 19.3 7.4***

500+ employees 13.1 7.4 5.7***

Don’t know or not determined(b) 21.1 39.5 –18.4***

I. BensIdoun, d. TrancarT

44

professionals have a gap of similar magnitude (25%). The wage gaps are significant and lower for engineers or technical professionals (18.6%), intermediate occupations (7-16%) and civil service professionals (9%)

The characteristics used to explain these differences in the distribution of jobs and wages by occupational category of the individuals in our sample are summarized in Table 3. They show that the women, on average, are slightly older than the men (6 months) but that their experience(11) on the labour market is shorter (a difference of slightly less than 4 months). Experience measured here by means of a month-by-month description in a work diary of all positions occupied from the date of leaving the education system up to the survey date reflects the actual experience of individuals and not a potential experience, as is often the case. While women’s experience is shorter than men’s, their educational level is higher: 40% of men have a qualification no higher than a lower secondary certificate, versus just 26% of women, while almost a quarter of women have at least three years of post-secondary education, versus 19% of men. Moreover, one quarter of men had already retaken a school year by the time they reached secondary school, versus less than 18% of women.

With a shorter working history than men, women differ from their male counterparts mainly in terms of mean working hours. While 97% of men work full time, among women the proportion is just 72% (18% work four-fifths time). Differences in experience and weekly presence on the labour market may be linked to differences in family characteristics. For example, ten years after leaving education, women are more often in a union than men and, above all, a larger number are already parents: almost 2/3 already have at least one child versus less than half of the men.

In terms of occupational characteristics, gender segregation is similar to that described above, with men being more present in industrial sectors and

(11) i.e. the number of months in employment, including maternity or paternity leave but not parental leave.

Table 3 (cont'd). Individual, family and employment characteristics in 2008 and preferences in 2001(a)

Men Women M/F difference

Preferences and attitudes

Optimism 85.0 79.9 5.1***

Career 23.8 19.6 4.2***

Risk 37.2 22.3 14.9***

(a) The “qualifications” and “family characteristics” variables determine the choices of OC and wages. For wages, the model includes experience, occupational characteristics and working hours, and for choice of OC, age is added. (b) Non-response is higher for women as it mainly corresponds to public-sector jobs where the proportion of women is higher. In farming, non-response accounts for 90% of the total. Statistical significance: * p < 0.10; ** p < 0.05; *** p < 0.01. Source: Authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

45

women in services, above all in the administrative, education, health and social domains (almost 50% of women, of which two-thirds in the public sector). Not surprisingly, women more often work in the public sector (37% versus 20%) and are slightly less inclined to work non-standard hours. They slightly less often have an open-ended contract and, above all, less frequently have supervisory responsibilities (one woman in five versus more than one-third of men). There are also fewer women in large companies.

The set of characteristics presented here concerns young people, aged 31.5 years on average, i.e. 10 years younger than employees in France as a whole (INSEE, Labour Force survey 2008). For this reason, educational levels are higher for both sexes: only 40% of young men and 26% of young women do not have an upper secondary qualification (baccalauréat), compared with 55% of men and 42% of women in general. The young men’s family situation is also atypical: only two-thirds are in a union and just 23% have a dependent child, versus 72% and 36% of male wage employees in general. The young women’s family situation, on the other hand, is similar to that of female wage employees overall. Doubtless for this reason, women’s working hours are very similar in both samples: 27.5% of the young women work part-time versus 32% of female wage earners in general. The gender wage gaps are comparable in both populations, with the young women earning 21.6% less than the young men, compared with a gap of 24.3% for the entire population of wage employees.(12) Ten years after leaving the education system, the wage differences between men and women are already well entrenched.

2. Work preferences and attitudes

The database used here gives an initial insight into the possible role of preferences and attitudes – the priority given by individuals to their career, their appetite for risk and their optimism about future career prospects – in employment decisions and wages. To limit the risks of endogeneity, i.e. that these variables may be influenced by individuals’ situations on the labour market (notably their wage), the responses given in 2001 were used although the analysis was conducted for 2008.

Respondents were asked a first question about their career priorities: “Is your priority over the next three years mainly to: 1) find a stable job; 2) get ahead in your career; 3) ensure a good work-life balance?”. A dichotomous variable based on the response “get ahead with your career” was constructed. This preference expressed by individuals – probably influenced by gender stereotypes or social norms, with women more often feeling a “duty” to invest in the family sphere and men in the work sphere – may lead to certain career choices over others, and to higher wages, by encouraging those who invest in

(12) Note that the difference between the two populations is greater if men’s wages are considered in relation to women’s wages: young men’s wages are 27.6% higher on average than those of young women, versus 32.1% for all wage-earners.

I. BensIdoun, d. TrancarT

46

the work sphere to press harder for a job change or wage increase in order to satisfy their ambition (Fortin, 2008; Grove et al., 2011).

A second question on perceived future career prospects was also used. Based on the answers to the question “How do you feel about your career prospects? 1) quite worried; 2) quite optimistic; 3) don’t know”, a dichotomous variable was constructed which sets the “quite optimistic” response against the two others. Several studies have shown that employment insecurity affects wage levels as the individuals concerned tend to moderate their wage demands or limit external mobility that might lead to a better paid job at the start of their career (Aaronson et Sullivan, 1998; Campbell et al., 2007; Hakim, 2004; Simonnet, 1996).

Last, the third question: “Do you plan to set up your own business one day? 1) Yes, I do; 2) Yes, maybe; 3) No; 4) Don’t know”, was used to construct a dichotomous variable with, on one side, those who replied “Yes I do” and “Yes, maybe”, and on the other, those who chose the two other options. The variable constructed in this way is considered as a marker of a positive attitude to risk, since individuals who report plans to set up their own business have a lower level of risk aversion than the others. Being self-employed involves risk – not only financial but also personal – and social insurance is less generous. Numerous studies have shown that the least risk-averse individuals are more likely to become self-employed. Cramer et al. (2002) for the Netherlands, Ekelund et al. (2005) for Finland, Brown et al. (2011) and Ahn (2010) for the United States show that appetite for risk is a key determinant of self-employment. For France, based on experimental studies, Colombier et al. (2008) and Masclet et al. (2009) also showed that self-employed workers are significantly less risk-averse. Individual attitudes to risk may influence career choices, with the most risk-averse opting for occupations where earnings variance is low (Bonin et al., 2007), or in the public rather than private sector ( Jung, 2013). Risk aversion may also result in lower pay due to the earnings differential associated with lower risk-taking (Bertrand, 2010).

In 2001, three years after leaving education, the distributions of these three variables show significant differences between men and women, in line with other study findings. Women are, on average, significantly less optimistic, less frequently express career ambitions, and are more risk-averse than men (Table 3). These gender differences persist even after controlling for differences in educational level and skillset.

The variables of preferences and attitudes were measured seven years before the date concerned by our analysis, but after the respondents’ labour market entry (i.e. after 1998), so they may reflect individuals’ labour market situation and not their “true” preferences. This means that the results of our analysis in terms of gender wage gaps may be contaminated by the labour market situation if this factor influences men and women differently. To test for this potential bias, we used logistic models to estimate the influence of the number

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

47

of months of unemployment between 1998 and 2001 on work preferences and attitudes (controlling by educational level) and tested whether the impact was different for men and women.

Table 4, which gives the marginal effects of unemployment, sex and the interaction between these two variables, shows that optimism and giving priority to one’s career are affected by unemployment: all other things being equal, the longer individuals remains unemployed, the less optimistic they are about their career prospects and the less priority they give to their working career. Moreover, and in line with the results of the descriptive statistics, women are less frequently optimistic, less frequently express the desire to have a career, and are more risk-averse than men. But above all, we observe that the interaction of the gender effect and of months spent unemployed is not significant for any of the variables studied: while unemployment does indeed affect the respondents’ responses about their career prospects or the desire to give priority to their career, there is no difference between men and women in this respect

IV. Results and discussion

The decomposition of wage gaps used here enables us to identify the share of these gaps that is linked to differences in male and female characteristics (explained gap), and the unexplained share. It also allows us to determine the share stemming from work segregation, i.e. the fact that men and women are not equally distributed across occupational categories (inter-OC wage gap). The results are first presented at overall level, then at detailed level, to identify the factors behind the overall wage differences. The section will conclude with a discussion of the reasons why our results on the gender wage gap in France differ from those published elsewhere.

Table 4. Preferences and attitudes to work and unemployment

Mean marginal effects Optimism Career Risk

Months of unemployment 1998-2001 –0.0056*** –0.0067*** –0.00041

(0.00059) (0.011) (0.00086)

Sex (Ref.: male) –0.0605*** –0.043*** –0.137***

(0.0105) (0.0108) (0.0122)

Sex * months of unemployment 0.00079 0.0025 –0.0014

(0.00079) (0.00226) (0.0016)

Logit model, standard deviations in brackets. Note: Marginal effects of educational level not given here. Statistical significance: * p < 0.10; ** p < 0.05; *** p < 0.01. Source: Authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

I. BensIdoun, d. TrancarT

48

1. Overall wage gap decomposition

How much can be explained?

Table 5 gives the results of the BMZ(13) decomposition for the estimates without (column 1) and with (column 2) variables of preferences and attitudes, and for those of the Oaxaca-Blinder decomposition(14) (column 3).

Comparison of the second and third columns shows that, as expected, the explicit inclusion of labour market gender segregation in the decomposition of the gender wage gap (column 2) reduces the explained component. While the explained part accounts for 60% of the wage gap when the gender distribution across occupational categories is considered as stemming solely from personal choice (column 3), it accounts for no more than 40% when the decomposition assumes that these distributions may also reflect discriminatory behaviour on

(13) The results presented here consider that choice of OC and wage are independent, given that taking account of possible occupational selection linked to unobservable characteristics affecting both OC choices and wages gives non-significant results (Bensidoun and Trancart, 2015).

(14) Selection linked to the fact that only wages of employed individuals are observed was tested for both sexes, taking parents’ social origin and mother’s occupation as exclusion variables. However, the inverse of the Mills ratio was not significant in either the men’s or the women’s wage equation. This is probably due to the fact that the women in our sample have a labour force participation rate that is high (92%), because they are relatively young, and similar to that of men (98%).

Table 5. Decomposition of the gender wage gap

Without variables of preferences and attitudes

(1)

With variables of preferences and attitudes

(2) Oaxaca-Blinder (3)

Wage gap % of gap Wage gap % of gap Wage gap % of gap

Total difference 24.4 100 24.4 100 24.4 100

Total explained 7.3*** 30.1 9.3*** 38.1 14.7*** 60.3

Total unexplained 17.0*** 69.9 15.1*** 61.9 9.7*** 39.7

Intra-OC 19.3*** 79.3 19.3*** 79.1

Explained 10.7*** 43.8 11.3*** 46.3

Unexplained 8.6*** 35.5 8.0*** 32.8

Inter-OC 5.0*** 20.7 5.1*** 20.9

Explained –3.3*** –13.7 –2.0*** – 8.2

Unexplained 8.3*** 34.4 7.1*** 29.1

The differences in log wages were multiplied by 100 to make the table more legible. Statistical significance: * p < 0.10; ** p < 0.05; *** p < 0.01 based on 200 bootstrap sample replications, except for the Oaxaca-Blinder column. Source: Authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

∑ p j w β

j m _X

j m – X

j wi

j

ˆ − −

∑ p j w X

j w _β

j m – β

j wi

j

ˆ ˆ−

∑ lnW j m _p

j m – p

j wi

j

ˆ

∑ lnW j m _p

j w – p

j wi

j

ˆ

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

49

the part of employers. Introducing preferences and attitudes, on the other hand, reduces the unexplained component, which falls from 70% without preferences and attitudes (column 1) to 62% (column 2). The unexplained intra-OC gap is reduced, but also the unexplained inter-OC gap, thereby justifying the use of a decomposition that takes into account the indirect effects of preferences and attitudes on careers.

What is the role of occupational segregation?

Looking now at the shares of the wage gap attributable to intra-OC and inter-OC differences (column 2), we see that almost 80% of the wage gap is due to wage differences between men and women within the different occupational categories, with 46% stemming from differences in characteristics and 33% remaining unexplained. The wage gaps attributable to gender segregation on the labour market – the fact that men and women are not equally distributed across OCs – account for just 20%,(15) although this figure is the sum of a negative explained component (–8.2%) and an unexplained component of almost 30%. This first negative component signifies that if women had the same opportunities as men to work in the various occupational categories, their characteristics would enable them, on average, to obtain jobs with higher wages than men. Figure 1 illustrates this situation. It shows three distributions: that of men, that of women and the counterfactual distribution that would be observed if women had the same opportunities to work in the various OCs as men. The OCs are ranked by decreasing level of mean male wages.

We see that many OCs should account for a larger share of women’s employment than that of men.(16) Only the OCs of engineer and, above all, manual worker are “legitimately” masculine, in the sense that even if women had equal access to the various OCs, these two categories would still represent a larger share of male jobs than female jobs due to the characteristics which more frequently orient men towards them. Yet, given that the OCs where they are legitimately more strongly represented (among manual workers) are also those where mean wages are low, the gender wage gap should favour women, given their characteristics, if they have the same level of access to the various OCs as men. This graph also shows that the counterfactual distribution of women is very different to the actual one. There should be higher proportions of women who are administrative and commercial professionals, engineers, civil service professionals, manual workers or associate professionals in fields other than health, social services and the civil service. Conversely, the proportions who are skilled or unskilled clerical workers or associate professionals in health, social services and the civil service should be much smaller. The wage gaps linked to occupational segregation remain unexplained

(15) This low share of the inter-OC component is even more pronounced in the study by Cobb-Clark and Tan (2011) presented in Table 1, but also, in a similar proportion, in that of Meng and Meurs (2001) on the gender wage gap in France in 1992.

(16) The OCs for which the female counterfactual bar is higher than the male bar in the figure.

I. BensIdoun, d. TrancarT

50

here. They may be linked to determinants other than those used here, or to discrimination against women in access to various occupations.

2. Detailed decompositions: the factors behind gender wage gaps

The detailed decompositions of wage gaps within and between occupational categories enable us to identify the characteristics that account for the overall gaps analysed so far.(17) In this respect, the first columns of Table 6 show that more than one-third of the total wage gap is explained by higher levels of part- time working among women,(18) and 13.5% by gender differences in occupational characteristics. For example, the fact that more men than women occupy

(17) The detailed decompositions were performed using the STATA Oaxaca program developed by Jann (2008) to each OC. The mean contributions of each variable were then obtained by weighting the different contributions per OC by the distribution of female jobs by OC for the intra-OC part and the distribution of male jobs by OC for the inter-OC part.

(18) By considering working hours and managerial responsibility as elements that justify the wage gap, the model used here assumes that the observed situation is the result of individual choice. Bensidoun and Trancart (2015) consider that in 25% to 36% of cases, women in part-time work did not choose their working hours. Our decomposition therefore underestimates the unexplained part of the gender wage gap.

Figure 1. Observed distribution of women’s and men’s jobs by occupational category and counterfactual distribution of women (%)

0

5

10

15

20

25

30

CAC Engineer CFP Other PI PI_SS PI_FP Skilled clerical

Skilled manual

Unskilled manual

Unskilled clerical

6.6

6.8

7.0

7.2

7.4

7.6

7.8

8.0

8.2

Men Women counterfactual

Women

Mean male wage (log), right-hand scale INED

001A18

Percentage Mean log of wages

Note: CAC: Administrative and commercial professional; CFP: Civil service professional; Other PI: Administrative and commercial associate professional, technician and supervisor, etc.; PI_SS: Health and social

work associate professional; PI_FP: Civil service associate professional. Statistical significance: * p < 0.10; ** p < 0.05; *** p < 0.01.

Source: Authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

51

managerial positions justifies their higher pay levels.(19) Their greater professional experience also explains 3.5% of the wage gap in favour of men. Conversely, women’s level of education and the fact the more women than men live with a partner and have children at these ages plays in the opposite direction.

The decomposition by characteristic of the unexplained intra-OC wage gap (columns 3 and 4) shows first that most of this gap is linked to differences in the constants estimated for men and women, and second, that the returns of men’s and women’s characteristics are not significantly statistically different, apart from the return of their family characteristics, which is unfavourable to women. This tallies with the conclusions of Filer (1983) and of Nyhus and

(19) The glass ceiling on women’s careers suggests that the low proportion of women in managerial positions is not simply the result of individual choice, but also of discriminatory behaviour. If so, as is the case for involuntary part-time working, our decomposition underestimates the unexplained part of the gender wage gap.

Table 6. Detailed decomposition of the gender wage gap(a) 

Intra-OC Inter-OC (b)

Explained %

of gap Un-

explained %

of gap Explained

% of gap

Un- explained

% of gap

(1) (2) (3) (4) (5) (6) (7) (8)

Experience/age 0.8** 3.5 –0.3 –1.4 –0.01 0.0 –0.08 –0.4

Educational level –1.5*** –6.0 –2.6 –10.7 4 –3.0*** –12.5 –0.2 –0.5

Speciality 0.5 2.0 0.1 0.3

Occupational characteristics

3.3*** 13.5 –2.5 –10.4

Working hours 8.2*** 33.8 0.7 2.8

Residence in Paris region

0.1 0.3 1.0 3.9 0.07 0.3 –0.4 –1.8

Family characteristics

–1.0*** –3.9 2.1*** 8.8 –0.3 –1.1 0.5*** 2.2

Optimism Career Risk

0.4** 0.3* 0.0

1.9 1.4 0.1

–0.5 –0.4 –0.7

–1.8 –1.6 –2.9

0.1** 0.3*** 0.3***

0.5 1.1 1.3

–0.2 –0.6* 0.0

–0.8 –2.5 0.1

Constant 11.1 45.8 8.4*** 34.7

Total explained / unexplained

11.3*** 46.5 8.0*** 32.8 –2.5*** –10.4 7.6*** 31.1

Total intra-OC / inter-OC

19.3*** 79.3 %

5.0*** 20.7 %

M-F Wage gap 24.4

(a) The differences in log wages were multiplied by 100 to make the table more legible. (b) The detailed decomposition on the inter-OC wage gap is based on a linear model of choices of OC, whereas the overall decomposition (Table 5) is based on a non-linear model (multinomial logit). This change leads to relatively modest differences in the respective contributions of the explained and unexplained components:–2.0 / 7.1 (for the non-linear model) versus –2.5 / 7.6 (for the linear model). Statistical significance: *  p < 0.10; ** p < 0.05; *** p < 0.01 based on 200 bootstrap sample replications. Source: Authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

∑ p j w β

j m _X

j m – X

j wi+∑ p

j w X

j w _β

j m – β

j wi

jj

ˆ ˆ ˆ− − ∑ lnW j m γ

j m _Xom – Xowi+∑ lnW

j m Xow _γ

j m – γ

j wi

j j

ˆ ˆ ˆ

I. BensIdoun, d. TrancarT

52

Pons (2012) whereby most of the returns of characteristics explaining wages are similar for both sexes.

Columns 5 and 6 show that it is women’s level of education – notably the fact that a much smaller proportion of women have no qualifications or a lower-secondary industrial qualification – that would justify their presence in occupations with higher wages than those where they are actually employed (negative contribution). For the unexplained component of the inter-OC wage gap (columns 7 and 8), as for the intra-OC wage gap, most returns are not statistically different for men and women, so most of the gap stems from differences between the estimated constants.

Hence, most of what is sometimes qualified as discrimination is not linked to the fact that the factors explaining wages or choice of OC have higher returns for men (excepting family characteristics, which penalize women), but to gender differences in treatment that extend beyond these factors.

Regarding the influence of preferences and attitudes (Table 6, shaded lines), we observe that optimism and giving priority to one’s career explain 3.3% of the overall wage gap – almost as much as experience. The difference in attitudes to risk, on the other hand, has no influence on the wage gap. 

Overall, preferences and attitudes explain more than 6% of the observed gender wage gap, more than the traditional variables of human capital, experience and educational level, especially since women’s educational advantage should also give them a wage advantage. A similar proportion is found in other studies. In those examined in this article (Section I), non- cognitive factors explain 8.2% of the wage gap at most. Taking account of non-traditional factors linked to individual preferences and attitudes thus makes it possible to reduce the unexplained component by rendering observable what is usually counted as unobservable. Within the unexplained component, only the factor linked to preference for a career gives men a significant advantage over women by facilitating access to well-paid OCs. This unexplained component remains strong however, representing more than 60% of the observed gender wage gap.

3. Comparison with other studies of France

Recent analyses (Bozio et al., 2014; Meurs and Ponthieux, 2006) on the origin of the gender wage gap in France produce results that differ from those presented here for several reasons.

The first concerns the decomposition method used. These two studies are based on an Oaxaca-Blinder decomposition, and therefore consider that the choice of OC is exogenous; in other words, they do not consider the potentially discriminatory nature of occupational segregation. Consequently, the explained part of the wage differences – 76.2% in 2002 for Meurs and Ponthieux (2006) and 71.6% in 2012 for Bozio et al. (2014) – is much higher than the proportion

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

53

found by us (36.1%).(20) As described earlier, if this decomposition method was applied to our data it would likewise result in a much higher explained component (60.3%).

The second reason is linked to the populations concerned. While our study is based on young people who left the education system ten years earlier, in 1998, these two studies are based on all wage-earners. Our results therefore concern a younger population. In fact, if our results are compared with those of the two other two studies using the same decomposition method (Table 7), we observe that working hours and occupational characteristics are the two main sources of explained wage gaps in all cases, but that they are smaller in our own estimation.

In terms of working hours, the higher contribution observed in these studies (48% and 44.3 % versus 36.1 % in ours) is linked to the fact that they take differences in weekly working hours into account in addition to differences in percentage of full-time working. Differences in occupational characteristics, for their part, represent between 30% (Bozio et al., 2014) and 34 % (Meurs and Ponthieux, 2006) of wage differences, versus 26% in our study, due to the age differences of the populations concerned. As the individuals in our sample are younger, the differences in the structure of employment are smaller than for

(20) Or the 35.6% obtained by Chamkhi and Toutlemonde (2015), resulting from the use of a decomposition that takes account of the partially discriminatory nature of occupational segregation.

Table 7. Comparison of results for France: Oaxaca-Blinder decomposition of the monthly gender wage gap(a) (%)

Study Meurs and Ponthieux (2006) Bozio et al. (2014) Bensidoun and Trancart

Observation year 2002 2012 2008

Explained gap 76.2 71.6 60.3

Experience 0.4 1.1 4.9

Education –6.0 –4.1 –8.7

Speciality – – 3.1

Occupational characteristics 33.7 30.2 25.8

Working hours 48.0 44.3 36.1

Family characteristics – – –3.4

Work preferences and attitudes

– – 2.4

Unexplained gap 27.4 25.6 39.7

Selection effect –2.4 2.8 –

Total 100 100 100

Wag gap (log) 0.252 0.281 0.244

(a) The variables used in Bozio et al (2014) and Meurs and Ponthieux (2006) are identical but different from ours. Experience is real in our study but potential in the two others. Occupational characteristics, occupational cate- gories (OC) and job characteristics overlap, with the exception of managerial functions and firm size which are present in our study only. Working hours include the percentage of full-time working in all three studies, and hours per week only in 2002 and 2012. Given that female labour force participation is high in our study (92%) because the women in the sample are relatively young, selection linked to female labour force participation was not used as the results are not significant. Sources: Bozio et al. (2014), Meurs and Ponthieux (2006), and authors’ calculations based on CÉREQ data, Génération 98 à 10 ans.

I. BensIdoun, d. TrancarT

54

the population in general. The more negative contribution of educational level in our study is, here too, linked to the differences in age of the study populations and reflects women’s strong investment in education. Differences in the contribution of experience doubtless reflect differences in the ways this variable is measured, i.e. real experience in our study versus potential experience in the two others.

In sum, the differing results for the scale of the explained component of gender wage gaps are due essentially to the decomposition method used, differences in the way working hours are taken into account and differences in the populations concerned. In all three studies, however, the characteristics which contribute most to this explained gender wage gap are the same, namely gender differences in working hours and in occupational characteristics.

Conclusion

Differences between men’s and women’s preferences and attitudes have been put forward as one of the reasons behind the stagnation of the gender wage gap over the last two decades. Based on data from the United States, Australia and the Netherlands, it has been shown that in most cases, the wage gap is partly explained by differences in personality traits or in preferences. In France, no surveys have been conducted to measure individual preferences or personality traits in as much detail as in other countries. The data used here nonetheless provide a number of pointers, and we hope that these new insights will attract sufficient interest to prompt the inclusion of questions on preferences and attitudes in future surveys.

First insight: ten years after leaving the education system, the fact that women’s wages are 21.6% below those of men is only 20% attributable to their presence in different occupational categories, and 80% to the fact that within an identical OC, their wages are lower than those of their male counterparts.

Second insight: while 40% of the wage difference can be explained by gender differences in characteristics, more than 60% remains unexplained. These differences in characteristics should result in female wages that are just 8% below those of men.

Third insight: differences in preferences and attitudes – optimism, giving priority to one’s career and appetite for risk – matter (6.3% of the total wage gap, i.e. almost twice as much as experience), with as much (direct) influence on wages as on choice of OC. Indeed, regarding the choice of OC, differences in career priorities and attitudes exert a significant influence, in a direction which explains why men opt for better-paid OCs than women.

While these characteristics reduce the unexplained component of the gender wage gap, it nonetheless remains large, especially since wage differences across OCs are generally unjustified, unlike differences within OCs, which

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

55

are partly explained by gender differences in working hours, occupational characteristics or preferences and attitudes.

Among the unexplained components of wage gaps, only the differences in returns of family characteristics are significant (to women’s disadvantage), which signifies that most of these unexplained differences are not explained by the specific variables identified in the analysis (the explanatory variables selected to determine wages and occupational choices). The unexplained differences are thus not linked to men making better use than women of their characteristics, i.e. their qualifications, their experience, their occupational characteristics or their work preferences and attitudes, but to unexplained reasons which result in men receiving higher wages.

Once potential discrimination against women has been eliminated, these gender differences in preferences and attitudes, like the differences that explain the largest share of their wage gaps – time spent in the labour force or managerial responsibilities – reflect the differences probably stemming from the gendered roles attributed to each sex. In this context, Akerlof and Kranton (2000) have shown that the identity of individuals, their desire to comply with the social norms prevailing in their social group, may influence their economic decisions via the utility associated with them.(21) Hence, beyond questions of discrimination which, while important in our results, must always be viewed with caution in this type of exercise, our work points to the importance of taking measures to change mentalities. Government policies designed to deconstruct gender prejudice and to foster gender equality from an early age are key to progress in this area.

(21) Or the disutility that would result from behaviour that deviates from the norms of the group to which the individual belongs.

I. BensIdoun, d. TrancarT

56

RefeRences

aaronson Daniel, suLLIVan Daniel g., 1998, “The decline of job security in the 1990s: Displacement, anxiety, and their effect on wage growth”, Federal Reserve Bank of Chicago, Economic Perspectives (First Quarter), pp. 17-43.

aHn Taehyun, 2010, “Attitudes toward risk and self-employment of young workers”, Labour Economics, 17(2), pp. 434-442.

aKerLof Georges, KranTon Rachel, 2000, “Economics and identity”, Quarterly Journal of Economics, 115(3), pp. 715-753.

BensIDoun Isabelle, TrancarT Danièle, 2015, “Écarts de salaires hommes-femmes : quels rôles des caractéristiques non cognitives ?”, Document de travail CEE, 177, 53 p.

BerTranD Marianne, 2010, “New perspectives on gender”, in Ashenfelter O., Card D. (eds.), Handbook of Labor Economics, vol. 4b, pp. 1545-1592.

BLInDer Alan Stuart, 1973, “Wage discrimination: Reduced form and structural estimates”, The Journal of Human Resources, 8(4), pp. 436-455.

BonIn Holger, DoHmen Thomas, faLK Armin, Huffman David, sunDe Uwe, 2007, “Cross sectional earnings risk and occupational sorting: The role of risk atti- tudes”, Labour Economics, 14(6), pp. 926-937.

BrInBaum Yaël, Tr anc arT Danièle, 2015, “Educational pathways and gender differences at labor market entr y in France”, in Blossfeld H.-P., Buchhol z S., Skopek J., Triventi M. (eds.), Gender, Education and Employment: An International Compar ison of School-to-Work Transitions, Cheltenham, UK, Edward Elgar, 416 p.

BroWn Randall. s., moon Marilyn, zoLoTH Barbara s., 1980, “Incorporating occupational attainment in studies of male/female earnings differentials”, The Journal of Human Resources, 15(1), pp. 3-28.

BroWn Sarah, DIeTrIcH Michael, orTIz-nuñez Aurora, TayLor Karl, 2011, “Self-employment and attitudes towards risk: Timing and unobserved heterogeneity”, Journal of Economic Psychology, 32(3), pp. 425-433.

BozIo Antoine, DormonT Brigitte, garcIa-PeñaLosa Cecilia, 2014, “Réduire les inégalités de salaires entre femmes et hommes”, Notes du Conseil d’analyse économique, 17, 12 p.

camPBeLL David, carruTH Alan, DIcKerson Andrew, green Francis, 2007, “Job insecurity and wages”, Economic Journal, 117(518), pp. 544-566.

cHaInTreuIL Lydie, ePIPHane Dominique, 2013, “‘Les hommes sont plus fonceurs mais les femmes mieux organisées’ : quand les recruteur-e-s parlent du sexe des candidat-e-s”, Bref Céreq, 315, 4 p.

cHamKHI Amine, TouTLemonDe Fabien, 2015, “Ségrégation professionnelle et écarts de salaire femmes-hommes”, Dares Analyses, 082, 12 p.

cHarDon Olivier, 2001, “Les transformations de l’emploi non qualifié depuis vingt ans”, Insee première, 796, 4 p.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

57



coBB-cLarK Deborah a., Tan Michelle, 2011, “Non cognitive skills, occupational attainment and relative wages”, Labour Economics, 18(1), pp. 1-13.

coLomBIer Nathalie, DenanT-BoemonT Laurent, LoHeac Youenn, mascLeT David, 2008, “Risk aversion: An experiment with self-employed workers and salaried workers”, Applied Economics Letters, 15(10), pp. 791-795.

cramer Jan Salomon, HarTog Joop, JonKer Nicole, Van Praag c. Mirjam, 2002, “Low risk aversion encourages the choice for entrepreneurship: An empirical test of truism”, Journal of Economic Behavior and Organization, 48, pp. 29-36.

eKeLunD Jesper, JoHansson Edvard, JärVeLIn Marjo-Riitta, LIcHTermann Dirk, 2005, “Self-employment and risk aversion – evidence from psychological test data”, Labour Economics, 12(5), pp. 649-659.

esWaran Mukesh, 2014, Why Gender Matters in Economics, Princeton University Press, 408 p.

faLTer Jean-Marc, WenDeLsPIess cHáVez Juárez Florian, 2012, “Can gender traits explain job aspiration segregation?”, University of Genova, Mimeo.

fILer Randall, 1983, “Sexual differences in earnings: The role of individual persona- lities and tastes”, The Journal of Human Resources, 18(1), pp. 82-98.

fILer Randall, 1986, “The role of personality and tastes in determining occupational structure”, Industrial and Labor Relations Review, 39(3), pp. 412-424.

forTIn Nicole, 2008, “The gender wage gap among young adults in the United States”, The Journal of Human Resources, 43(4), pp. 884-918.

groVe Wayne a., Hussey Andrew, JeTTer Michael, 2011, “The gender pay gap beyond human capital: Heterogeneity in noncognitive skills and in labor market tastes”, The Journal of Human Resources, 46(4), pp. 827-874.

HaKIm Catherine, 2004, “Lifestyle preferences versus patriarchal values: Causal and non-causal attitudes”, in Giele Zollinger Janet, Holst Elke, Changing Life Patterns in Western Industrial Societies, Oxford, Elsevier, pp. 69-91.

Ham Roger, JunanKar P.n. Raja, WeLLs Robert, 2009, “Occupational choice: Personality matters”, IZA discussion paper, 4105, 38 p.

Jann Ben, 2008, “The Blinder-Oaxaca decomposition for linear regression models”, The Stata Journal, 8(4), pp. 453-479.

Jung Seeun, 2013, “The gender wage gap and sample selection via risk attitudes”, Paris School of Economics, Mimeo, 64 p.

JoHn Katrin, THomsen Stephan L., 2012, “Heterogeneous returns to personality: The role of occupational choice”, University of Hanover, Mimeo, 41 p.

Lee Lung-Fei, 1983, “Generalized econometric models with selectivity”, Econometrica, 51(2), pp. 507-512.

mascLeT David, coLomBIer Nathalie, DenanT-BoemonT Laurent, LoHeac Youenn, 2009, “Group and individual risk preferences: A lottery-choice experiment with self-employed and salaried workers”, Journal of Economic Behavior and Organization, 70(3), pp. 470-484.

meng Xin, meurs Dominique, 2001, “Différences de structure des emplois et écart salarial entre hommes et femmes en France”, Économie et prévision, 148(2), pp. 113-126.

meron Mong, oKBa Mahrez, VIney Xavier, 2006, “Les femmes et les métiers : vingt ans d’évolutions contrastées”, INSEE, Données sociales, La société française, 10 p.

meurs Dominique, PonTHIeux Sophie, 2006, “L’écart de salaires entre les femmes et les hommes peut-il encore baisser ?”, Économie et statistique, 398-399, pp. 99-129.

I. BensIdoun, d. TrancarT

58

mueLLer Gerritt, PLug Erik s., 2006, “Estimating the effect of personality on male- female earnings”, Industrial and Labor Relations Review, 60(1), pp. 3-22.

nyHus Ellen K., Pons Empar, 2012, “Personality and the gender wage gap”, Applied Economics, 44(1), pp. 105-118.

oaxaca Ronald, 1973, “Male-female wage differentials in urban labor markets”, International Economic Review, 14(3), pp. 693-709.

oaxaca Ronald L., ransom Michael, 1999, “Identification in detailed wage decom- positions”, The Review of Economics and Statistics, 81(1), pp. 154-157.

reILLy Barry, 1991, “Occupational segregation and selectivity bias in occupational wage equations: An empirical analysis using Irish data”, Applied Economics, 23(1), pp. 1-7.

sImonneT Véronique, 1996, “Mobilité professionnelle et salaire : des différences entre hommes et femmes”, Économie et statistique, 299(1), pp. 59-71.

yun Myeong-Su, 2005, “A simple solution to the identification problem in detailed wage decompositions”, Economic Inquiry, 43(4), pp. 766-772.

Career ChoiCes and the Gender Pay GaP: the role of work PreferenCes and attitudes

59

Isabelle BensiDoun, Danièle TranCart • cAreer choices AnD the genDer pAy gAp: the role oF work preFerences AnD AttituDes

The gender wage gap has barely narrowed over the last two decades. This may be partly attributable to gender differences in work preferences and attitudes. Drawing on data from the “Génération 1998 à 10 ans survey” conducted by CÉREQ in France, this article examines the potential roles of career priorities, appetite for risk or optimism about future career prospects on gender wage gaps. As these factors are liable to influence wages, but also career choices, this study uses a wage gap decomposition that takes them into account. It also factors in the potentially discriminatory nature of occupational segregation. Differences in preferences and attitudes account for 6.3% of the total wage gap, nearly twice as much as experience. They also reduce the unexplained component of wage differences, which nonetheless remains large. Ten years after leaving the education system, although the gender wage gap should be just 8%, women’s wages are 21.2% below those of men.

Isabelle BensiDoun, Danièle TranCart • choix proFessionnels et écArts De sAlAires entre hommes et Femmes : le rôle Des DiFFérences De préFérences et D’AttituDes FAce Au trAvAil

La réduction des écarts de salaires entre les hommes et les femmes est depuis maintenant deux décennies au point mort. Le fait que les unes et les autres se distinguent en matière de préférences et d’attitudes face au travail constitue une des raisons qui pourrait l’expliquer. Dans cette étude, on examine à partir de l’enquête Génération 1998 à 10 ans réalisée par le Céreq en France, le rôle que les préférences en matière de carrière, l’attitude face au risque ou le rapport à son avenir professionnel peuvent avoir sur les écarts de salaires. Comme ces facteurs sont susceptibles d’influencer non seulement les salaires mais aussi les choix professionnels, une décomposition des écarts de salaires qui en tient compte est retenue ici. Celle-ci permet en outre de prendre en considération le caractère potentiellement discriminatoire de la ségrégation professionnelle. Les différences de préférences et d’attitudes comptent pour 6,3  % de l’écart de salaires total, soit près de deux fois plus que l’expérience. Elles permettent de réduire la composante inexpliquée des écarts de salaire qui reste toutefois importante. Dix ans après la sortie du système éducatif, le salaire des femmes, inférieur de 21,2 % à celui des hommes, ne devrait en effet l’être que de 8 %.

Isabelle BensiDoun, Danièle TranCart • opciones proFesionAles y DesiguAlDAD sAlAriAl entre homBres y mujeres: el pApel De lAs DiFerenciAs en lAs preFerenciAs y en lAs ActituDes Frente Al trABAjo

La reducción de la desigualdad salarial entre los hombres y las mujeres está, desde hace ahora dos decenios, paralizada. El hecho de que unos y otros se distinguen en las preferencias y las actitudes frente al trabajo podría ser una de las razones de esa situación. A partir de los datos de la encuesta Generación 1998 a los diez años, realizada por el Cereq (Centre d’études et de recherches sur les qualifications), examinamos aquí el papel que las preferencias de carrera, la actitud frente al riesgo o la relación al futuro profesional pueden tener sobre la desigualdad salarial. Como estos factores son susceptibles de influir no solo en los salarios sino también en las opciones profesionales, hemos operado una descomposición de las diferencias de salario que tenga cuenta de ello. Esta descomposición permite además tomar en consideración el carácter potencialmente discriminatorio de la segregación profesional. Las diferencias en las preferencias y en las actitudes dan cuenta del 6,3% de la diferencia total entre los salarios, o sea casi dos veces más que la experiencia profesional. Dichas diferencias permiten reducir la componente inexplicada de la desigualdad salarial, la cual continua siendo no obstante importante. Diez años después de la salida del sistema educativo, el salario de las mujeres, inferior de 21,2 % al de los hombres, no debería serlo que de 8 %.

Keywords: gender wage gap, Brown-Moon and Zoloth wage decomposition, work preferences and attitudes, occupational segregation, discrimination

Translated by Catriona Dutreuilh

I. BensIdoun, d. TrancarT

60

Copyright of Population (16342941) is the property of Institut National d'Etudes Demographiques and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

EEOC Striving to Make Equal Pay for Equal Work the New Norm

Charlotte A. Burrows (interviewed by Patrick Dorrian)

Charlotte A. Burrows (D) has served as a commissioner of the Equal Employment Opportunity Commission since Jan. 13. In an interview with BBNA, she discussed pay dis- crimination based on sex, race and other characteristics, and highlighted the commission’s commitment to and efforts toward rooting out this form of workplace bias. Bur- rows says employers can head off compliance issues by, among other things, making sure managers know they’re the first line of defense.

BBNA:

You have publicly addressed the issue of the gen- der pay gap and the Equal Employment Opportu- nity Commission’s efforts to close the gap. What is the gender pay gap currently?

Burrows: In 1963, when President Kennedy signed the Equal Pay Act, the gender pay gap was 41 cents—that is, women earned on average just 59 cents to the dollar when their salaries were com- pared to those of men. Although there’s been prog- ress since, it’s been far too slow, and we’ve only been able to cut the pay gap approximately in half in all these years. So according to recent census data, in 2014, more than 50 years after passage of the Equal Pay Act, women on average still earned just 79 cents for every dollar earned by men. The comparisons are even worse for women of color.

The persistence of such a large pay gap more than 50 years after passage of the Equal Pay Act makes clear that a great deal more must be done to address it. In fact, a recent study from the Institute for Women’s Policy Research estimates that we won’t achieve gender pay equity for another 44 years, when nearly every woman now in the work- force will have retired.

Clearly, we can’t afford to wait that long. The damage caused by pay discrimination is far- reaching, and addressing the ongoing gender wage gap is a national priority for the EEOC and a top priority for me personally. It hurts not only women who take home less money than they have rightfully

earned, but it hurts the well-being of families, em- ployers, and our nation’s economy. The wage gap af- fects all stages of earnings, widening over women’s careers and impacting Social Security and retire- ment.

BBNA: What are the primary causes of gender- based unequal pay?

Burrows: Differences in pay between men and women result from a variety of factors, but unfortu- nately, workplace discrimination remains a signifi- cant part of the problem. The commission takes se- riously its responsibility to address the part of the pay gap that results from employment discrimina- tion.

The discrimination that contributes to the pay gap of course includes direct discrimination in wages, such as when women are offered lower start- ing salaries than men for doing the same job. But other forms of discrimination can affect women’s paychecks just as severely.

A woman who is denied hire, training or a de- served promotion, steered into a lower paying job than her qualifications warrant, or has to change jobs because of sexual harassment or pregnancy dis- crimination, sees the results in her paycheck, just as do those who are paid less for doing the same job as their male counterparts.

All of these dynamics limit women’s full economic potential and contribute to gender pay disparities. Sadly, we continue to see all of these fact patterns on a regular basis in the charges of discrimination filed with the commission. Just in the past few months, the EEOC has filed several pay discrimination law- suits, including a lawsuit against SOCI Petroleum/ Santmyer Oil Co., alleging that the company violated federal equal pay laws by paying a female human re- sources manager less than a male predecessor for performing substantially equal work.

BBNA: Is it known how much of the gender pay gap is directly attributable to sex discrimination?

10 (No. 1)

Interview

January 2016 COPYRIGHT � 2016 BY THE BUREAU OF NATIONAL AFFAIRS, INC. HRF ISSN 1059-6038

Burrows: Although it is difficult to say exactly how much of the gender wage gap is caused by sex discrimination, research has shown that even when controlling for factors like experience and education, a significant portion of the wage gap remains unex- plained. As discussed earlier, for many women in the workforce, particularly those who pursue tradition- ally male careers, their pay is depressed to some de- gree by discrimination in a variety of areas.

An additional challenge in addressing pay dis- crimination is that pay differentials are notoriously difficult to detect, because salaries generally are not public. Because many women have no way of know- ing when they are shortchanged, it’s hard for them to contest pay discrimination on their own. That’s why it’s so important for the EEOC to focus on this.

Some workplaces have official policies that ban employees from disclosing or inquiring about their own wages or the wages of a co-worker. In fact, ac- cording to a 2010 survey by the [IWPR], more than 60 percent of private sector workers reported that their employer either prohibits or strongly discour- ages employees from discussing their wages.

Such punitive pay secrecy policies send the mes- sage that there’s something to hide, and they can be a significant obstacle to equal pay for women. They may perpetuate pay discrimination by making it dif- ficult for individuals to learn about unlawful dispari- ties and by leaving workers afraid to inquire.

Last year, President Barack Obama signed an ex- ecutive order banning federal contractors from re- taliating against employees who disclose their own wages were key steps toward addressing this prob- lem.

The rule covers approximately 26 million people employed by companies that contract with the fed- eral government. It’s my hope that the rule will prompt companies to take action proactively to end pay secrecy policies, and to reexamine pay policies to ensure that men and women are paid equally for equal work.

BBNA: Studies show that the gender pay gap is even worse for black and Hispanic women, and that

a pay gap also exists between white men and men of other races. Is the EEOC also focused on this?

Burrows: Absolutely. The EEOC is committed to rooting out pay discrimination in all its forms.

Unfortunately, as you note, the pay gap is even more substantial for many women of color. Accord- ing to recently released census data, in 2014, African American women on average earned roughly 60 cents for every dollar earned by white, non-Hispanic men each year; the annual earnings of Latinas were approximately 55 cents for every dollar earned by white men.

There is also an extremely troubling racial pay gap that depresses the wages of men of color. For in- stance, in 2013, African American men earned ap- proximately 75 cents for every dollar earned by white men, and Hispanic men earned roughly 67 cents on the dollar.

The EEOC vigorously investigates claims of race- based pay discrimination. For example, in EEOC v. Mitsuwa Corporation, the commission alleged that a group of Hispanic employees was discriminated against based on their national origin. Mitsuwa rou- tinely paid Hispanic employees less than non- Hispanics doing the same work because of their na- tional origin. The case settled for $250,000 and in- junctive relief.

We’ve also pursued claims of wage discrimination that result from other components that make up earnings—such as overtime pay or bonuses. For ex- ample, in EEOC v. Baird Tree, the commission al- leged that the company maintained a policy and practice of failing to pay Hispanic employees over- time pay while paying non-Hispanic workers such wage premiums. Our investigation also found that the company threatened to fire employees after they complained about the wage discrimination. The case settled for back pay, compensatory damages and in- junctive relief.

BBNA: What are the primary causes of race- based unequal pay?

(No. 1) 11

Interview

HRFOCUS ISSN 1059-6038 BNA January 2016

Burrows: As with the gender wage gap, there’s no doubt that discrimination contributes, at least in part, to racial disparities in earnings.

Racial discrimination in wages is unfortunately a long-standing problem in this country. Slavery was the most obvious example, but after Reconstruction, race-based wage discrimination has continued to persist in various forms and to varying degrees un- til the present day.

In April 1965, just a few short months before the commission first opened its doors, Dr. Martin Luther King, Jr., noted that 88 percent of African Ameri- cans earned less than $5,000 a year, significantly less than the average earnings of the white population. At the time, Latinos and Asian Americans also faced significant limitations in employment opportunity.

And as Robert Kennedy noted in a 1963 speech to the National Congress of American Indians, Native Americans’ income was between a third and a quar- ter of the American average at the time.

There has been enormous progress since the 1960s, but we haven’t completely solved the prob- lem. Some of this discrimination occurs long before men and women of color enter the workforce—in education, training opportunities, and through con- tact with the criminal justice system. But the charges of discrimination we see, and the evidence developed in our investigations, make clear that em- ployment discrimination in all forms is also a signifi- cant factor.

The racial pay gap is affected by discriminatory salary setting, promotion decisions, performance ap- praisals, assignment and training opportunities, and the relegating of women and workers of color to lower paying jobs.

BBNA: Is it known how much of the racial pay gap is directly attributable to racial discrimination?

Burrows: While it is difficult to quantify an exact number, we see from our investigations that race discrimination in compensation, hiring, and promo- tion decisions continues to impact pay. Research has

made it clear that the racial wage gap is at least par- tially explained by labor market discrimination.

On average, white workers are in higher paying jobs than their minority counterparts, but even when white workers and workers of color are in the same field—such as retail—racial income gaps per- sist. According to a research study by Demos and NAACP, African American and Latino retail sales- persons make just 75 cents for every dollar earned by white employees. Moreover, even when African Americans work in high-skill fields they continue to earn less than their white peers.

Women of color, in particular, face multiple barri- ers to achieving equal pay. According to a new study from LeanIn.Org and McKinsey & Company, Afri- can American, Latina, and Asian women in corpo- rate America are more interested in being promoted than white employees of both genders but often do not receive the mentoring and networking opportu- nities needed to advance.

For example, in June 2014, the EEOC settled a lawsuit filed against Chapman University alleging that the university discriminated against an assis- tant professor of marketing by denying her tenure and promotion to associate professor because of her race. The EEOC’s investigation determined that the assistant professor was the first black professor to have been allowed to apply for tenure at the univer- sity’s School of Business and Economics, and she was subjected to a higher standard for obtaining tenure and promotion than her non-black peers.

Cases like this illustrate that bias continues to op- erate to deny women of color these opportunities, despite their strong interests.

BBNA: What is occupational segregation and how big of a factor is it in gender- and race-based un- equal pay?

Burrows: Occupational segregation occurs when workers are steered into particular positions or in- dustries based on their race, national origin, gender, or other prohibited bases. Unfortunately, occupa- tional segregation continues to occur across a vari-

12 (No. 1)

Interview

January 2016 COPYRIGHT � 2016 BY THE BUREAU OF NATIONAL AFFAIRS, INC. HRF ISSN 1059-6038

ety of industries, and it remains a major factor in race and gender disparities in pay.

Women are often segregated into traditionally fe- male jobs that pay less than traditionally male jobs. Indeed, it’s estimated that over half of women are employed in lower-paying sales, service, and admin- istrative support positions, and many of these women are women of color.

For example, in 2011, an EEOC systemic investi- gation found that the Western Sugar Cooperative denied women training and promotions, gave them less desirable work assignments, segregated posi- tions by gender, denied women year-round employ- ment, and paid lower wages to women. Western Sugar agreed to settle the matter by paying a class of women $550,000 and furnishing significant reme- dial relief.

Similarly, in restaurants and other service indus- tries, we see workers of color being assigned ‘‘back of the house’’ positions that do not involve interac- tion with customers. Moreover, these segregated, ‘‘back of the house’’ positions come with substan- tially less pay, either in salary, tips, or other forms of compensation.

The employment laws do not allow employers to rely on discriminatory preferences of co-workers, customers, or clients to justify this kind of segrega- tion of employees. Employers also cannot rely on a specific ‘‘appearance’’ or ‘‘image’’ policy that serves as a proxy for discriminatory customer preference to justify segregating employees.

This is why two of the commission’s six national priorities address issues related to occupational seg- regation: 1) eliminating barriers in recruitment and hiring and 2) protecting immigrant, migrant and other vulnerable workers through combating, among other things, job segregation.

In recent years, the commission has brought cases challenging practices that either denied hire to women and minorities, or limited them to low- ranking positions with little chance of advancement. We’ve had some successes, such as the establish- ment in 2009 of a $19 million claimant fund in a case

in which the EEOC alleged that the Outback Steak- house restaurant chain had a pattern or practice of failing to promote women to the higher-level profit- sharing management positions.

In addition, in 2014, McCormick & Schmick’s agreed to pay $1.3 million and provide significant eq- uitable relief to settle a pattern-or-practice race dis- crimination suit filed by the EEOC, which alleged that the company excluded African American appli- cants from front-of-the-house positions at its Balti- more locations in violation of Title VII.

BBNA: What are some of the causes of occupa- tional segregation?

Burrows: Occupational segregation often occurs as the result of stereotypes about what men and women can and should do, and what their work is worth. It’s often for these reasons that people in a protected class are channeled into jobs with lower pay or lower long-term pay potential than other similar jobs.

For example, one of the causes of occupational segregation of women is hiring discrimination in fields traditionally dominated by men, such as truck- ing, mining, oil and gas, construction, and warehouse work, based on a generalized view of what consti- tutes ‘‘women’s work.’’

Employers in such fields may make discrimina- tory assumptions that women cannot perform this work or don’t belong in these workplaces.

Occupational segregation also occurs when infor- mal networks function to channel job opportunities exclusively to those of the same race or gender as current managers or employees, reflecting manag- ers’ preferences for employees who have back- grounds similar to their own.

The commission has vigorously enforced discrimi- nation against females in traditionally male occupa- tions, including the following recent resolutions:

• EEOC v. New Prime, Inc., 42 F. Supp. 3d 1201, 124 FEP Cases 227 (W.D. Mo. 2014) (holding in August 2014 that the policy of re-

(No. 1) 13

Interview

HRFOCUS ISSN 1059-6038 BNA January 2016

quiring same-sex trainers for truck drivers is discriminatory).

• EEOC v. Unit Drilling Co., N.D. Okla., No. 13-147, settlement approved 4/21/15 (settled for $400,000 in 2015 where the EEOC alleged defendant failed to hire women to work on oil rigs).

In addition, the commission has pending systemic discrimination litigation against Mavis Discount Tire Inc., alleging that the company did not hire women for manager, assistant manager, tire installer, and similar jobs because of their sex, and another case alleging that Performance Food Group refused to hire women for operative positions such as selector, receiving clerk, driver, and yard jockey. We also have recently brought litigation against Workplace Staff- ing Solutions, a Louisiana staffing firm, alleging that it failed to hire at least 34 qualified women for trash- can collector jobs because of their sex.

BBNA: What steps has the EEOC undertaken to close the gender-based pay gap?

Burrows: The EEOC’s Strategic Enforcement Plan for Fiscal Years 2013-2016 prioritizes enforce- ment of equal pay laws to help close the gap.

The EEOC also serves as a key member of the National Equal Pay Enforcement Task Force, a fed- eral government initiative focused on ending the gender pay gap.

We have recovered over $85 million for victims of sex-based wage discrimination since 2010 through our administrative enforcement process alone.

The commission also aggressively pursues pay discrimination violations under Title VII and the Equal Pay Act in federal court. For example, in ad- dition to the other cases I’ve mentioned:

• In July 2014, Royal Tire Inc., a commercial and retail tire company based in St. Cloud, Minn., agreed to pay $182,500 and be subject to a detailed consent decree to settle a lawsuit filed by the EEOC alleging that the company violated the Equal Pay Act and Title VII by

paying its female human resources director $35,000 less per year than her male predeces- sor, despite the fact that her duties were equal to her predecessor’s and being performed un- der similar working conditions.

• In April 2014, Checkers, a fast food restau- rant franchise, agreed to pay $100,000 and fur- nish significant equitable relief to settle a law- suit filed by the EEOC alleging that the fran- chise paid female shift managers and female cashiers/sandwich makers lower wages than their male counterparts for substantially equal work and suppressed their wages through dis- criminatory job assignments in violation of the Equal Pay Act and Title VII.

• In February 2014, Extended Stay Hotels agreed to pay four female guest services rep- resentatives $75,800 and provide significant equitable relief to settle a lawsuit filed by the EEOC alleging that the company unlawfully paid the female employees lower wages than those paid to male guest services representa- tives for performing equal work in violation of the Equal Pay Act and Title VII.

• In September 2013, the EEOC settled a law- suit alleging that the Worcester County De- partment of Liquor Control violated the Equal Pay Act by paying three female clerks lower wages than male retail clerks, even though they were doing substantially equal work un- der similar working conditions. Worcester County, Md., agreed to settle the matter by paying $60,000 and agreeing to provide signifi- cant equitable relief.

• In May 2012, Health Management Group agreed to pay $260,000 to settle an Equal Pay Act and Title VII lawsuit filed by the EEOC, which alleged that two female directors of franchise development were paid less than a male colleague who performed substantially equal work.

BBNA: What steps has the agency undertaken to close the race-based pay gap?

14 (No. 1)

Interview

January 2016 COPYRIGHT � 2016 BY THE BUREAU OF NATIONAL AFFAIRS, INC. HRF ISSN 1059-6038

Burrows: The EEOC remains committed to en- forcing Title VII and targeting compensation sys- tems and other unlawful practices that discriminate based on race. The commission’s Strategic Enforce- ment Plan identifies the elimination of barriers in re- cruitment and hiring, along with enforcing equal pay laws, in its list of six national priorities.

The agency continues to investigate and, where violations of Title VII are found, litigate charges al- leging race discrimination in compensation, recruit- ment, hiring and other barriers to equal pay.

For example, the EEOC had sued the operators of a Hampton Inn in Indianapolis for firing African American housekeepers because of their race and in retaliation for complaints about race discrimination. We also charged that the hotel paid lower wages to black housekeepers and excluded black housekeep- ing applicants on a systemic basis. In September 2012, the court entered a consent decree providing $355,000 in monetary relief and injunctive relief. In March 2015, the hotel operators were held in con- tempt by a federal judge for violating the consent decree, which led to, among other remedies, an in- creased back pay order for affected employees, ad- ditional fines, and a two-year extension of provisions of the original consent decree.

In another example of our work to combat dis- criminatory practices that contribute to the racial wage gap, the EEOC collaborated with the Depart- ment of Justice in investigating and attempting to conciliate claims that the South Dakota Department of Social Services engaged in a pattern or practice of canceling vacancy announcements rather than hire well-qualified Native American applicants into certain state government positions. Such wholesale exclusion of persons of a particular race from certain jobs helps fuel the racial pay gap, by limiting em- ployment opportunities for persons of color. The case is currently in litigation.

BBNA: You’ve also mentioned publicly that em- ployers can play a proactive role in helping the EEOC close the gender- and race-based pay gaps. What do you see employers doing in this area?

Burrows: Although litigation is important, the EEOC sees outreach and technical assistance to both employers and employees as crucial to protect- ing American workers, because they can help pre- vent wage discrimination before it happens.

We value those partners in corporate America that understand that equal employment opportunity is in companies’ best interests. We want to help em- ployers do their jobs and reach the best outcomes for their employees.

One way employers can make a difference is to be proactive in making equal pay a priority. Employers shouldn’t wait for the EEOC or other agencies to tell them there’s a problem. It’s in everyone’s inter- ests if employers take the initiative to examine their own payrolls and compensation practices to ensure they are fair.

Specifically, employers should take three steps. First, they should make clear to their workforce that pay equity is a priority at the very highest levels of the corporation by having top executives articulate a clear commitment to that goal. If managers know fair pay is important to corporate leaders, they will be motivated to prevent pay disparities before they happen.

Second, employers should make sure their house is in order. They can do this by examining payrolls, or through statistical analyses for large employers, to determine whether pay disparities exist between men and women who are doing the same job. Be- cause the effects of an unjust pay disparity at the start of one’s tenure can have ripple effects through- out a woman’s career, when possible, this sort of pay analysis should be conducted when determining starting salaries.

Employers should also periodically review com- pensation for purposes of determining pay increases and should designate specific individuals with the re- sponsibility to monitor pay practices. And of course, employers that discover they have an unjustified pay disparity in part of the company should fix it.

(No. 1) 15

Interview

HRFOCUS ISSN 1059-6038 BNA January 2016

The third proactive step that every company should take is to let employees know that, if they wish, they can voluntarily share pay data with their co-workers. As discussed above, many workplaces have official policies that ban employees from dis- closing or inquiring about their own wages or the wages of a co-worker.

Eliminating these punitive pay secrecy policies may serve the employers’ best interests since these policies prevent employees from sharing information that might avoid litigation in the first place. A worker who suspects discrimination, but is subject to a punitive pay secrecy rule, may have no option other than discovery to learn what her colleagues are paid, for fear of being terminated for asking about pay in informal discussions with co-workers.

BBNA: What can employers do to help bring down the existing level of race-based unequal pay?

Burrows: All of the steps that I discussed for em- ployers with respect to examining gender-based pay disparities apply here as well. It’s important for em- ployers to take proactive steps to ensure that mid- level managers, who often conduct salary negotia- tions with prospective employees, understand that pay equity is a priority for the employer, and that they will be asked to justify racial pay disparities for employees doing the same job.

Employers should also ensure that promotions and entry-level opportunities are advertised broadly, to ensure a diverse applicant pool and avoid limiting opportunities for persons of color only to certain jobs.

Data collected by the commission since 1966 show that African Americans and Latinos have had rela- tively greater success in entering the ranks of pro- fessionals where the requirements for entry are relatively clear and hiring criteria tend to be based on merit.

By contrast, they seem to have had the most dif- ficulty entering positions for which hiring and pro- motion practices rely heavily on personal contacts. This suggests that informal employment practices

disadvantage candidates of color by permitting deci- sion makers’ biases and preferences to influence the process.

Avoiding legal liability isn’t the only reason com- panies should ensure that employment practices do not unfairly exclude qualified persons of color. There’s also a compelling business case for diversity, as many companies have already discovered.

The U.S. is becoming increasingly diverse, and the Census Bureau projects that by 2043, the major- ity of the population will be persons of color. That’s already the case in the states of California, Texas, New Mexico, and Hawaii, and in many major Ameri- can cities, including New York, Washington, Boston, and Miami.

American companies are also increasingly doing business in global markets. This means that employ- ers that recruit, integrate, and retain employees of color will increasingly have an advantage in attract- ing top talent, reaching consumers, and competing in the global marketplace.

My hope is that the EEOC, through the technical assistance and outreach we provide to employers—as well as through litigation, when necessary—can help change enough industry prac- tices to ensure that equal employment opportunity and equal pay for equal work becomes the norm in American workplaces, even as those workplaces be- come more diverse. I’m confident that this is both necessary and possible as the EEOC continues its work in this area. A graduate of Princeton University and Yale Law School, Commissioner Burrows was confirmed by the Senate Dec. 3, 2014, for a term expiring July 1, 2019.

Prior to her appointment, she served as associate deputy attorney general at the Department of Justice, where she worked on a broad range of legal and policy issues, and prior to that as general counsel for civil and constitu- tional rights to Sen. Edward M. Kennedy (D) on the Senate Committee on Health, Education, Labor and Pensions in 2009 and on the Senate Judiciary Committee from 2007 to 2008, and as legal counsel on the Senate Judiciary Com- mittee from 2003 to 2007.

Before arriving on Capitol Hill, Burrows served in the Civil Rights Division’s Employment Litigation Section at the Department of Justice as a trial attorney, special litigation counsel and deputy chief.

16 (No. 1)

Interview

January 2016 COPYRIGHT � 2016 BY THE BUREAU OF NATIONAL AFFAIRS, INC. HRF ISSN 1059-6038

Copyright of HR Focus is the property of Bloomberg BNA and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Pay Secrecy and the Gender Wage Gap in the United States*

MARLENE KIM

Legislators and advocates claim that pay secrecy perpetuates the gender wage gap and that the Fair Labor Standards Act (FLSA) should be amended to outlaw these practices. Using a difference-in-differences fixed-effects human-capital wage regression, I find that women with higher education levels who live in states that have outlawed pay secrecy have higher earnings, and that the wage gap is conse- quently reduced. State bans on pay secrecy and federal legislation to amend the FLSA to allow workers to share information about their wages may improve the gender wage gap, especially among women with college or graduate degrees.

There has been a sea change in the workforce since the Fair Labor Standards Act (FLSA) was passed in 1938. One of the more notable changes is that few women worked for pay when the law was passed, while today, most do.1 To address the needs of the modern workforce, legislation has amended the FLSA to account for these changes. These amendments include the Equal Pay Act of 1963, which mandates equal pay for equal work, regardless of sex. But there have been other proposed amendments that purport to meet the needs of working women. This paper examines one of the least known policy proposals: amending the FLSA in order to outlaw pay secrecy. In this paper, I discuss pay secrecy, its extent, proposed legislation to amend

the FLSA in order to prohibit these practices, and the likely effect of amend- ing the FLSA to outlaw pay secrecy on earnings and the gender wage gap.

*The author’s affiliation is University of Massachusetts Boston, Boston, Massachusetts. Email: Marlene. Kim@umb.edu.

The author offers thanks to the U.S. Women’s Bureau staff and former director for their support, the staff in many of the states that were interviewed for this article, and to the legislators and activists in these states for their helpful information and comments. Thanks to Reagan Baughman, Michael Carr, Bill Dickens, Andrew Houtenville, Ju-Chin Huang, Emily Weimers, Catherine Weinberger, Paul Wolfson, Linus Yamane, and the University of New Hampshire’s Economics Department Seminar for their help, insights, and sugges- tions. Thanks also to Jesse Rothstein and two anonymous reviewers who helped improve, tighten, clarify, and find all the typos and errors in this paper. All remaining errors are the author’s.

1 In 1948, 32 percent of women were in the labor force; in 1938, even fewer were (Federal Reserve Board of St. Louis 2013; Smith 1979;). In contrast, in 2012, 58 percent of women participated in the labor force (U.S. Bureau of Labor Statistics 2013b).

INDUSTRIAL RELATIONS, Vol. 54, No. 4 (October 2015). © 2015 Regents of the University of California Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington

Road, Oxford, OX4 2DQ, UK.

648

Using a natural experiment of states that prohibit pay secrecy compared to those that do not, I examine whether states that outlaw pay secrecy reduce the gender wage gap. I find that in states that have outlawed pay secrecy, earnings for college-educated women are greater, reducing the gender pay gap.

Pay Secrecy, Its Prevalence, and the Law

Pay secrecy includes rules, policies, and practices that prohibit workers from discussing or sharing information about their earnings (Bierman and Gely 2004; Edwards 2005; Gely and Bierman 2003). These include formal policies written in employee handbooks and informal policies conveyed to workers sometime during their employment (Gely and Bierman 2003). Advocates and legislators who have proposed to amend the FLSA by outlawing pay secrecy argue that pay secrecy perpetuates the gender wage gap: if women don’t know what other workers are paid, gender discrimination in earnings can continue. Lilly Ledbetter illustrates this argument. For 20 years, Ledbetter was the

only female supervisor among sixteen male supervisors for Goodyear Tire in Alabama. She earned less than all these men, including some who had less seniority, yet she did not know that she was underpaid because her workplace prohibited employees from discussing their pay. It was only after she received an anonymous note that revealed the earnings of some of these male managers that she realized she was underpaid (Greenhouse 2007; National Women’s Law Center 2013). Ledbetter is not alone. In the United States, most employees are prohibited

from discussing their earnings. According to a survey conducted in 2010, 61 percent of private-sector workers are either formally forbidden or informally discouraged from discussing their pay with their colleagues (Institute for Women’s Policy Research 2010). About one-third of private-sector workers are explicitly forbidden from doing so because of formal rules or policies not to discuss their pay, with another third informally discouraged from doing so (Bamberger and Belogolovsky, 2010; Bierman and Gely 2004; Colella et al. 2007; Edwards 2005; Gely and Bierman 2003). Yet, for most of these workers, these pay-secrecy policies are illegal.

Section 7 of the National Labor Relations Act (NLRA) protects workers in “concerted activities for the purpose of collective bargaining or other mutual aid or protection.”2 The National Labor Relations Board (NLRB), which enforces the NLRA, has consistently ruled that discussions of wages are a

2 29 U.S.C. § 157 (2003).

Pay Secrecy and the Gender Wage Gap / 649

form of “protected concerted activity”; thus, prohibiting discussions of earnings is illegal (Bierman and Gely 2004; Gely and Bierman 2003). The NLRB views sharing information about pay as integral to organizing workers into unions, even if a union-organizing campaign is not in progress.3

Dissatisfaction due to low wages is the grist on which concerted activ- ity feeds. Discord generated by what employees view as unjustified wage differentials also provides the sinew for persistent concerted action. (Jeannette Corp v. NLRB, 532 F.2d 916, 919 [3d Cir. 1976], cited in Gely and Bierman 2003: 128)

Thus, the NLRB has ruled very broadly that employers are in violation of the law if they discourage or prohibit workers from sharing information about their pay (Bierman and Gely 2004; Gely and Bierman 2003). Such prohibi- tions include informal or formal pay-secrecy policies, even if not enforced. Policies that in any way restrict employees from sharing information about their earnings are forbidden, including employers’ preventing employees from opening their paychecks among other workers. The NLRB has ruled that these prohibitions hamper employees’ rights under the NLRA (Bierman and Gely 2004; Gely and Bierman 2003).4 These protections extend to both unionized and nonunion workers. Why is pay secrecy so widespread if it is illegal? First, many employees

are not covered by the NLRA: supervisors and managers are excluded. Hence if Lilly Ledbetter had violated her company’s policy and had asked how much her male counterparts earned, she could have been fired with no legal recourse. Second, most employees do not know that pay secrecy is illegal (Gely and Bierman 2003). Third, the penalties from violating the NLRA are mild, so employers commonly break the law (Bronfenbrenner 1994; Freeman and Medoff 1984; Gely and Bierman 2003). Fourth, many employees favor pay-secrecy policies, in part because the culture in the United States dictates that one does not discuss one’s earnings (Colella et al. 2007). Fifth, employers favor pay secrecy (Gely and Bierman 2003). They believe that morale and productivity would decline, relationships among workers would be strained, and conflict could occur if workers knew how much others earned (see Colella et al. [2007] for a review of

3 An NLRB Board member confirms that “The right of employees to talk to each other about pay is as fundamental as any activity intended to receive NLRA protection, given that pay discussions among disgrun- tled employees are often at the heart of unionization activity” (Bierman and Gely 2004: 169, citing John E. Higgins, an NLRB Board member).

4 The only employer prohibitions the NLRB has allowed involve revealing the entire pay structure, because pay structures are viewed as proprietory (see Gely and Bierman [2003] and Bierman and Gely [2004]).

650 / MARLENE KIM

this literature). As one employer states, “jealously and strife among employ- ees” would prevail if employees knew what others were paid but not the justifications for these wage differentials (Bierman and Gely 2004). Pay secrecy may also prevent employees from leaving their employers for com- panies that offer better pay (Colella et al. 2007; Danziger and Katz 1997). Thus, because most managers believe that pay secrecy is a good policy, the unwritten rule in most workplaces is that employees must keep their mouths shut about their pay (Bierman and Gely 2004; Gely and Bierman 2003).

Pay Secrecy and the Gender Wage Gap

Some scholars argue that pay secrecy can contribute to the gender pay gap (Eisenberg n.d.) because it can help companies “avoid perceptions of unfairness when pay inequities do exist and can minimize claims of discrim- ination” (Colella et al. 2007). Hence, lack of knowledge about what others earn can contribute to the existence of pay discrimination and thus to the gender wage gap. Eisenberg (n.d.) argues that pay transparency is important so that the employer is motivated to establish fair pay systems, and so that employees can monitor, complain about, and remedy any unfair pay. Although market wages are supposed to discipline both workers and employers in compensation, much information is unavailable to workers (such as what their colleagues earn), so market discipline may not work. Instead, without standard salary scales, and with salaries based on previous salary, the gender pay gap can perpetuate (Eisenberg n.d.). In addition, with salaries open to negotiation, women may be underpaid because they do not negotiate as hard as men for higher salaries, in part because if they do, they are seen as too demanding and unpleasant to work with. Thus, for a woman, negotiating over salary may lead to loss of a job offer. Men, how- ever, are able to negotiate for higher salaries without such adverse conse- quences (Eisenberg n.d.). With salaries based on previous earnings, and with women earning less than

men on average and unable to negotiate for higher pay, women will continue to earn less than men (Eisenberg n.d.). Salary transparency, however, would enable women to know what others are earning and to negotiate for similar pay. It would also allow women to know if they are underpaid compared to similar men and to correct these disparities, either informally or through the court system (Eisenberg n.d.).

Pay Secrecy and the Gender Wage Gap / 651

Federal Legislation to Outlaw Pay Secrecy: Amending the FLSA

Recognizing the importance of sharing information on pay for women’s earnings, Congress has introduced twenty-two pieces of legislation that would amend the FLSA to outlaw pay secrecy; however, none have passed. Proposed legislation includes the Paycheck Fairness Act (introduced eighteen times), the Fair Pay Act of 2001 and 2011, the Enhancing Economic Security for Amer- ica’s Working Families Act in 2001, the Fairness and Individual Rights Neces- sary to Ensure a Stronger Society: Civil Rights Act of 2004, and the Wage Awareness Protection Act in 2000.5 The Wage Awareness Protection Act was the only bill in which pay secrecy was the sole content; all the other bills included broader legislation to reduce the gender wage gap (such as mandating comparable worth or harsher penalties for findings of discrimination). These

5 See for example, S. 71: Paycheck Fairness Act. 105th Congress (introduced January 21, 1997) http:// www.gpo.gov/fdsys/pkg/BILLS-105s71is/pdf/BILLS-105s71is.pdf; H.R. 2023: Paycheck Fairness Act (intro- duced June 24, 1997) http://www.gpo.gov/fdsys/pkg/BILLS-105hr2023ih/pdf/BILLS-105hr2023ih.pdf; S 74, 106 Congress Paycheck Fairness Act (introduced January 19, 1999) http://www.gpo.gov/fdsys/pkg/BILLS- 106s74is/pdf/BILLS-106s74is.pdf; H.R. 541 Paycheck Fairness Act (introduced February 3, 1999) http:// www.gpo.gov/fdsys/pkg/BILLS-106hr541ih/pdf/BILLS-106hr541ih.pdf; H.R. 2397 Paycheck Fairness Act (introduced June 30, 1999) http://www.gpo.gov/fdsys/pkg/BILLS-106hr2397ih/pdf/BILLS-106hr2397ih.pdf; S. 77 Paycheck Fairness Act (introduced January 22, 2001) http://thomas.loc.gov/cgi-bin/query/z?c107:S.77; S. 8 Enhancing Economic Security for America’s Working Families Act (introduced January 22, 2001) http://www.gpo.gov/fdsys/pkg/BILLS-107s8is/pdf/BILLS-107s8is.pdf; H.R. 781 Paycheck Fairness Act (in- troduced February 28, 2001) http://www.gpo.gov/fdsys/pkg/BILLS-107hr781ih/pdf/BILLS-107hr781ih.pdf; S. 76 Paycheck Fairness Act, 108th Congress (introduced January 7,2003)http://www.gpo.gov/fdsys/pkg/ BILLS-108s76is/pdf/BILLS-108s76is.pdf; 10. H.R. 1688 Paycheck Fairness Act (introduced April 9, 2003) http://www.gpo.gov/fdsys/pkg/BILLS-108hr1688ih/pdf/BILLS-108hr1688ih.pdf; H.R. 1687 Paycheck Fair- ness Act (introduced April 19, 2005) http://www.gpo.gov/fdsys/pkg/BILLS-109hr1687ih/pdf/BILLS- 109hr1687ih.pdf; S. 841 Paycheck Fairness Act (introduced 4/19/2005) http://www.gpo.gov/fdsys/pkg/ BILLS-109s841is/pdf/BILLS-109s841is.pdf; H.R. 1338 Paycheck Fairness Act (introduced March 6, 2007) http://www.gpo.gov/fdsys/pkg/BILLS-110hr1338ih/pdf/BILLS-110hr1338ih.pdf; H.R. 12 Paycheck Fairness Act(introduced January 6, 2009) http://www.gpo.gov/fdsys/pkg/BILLS-111hr12ih/pdf/BILLS-111hr12ih.pdf; 15. S. 182 Paycheck Fairness Act (introduced April 19, 2005) http://www.gpo.gov/fdsys/pkg/BILLS- 109s841is/pdf/BILLS-109s841is.pdf and http://www.gpo.gov/fdsys/pkg/BILLS-111s182pcs/pdf/BILLS- 111s182pcs.pdf; S. 3772 Paycheck Fairness Act (introduced September 13, 2010) http://www.gpo.gov/fdsys/ pkg/BILLS-111s3772pcs/pdf/BILLS-111s3772pcs.pdf; S. 2966 Wage Awareness Protection Act (introduced July 27, 2000) http://www.gpo.gov/fdsys/pkg/BILLS-106s2966is/pdf/BILLS-106s2966is.pdf; Fair Pay Act of 2011. Introduced as HR 1493 in the House of Representatives on April 12, 2011 http://www.gpo.gov/fdsys/ pkg/BILLS-112hr1493ih/pdf/BILLS-112hr1493ih.pdf and as S 788 on April 12, 2011 in the Senate:http:// www.gpo.gov/fdsys/pkg/BILLS-112s788is/pdf/BILLS-112s788is.pdf; Fair Pay Act of 2001. Introduced as HR 1362 on April 3, 2001 http://www.gpo.gov/fdsys/pkg/BILLS-107hr1362ih/pdf/BILLS-107hr1362ih.pdf and S684, on April 3, 2001. http://www.gpo.gov/fdsys/pkg/BILLS-107s684is/pdf/BILLS-107s684is.pdf; Fair- ness and Individual Rights Necessary to Ensure a Stronger Society: Civil Rights Act of 2004 (introduced on February 11, 2004 as HR 3809: http://www.gpo.gov/fdsys/pkg/BILLS-108hr3809ih/html/BILLS- 108hr3809ih.htm and on February 12, 2004 as S 2088 in the Senate: http://www.gpo.gov/fdsys/pkg/BILLS- 108s2088is/html/BILLS-108s2088is.htm); S.862 114th Congress Paycheck Fairness Act (introduced in the House and the Senate March 25, 2015) https://www.congress.gov/bill/114th-congress/senate-bill/862?q=%7B %22search%22%3A%5B%22paycheck+fairness+act%22%5D%7D.

652 / MARLENE KIM

bills would have amended Section 15 of the FLSA—the Prohibited Acts—to make unlawful any policies or actions against employees who share informa- tion about their earnings. For example, the Wage Awareness Protection Act would have prevented

employers from taking any adverse employment action against employees who inquire about or discuss wages, or to “make or enforce a written or oral confi- dentiality policy that prohibits an employee from inquiring about, discussing, or otherwise disclosing the wages of the employee or another employee” (Bierman and Gely 2004: 186). Various Paycheck Fairness Acts introduced in Congress, most recently in March 2015, would amend Section 15 so that it would be unlawful for any person:

to discharge or in any other manner discriminate against, coerce, intimidate, threaten, or interfere with any employee or any other person because the employee inquired about, discussed, or otherwise disclosed the wages of the employee or another employee. (See S. 3772, 9/13/2000; S. 182 4/19/2005; HR 12 1/6/2009)

The motivation for federal legislation was to reduce the gender wage gap (see Mikulski 2015; Gely and Bierman 2003). For example, Congresswoman Eleanor Holmes Norton states that it is important

[t]o keep employers from gagging employees by threatening them with sanctions for freely discussing and learning the wages of their cowork- ers, enabling women to engage in self-help to demand wage increases where appropriate. . .. (cited in Gely and Bierman 2003: 132)

Because repeated attempts to outlaw pay secrecy at the federal level failed, on April 8, 2014, President Obama issued an Executive Order banning pay secrecy for federal contractors. Only one out of five workers are covered by this provision, however,6 but managers and supervisors are included.

State Laws on Pay Secrecy

Eleven states have passed their own laws banning pay secrecy: Michigan (1982), California (1984), Colorado (2008), Illinois (2004), Maine (2009), Ver- mont (2005), New Jersey (2013), New York (2015), Minnesota (2014), New

6 See https://www.hrc.org/resources/entry/an-important-step-toward-workplace-equality-an-executive-order- on-federal-c; http://thinkprogress.org/economy/2014/04/06/3423399/obama-secrecy-salary/.

Pay Secrecy and the Gender Wage Gap / 653

Hampshire (2014), and Connecticut (2015). These laws vary in terms of which employees are covered and under which circumstances (see Table 1).7 In gen- eral, states that prohibit pay secrecy through their labor laws (such as Califor- nia, Michigan, and Colorado) commonly include only private-sector workers, with Colorado further limiting employees to those covered by the NLRA (so that supervisors and managers are excluded from these protections). In con- trast, states that prohibit pay secrecy in their Equal Employment laws, such as Illinois, New Jersey, and Maine, do so for both public- and private-sector workers but only when employees are investigating unequal pay claims. Ver- mont is the only exception to this pattern, covering both private- and public- sector workers without limiting them to investigating unequal pay claims. Ver- mont also allows workers to file claims anonymously, by sending in employee manuals that reveal an employer’s pay-secrecy policy. (See Table 1 for a sum- mary of state pay-secrecy laws.) In states without laws prohibiting pay secrecy, little is known about the

extent to which workers face reprisals for sharing information about their

TABLE 1

STATE LEGISLATION ON PAY SECRECY

State Date Passed Date Effective Scope of Law Workers Covered

California 1984 January 1985

Policies and retaliation; wage disclosure only

Private sector

Colorado April 2008 August 2008

Policies and retaliation; wage inquiry, disclosure, comparisons, discussions

Private sector workers covered by the NLRA

Illinois May 2003 January 2004

Cannot prohibit wage disclosure or inquiry but only when exercising Equal Pay laws

Private and public sector

Maine June 2009 September 2009

Cannot prohibit wage disclosure or inquiry but only when exercising Equal Pay laws

Private and public sector

Michigan 1982 March 1983

Policies and retaliation; wage disclosure only

Private sector

Vermont 2005 July 2005 Policies and retaliation; wage disclosure only

Private and public sector

New Jersey 2013 August 28, 2013

Retaliation for wage sharing and inquiries

Investigation of discrimination only

7 Laws passed in 2014 and 2015 when this article was under review are not listed in Table 1 or included in this analysis. Louisiana also recently outlawed pay secrecy but only for State government workers.

654 / MARLENE KIM

earnings. However, research suggests that workers don’t seem to share infor- mation about their wages out of fear of such punishment (Bierman and Gely 2004; Gely and Bierman 2003). In the states that have outlawed pay secrecy, not much is known about the extent to which workers avail themselves of these laws or know about them. Interviews with state experts on pay secrecy indicate that few charges are filed regarding pay secrecy violations (P. Bass, personal communication, November 2011; G. Harris, personal communica- tion, November 2011; N. Hernandez, personal communication, 2011; D. Moy, personal communication, November 2011), and that few workers may know about these laws (P. Bass, personal communication, November 2011; D. Moy, personal communication, November 2011), so that more needs to be done about publicizing them (P. Bass, personal communication, November 2011). However, other state experts on pay secrecy discuss workers who, knowing

that they were protected by these laws, inquired about their colleagues’ pay, and when they discovered they were underpaid because of their gender, they complained to their human resource departments and demanded higher pay (S. Everett, personal communication, 2011; L. Meric, personal communication, November 2011). Of course, it could be that in some states the laws are not publicized adequately and that many workers do not know about them, but that those who are informed use the laws to investigate their pay and remedy any discrepancies, if warranted. The motivation for passing these state laws was the same as with the federal

proposals—to close the wage gap. Legislators and activists claim that for women’s earnings to increase, women need to be able to discover if they are underpaid (P. Bass, personal communication, November 2011; J. L. Donovan, personal communication, November 2011; G. Harris, personal communication, November 2011; T. Hayden, personal communication, November 30 and December 1, 2011; L. Meric, personal communication, November 2011). As one advocate elaborates:

As part of pay equity, workers need to know how their pay compares to other workers in order to understand if they are paid fairly, and if not, take action. Few workers know that sharing their wages is a “con- certed activity” protected by the NLRA. We wanted a clear statement in Colorado law so that employees knew they could share wage infor- mation without reprisals. (L. Meric, personal communication, Novem- ber 2011)

For Maine, the Lilly Ledbetter case was also a motivating factor in passing legislation (P. Bass, personal communication, November 2011). But was Lilly Ledbetter the exception, or are women indeed underpaid as these advocates

Pay Secrecy and the Gender Wage Gap / 655

and legislators believe—because women simply don’t know that they are paid less than men?

Pay Secrecy and Wages: An Overview and the Data

If advocates and legislators are correct that women feel empowered to inquire and remedy any gender differences in pay, state laws outlawing pay secrecy would increase wages for women and reduce the gender wage gap. I use the March Supplement of the Current Population Survey (CPS), also known as the Annual Social and Economic Supplement, from the Integrated Public Use Microdata Series (IPUMS), to investigate this. Data included 1977 to 2012 because prior to 1977, not all the variables were available. Civilian workers between 25 and 64 years of age, who were wage and salary workers, and who had positive earnings (and worked positive numbers of weeks) were included in the sample. With these restrictions, the sample contained more than 2.1 million observations; approximately 1.1 million men and 1 million women. A cross-sectional examination of the data in 2011 and 2012, when pay

secrecy had been outlawed in the six states included in my analysis,8 indicates that women’s wages are higher in states that have outlawed pay secrecy—but men’s wages are higher, too. As Table 2 shows, men earn an average of $30.15/hour in states that ban pay secrecy, compared to $28.32 in states with- out these bans. Similarly, women earn $24.20 in states that ban pay secrecy, and a lower amount, $21.97/hour, in states that do not have such bans (these differences are statistically significant). Of course, there are two possible rea- sons for these outcomes. First, employers’ worst fears may be realized when pay secrecy is illegal: wages may creep higher once employees know what others earn, because employers must increase wages that are too low. Second, workers in these states may be quite different (e.g., more highly educated), leading to higher wages. Notice that the gender wage gaps persist whether one is in a state with or

without laws banning pay secrecy. These gender wage differences are all sta- tistically significant at the 1-percent level. Most likely, these persistent gender gaps result from many factors (more on this later). The gender wage gap is smaller in states that have banned pay secrecy, although these differences are not statistically significant. The statistical insignificance may be due to the low

8 Since the data end in 2012, I could not include New Jersey or the other states that passed legislation after this date.

656 / MARLENE KIM

sample size (six each year) or to other factors that are influencing wage rates (more on this below). Obviously, however, many factors can explain these findings. Workers liv-

ing in states that have banned pay secrecy may be different than workers in other states, for example, with higher education or work experience levels. States that have banned pay secrecy may also have passed other laws that can increase wages, such as higher minimum wages or laws that allow advocating for labor unions. As Table 3 shows, those living in states that have outlawed pay secrecy are more likely to live in metropolitan areas and in central cities. These states also have fewer African American workers, more Hispanic and Asian American workers, and fewer workers with less than high school degrees. These differences are likely to explain the higher wage rates of both men and women in these states. Similarly, the smaller (albeit insignificant) wage gaps in states with laws that

ban pay secrecy may be due to two entirely different factors. First, it may be true that states with pay-secrecy laws increase the pay for women and thus lower the gender pay gap because women are informed if they are underpaid compared to men and remedy this problem. In this scenario, the law is work- ing as proponents claim. A second explanation, however, may be self-selection. States that pass pay-

secrecy laws may care more about nondiscrimination and thus have greater enforcement of workplace discrimination. They may also be more aggressive in other areas that affect the gender wage gap, such as in affirmative action

TABLE 2

HOURLY EARNINGS FOR FULL-TIME, YEAR-ROUND WORKERS, 2011–2012

In the States

Banning Pay Secrecy No Ban in States

A. Hourly Earnings Men 30.15***

(.2906) 28.32 (.1291)

N 11,388 46,396 Women 24.20***

(.2355) 21.97 (.0962)

N 8892 39,053 B. Women’s Earnings Compared to Men’s Earnings

Ratio of Women’s/Men’s Earnings 0.7985 (.0111)

0.7783 (.00544)

N 12 90

NOTES: Data are from the Current Population Survey, March Supplement 2011–2012 data. Wages are in 2012 dollars. Sta- tistically different between states with and without laws banning pay secrecy at the: 10-percent level; **5-percent level; ***1-percent level. Standard errors are in parentheses.

Pay Secrecy and the Gender Wage Gap / 657

and gender-neutral education. In this scenario, the gender pay gap may be lower compared to states without such laws, but this may be due to factors other than the pay-secrecy law, including a culture more supportive of working women and ending employment discrimination, a legislature more supportive of women’s rights, stronger laws on gender pay discrimination, or stronger enforcement of such laws. Thus, banning pay secrecy may not be what increases pay for women, but rather a larger culture that supports women. In this scenario, passing pay-secrecy laws in other states would not lower the gender pay gap; instead, changing the social climate around women’s pay may be effective.

Research Methods

In order to examine these competing explanations, control for factors that may explain these patterns, and increase the sample size, I ran a difference-in- difference-in-difference (DDD) human capital wage regression. Human capital regressions are commonly run to examine gender wage differentials, since this

TABLE 3

MEANS BY STATES WITH AND WITHOUT PAY SECRECY LAWS

No Pay Secrecy Laws Pay Secrecy Outlawed

Mean s.d. Mean s.d.

Ln real wage 2.8997632 0.6747491 2.9789798 0.6876831 Number of children 1.0198004 1.1864069 1.037327 1.2225702 Metroarea 0.7686778 0.4216779 0.8935097 0.3084642 Central city 0.2391953 0.4265922 0.3104914 0.4626948 Hispanic 0.0788398 0.2694886 0.1699448 0.3755843 Female 0.4718978 0.4992096 0.4566614 0.4981182 Black 0.1242244 0.3298374 0.0823992 0.2749719 Other race 0.038352 0.1920446 0.0814843 0.2735774 High school 0.220942 0.4148815 0.1851082 0.3883853 Some college 0.2404129 0.4273342 0.2690724 0.4434777 College 0.2133254 0.4096556 0.2327202 0.4225654 Advanced degree 0.0690064 0.2534649 0.0704794 0.2559533 Never married 0.1726349 0.3779313 0.1989451 0.3992067 Married 0.6610127 0.4733655 0.6384964 0.480436 Sep/div/wid 0.1663524 0.3723966 0.1625585 0.3689624 Pay secrecy 0 0 0.6238283 0.4844239 Female*pay secrecy 0 0 0.2862372 0.4520016 Work experience 23.6157704 10.7683 23.2651797 10.6324109 Work experience2 673.660898 556.689613 654.316749 545.627074 Child under 5 0.1517406 0.3587692 0.1589455 0.3656252 Number of observations 1,685,259 419,730

658 / MARLENE KIM

allows for key variables to be measured separately from the effect of other characteristics that affect wage rates, such as higher education levels, work experience, race, and ethnicity (see Kim 2013). DDD analysis is used to exam- ine the effect of policy changes on specific groups of people—in this paper, women (see Pischke 2005; Imbens and Wooldridge 2007). Following Pischke (2005), the DDD specification I ran was:

Lnðwi;s;tÞ ¼ a1Xi;s;t þ a2fi;s;t þ a3It þ a4Is þ a5ðpss;t � fi;s;tÞ þa6ðfi;s;t � IsÞ þ a7ðfi;s;t � ItÞ þ a8ðIs � ItÞ þ ei;s;t

ð1Þ

For each individual i in state s and time t, the dependent variable is the natural log of the real (in 2012 dollars) hourly wage, wi,s,t.

9 Independent variables (Xi,s,t) were included to account for factors other than gender that may affect wages. These include educational attainment, race, potential work experience, its square, living in a metropolitan area or central city, and marital status, all which are typical controls in wage regressions. The number of children, the presence of children under age 5, and broad industry and occupational dichotomous vari- ables were also included, because these can explain differences in wages by gen- der.10 A female dummy variable, fi,s,t, is also included to capture the effect of underpaying female workers, controlling for all these other factors. A matrix of fixed effects by year (It) controls for economic conditions that

vary over time, such as the business cycle. A matrix of state effects (Is) con- trols for variations in the cost of living, economic conditions, business cli- mates, and state laws. This matrix interacted with the female dummy variable (fi,s,t*Is) captures political climates toward women or other state laws that may affect the pay for women and the gender pay gap across states. A dichotomous variable, pss,t, indicates those living in a state s in which pay secrecy laws were in effect in year t. This pay-secrecy variable is interacted with the female dummy variable (pss,t*fi,s.t) to measure the effect of outlawing pay secrecy on women’s wages in particular. The coefficient of interest is a5.

9 This was calculated as the annual wage or salary earnings in the previous year divided by the product of the usual hours worked per week and the number of weeks worked in the previous year.

10 Men usually have positive coefficients on children variables, while women usually have negative coefficients. The explanation for this difference varies from women not working as hard as men (O’Neill 2004) to discrimination against mothers (Budig and England 2001; Correll, Benard, and Paik 2007). In addi- tion, some explain the occupational and industry variables as accounting for job preferences or unmeasured worker or job characteristics (O’Neill 2004); while others explain these variables as capturing job segrega- tion by gender (Hegewisch and Liepmann 2013). Thus, including these variables may underestimate wage penalties for women if job segregation by gender and the coefficients on the children variables are in part determined by employer discrimination rather than the qualifications, productivity, and preferences of work- ers.

Pay Secrecy and the Gender Wage Gap / 659

Interacting the female dummy variable with year effects (fi,s,t*It) controls for trends particular to women, such as the general tendency for the wage gap to decline over time. Interacting state and year effects (Is*It) controls for unob- servable factors by state-year cells. The last argument in equation 1 is the indi- vidual specific error term. All reported standard errors are clustered at the state level. Pay secrecy probably affects some workers more than others. Those with

college educations (or higher) are more likely to work in professional jobs in which negotiating pay is more common. These jobs are more likely to have wider variation in skill levels and also allow for more discretion in pay-setting by employers. Research confirms that college-educated workers have higher variation in their pay (Chay and Lee 2000; Lemieux 2006), and that this varia- tion has increased over the time period examined in this study (Lemieux 2006). Thus, separate regressions are also run on those with college degrees and those without, to see the effects of outlawing pay secrecy by educational differences. The expectation is that laws outlawing pay secrecy will have a greater effect on college-educated workers.11

Definitions of the variables and their means and standard deviations are included in the Appendix.

Research Results

To assess the effects of outlawing pay secrecy, Table 4 first shows the regression results using a simpler difference-in-difference specification. Here, only human-capital variables (including gender) and state and year fixed effects are included, omitting all interaction variables.12 The policy variable is shown without any interactions with gender in order to see if the policy had any effect at all. As specification 1 in this table indicates, the policy does not seem to affect wages, as the results are statistically insignificant. Separate regressions by gender also yield insignificant results.

11 Those covered by unions are also more likely to have standardized pay and less room to negotiate pay than those without unions, consequently having lower variance in earnings (Freeman and Medoff 1984). However, data on union coverage begins only in 1990, which would omit the effect of two states that imple- mented their laws before this time period. In addition, not all surveyed respondents were asked about being covered by a union, leaving much of the sample in the “missing variable” category even after 1990.

12 The regression results for the control variables (not shown) are consistent with previous findings in the research literature. Those living in metropolitan areas, who are not racial or ethnic minorities, and who had more potential work experience earn more. People with higher education levels and who are married (as opposed to never married) also earn more, and women earn less.

660 / MARLENE KIM

This confirms that the higher average wages in the states that outlawed pay secrecy in Table 2 results from the different characteristics of workers in these states: their higher level of education, higher likelihood of living in metropoli- tan areas, and lower numbers of African Americans. Enacting pay-secrecy laws does not appear to increase wages for all employees, as employers feared. Model 2 in Table 4 keeps the previous specification, but adds the policy

interacted with being female. Here, the policy by itself is once again insignifi- cantly different from zero. The coefficient on the interaction term of this policy and being female is positive and statistically significant, however. On average, women’s wages increased 4 to 5 percent after pay-secrecy laws were passed, leading to an increase in the gender wage ratio (the ratio of women’s to men’s pay) of 3 to 3.5 percentage points.13

TABLE 4

REGRESSION RESULTS: DIFFERENCE-IN-DIFFERENCE AND DIFFERENCE-IN-DIFFERENCE-IN-DIFFERENCE

RESULTS

Difference-in-Difference Model: State and Year Fixed Effects

All workers Full-time, year-round workers

1. Policy in effect (PS) –0.00502 (.02338)

–.00710 (.02327)

2. Policy in effect (PS) –0.02735 (.02090)

–0.02408 (.01779)

Policy in effect*female 0.04810**

(.02090) 0.04045**

(.0186)

Difference-in-Difference-in-Difference Model: (see equation [1] in text)

Coefficient on pay secrecy policy in effect (PS) * female for

3. All education levels 0.01267**

(.00578) 0.0102 (.008)

4. No college degree 0.00958 (.0235)

0.0116 (.02195)

5. College degree 0.02665* (.01318)

0.02469***

(.0095)

NOTES: Regressions are on log of real (2012) hourly earnings, 1977–2012. Data are from March CPS, Annual Social and Economic Supplement, 1977–2012 from IPUMS. Sample includes wage and salary earners with positive earnings and weeks worked, between 25–64 years of age. See text for control variables used. Models 3–6 include all interaction terms from equation (1) in text. N=1,540,179 for the full sample (1,094,816 men and 99,605 women) and 2,083,421 (892,921 men and 647,258 women) for full-time, year-round earners. Clustered standard errors are in parentheses. *Significant at 10-percent level; **significant at 5-percent level; *** significant at 1-percent level.

13 With average wages of $19.41 for women and $27.33 for men in the data, the wage ratio is 71.02 percent. For full-time year-round (FTYR) workers, average wages were $19.74 for women and $27.17 for men, for a wage ratio of 72.65 percent. If wages increased 4 percent for FTYR female workers and 5 per- cent for all workers, the wage ratios would be 74.57 percent for all and 75.56 percent for FTYR workers, for an increase of 3.55 and 3 percentage points, respectively. Wages were calculated as real wages, 2012=100.

Pay Secrecy and the Gender Wage Gap / 661

Model 3 uses the DDD specification in equation 1 for all education levels. Notice that in this specification, the effect on outlawing pay secrecy for women is much smaller. Women’s wages increase by only 1 percent, and this is not even statistically significant for full-time, year-round workers. When examined by education level, the results are quite different: Wage increases for women with low education levels, although positive, are statistically insignifi- cant (see model 4). In contrast, women with college degrees increased their pay 3 percent (see model 5). Thus, these results show some support for the claims of advocates that laws outlawing pay secrecy increase pay for women, especially those with college degrees. As a check on these results, I use a method to measure race or gender dis-

crimination (see Verdugo 1992; for examples, see McGuire and Reskin 1993; Green and Ferber 2005; Yamane 2002; Mar 2000; Kim 2009) and calculate the gender wage gap in states with and without pay-secrecy laws in effect. I then use a difference-in-difference (DD) model to see if this gender wage gap is lower in states in which pay secrecy is outlawed. To do this, first, a wage regression was run only for men from 1977 to 2012,

using the same human capital controls, X, in equation (1) and state fixed effects:

Lnðwi;s;tÞ ¼ b1Xi;s;t þ b2Is þ ei;t ð2Þ This regression provides estimates of the coefficients for men, b1

m and b2 m,

after controlling for the same education attainment, geography variables, and demographic variables as in equation (1). Using these estimated coefficients, a predicted wage, p, was estimated for every woman by using the characteristics of each of the women (the X’s) but the estimated coefficients (b1

m and b2 m)

from the regression of men:

pi;s;t ¼ bm1 Xi;s;t þ bm2 Is ð3Þ This predicted wage is the wage that women would earn if they had the

same returns to their characteristics as do men. It is often a measure of a nondiscriminatory wage—i.e., what women would earn in the absence of dis- crimination, because it measures their earnings as if they were treated as men (Verdugo 1992). The actual real log wage of each woman was then subtracted from this predicted wage to estimate the amount of discrimination in wages each woman faced—i.e., what they would have earned as a man minus what they actually did earn:

Di;s;t ¼ pi;s;t � lnðwi;s;tÞ ð4Þ where i is the ith woman in year t and state s. This discrimination variable, D, measures the adjusted gender wage gap (adjusted for human capital and other

662 / MARLENE KIM

characteristics). I used a difference-in-difference specification to examine if laws outlawing pay-secrecy policies reduced this wage gap:

Di;s;t ¼ c1pses þ c2It þ c3pss;t þ ei;s;t ð5Þ Whether or not a state ever had prohibited pay secrecy at any time is now a

control (pse), as these states may be different from those that never passed such laws. Year effects are included in this specification. The coefficient of interest is c3, which measures the effect of the policy on lowering the pay gap for women. The first three rows in Table 5 show the results from equation (5). Notice that

although the adjusted wage gap, or D, is indeed lower for women in states in which pay-secrecy laws are in effect, these are not statistically significant. But these effects are once again very different by education level. Those with college degrees receive 5 to 6 percent reductions in the gender wage gap. In contrast, those without college degrees have statistically insignificant changes in the wage gap. I ran a variation of this specification with state fixed effects instead of the

pse variable in equation (5). The measure of discrimination, D, was con- structed similarly in equations (2) and (3) but using year instead of state fixed effects.14 Rows 4–6 in Table 5 indicate that in this specification, the effects

TABLE 5

COEFFICIENT ESTIMATES FOR PAY SECRECY LAWS IN EFFECT (PS)

Model with pse variable in equation 5 (ever had pay secrecy outlawed in state) All Workers Full-Time, Year-Round

1. All education levels –.0232 (.0232)

–.0217 (.0212)

2. No college degree –.0086 (.0209)

–.0089 (.0203)

3. College degree –.0578**

(.0273) –.0485***

(.0209)

Model with state fixed effects instead of pse variable

4. All education levels –.1068***

(.0304) –.0830**

(.0312) 5. No college degree –.0785***

(.0224) –.0561**

(.0282) 6. College degree –.1468***

(.0550) –.1198***

(.0418)

NOTE See text for an explanation of variables used. Dependent variable is the adjusted gender wage gap; clustered standard errors are in parentheses. *Significant at 10-percent level; **significant at 5-percent level; ***significant at 1-percent level.

14 Thus for this specification Di,s,t = φ1Is + φ2It + φ3pss,t + ei,s,t (similar to equation 5), equation (2) is now: ln(wi,s,t) = w1Xi,s,t + w2 It + ei,t, and equation (3) is now pi,s,t = w1

mXi,s,t +w2 m It. D is still calculated

as Di,s,t = pi,s,t - ln(wi,s,t) in equation (4).

Pay Secrecy and the Gender Wage Gap / 663

are even greater, with reductions of the adjusted wage gap of 8–11 percent. The results by education level are also consistent with previous findings: Women with college degrees face much higher declines in the gender wage gap—12–15 percent. In comparison, women without such degrees experienced 6–8 percent decreases in the wage gap, and these are statistically significant. Taken together, these results show support that state laws that outlaw pay

secrecy increase earnings for women relative to men, especially among col- lege-educated women, and that the gender wage gap falls among this popula- tion as well. Thus pay-secrecy laws appear to help women determine if they are underpaid compared to men and may be useful to reduce the gender wage gap, especially among those with a college (or higher) education.

Conclusion

When the FLSA was first passed, gender differences in wages were accepted and legal (Kim 1999). However, to be relevant, the FLSA may need to be amended from time to time in order to meet the needs of the current work- force. Today, the labor force is vastly different than in 1938, in that it is com- prised of many more women. Currently, women earn 82 percent of what men earn,15 and although the gender wage gap has narrowed over several decades, advocates feel that with women now attending college at higher rates than men, the gender wage gap should be much smaller. Although much research has been conducted on various policies to see how

pay can increase for women, no one has examined the effect of pay secrecy on women’s earnings. Using a natural experiment of states that have outlawed employment practices that prevent workers from discussing pay, I find that wages are higher for women in states that have outlawed pay secrecy, espe- cially among those with college degrees. These women increase their earnings 3 percent in states that have outlawed pay secrecy. Additional analyses find that those with college degrees reduce the gender wage gap by 5–6 percent, or by 12–15 percent, depending on the specification and population of workers examined. These results provide support for state laws outlawing pay secrecy. Thus,

prohibiting pay secrecy in other states is likely to benefit college-educated women, increasing their pay and lowering the gender wage gap. National leg- islation has been introduced to amend the FLSA, including the Paycheck Fair- ness Act, to outlaw pay secrecy on a national level, but Congress has not

15 Author’s calculations from U.S. Bureau of Labor Statistics (2013a). The third quarter in 2013 reports weekly earnings of $860 for men and $706 for women for full-time workers.

664 / MARLENE KIM

passed any of this legislation. The results in this paper indicate that such legis- lation is likely to improve women’s pay. Thus, outlawing pay secrecy in the Fair Labor Standards Act and state legis-

lation should be considered another tool to lower the gender wage gap among college-educated women. In this way, the FLSA can be amended to address a problem that was not seen as critical when it was first passed: the underpay- ment of female workers.

REFERENCES

Bamberger, Peter, and Elena Belogolovsky. 2010. “The Impact of Pay Secrecy on Individual Task Perfor- mance.” Personnel Psychology 63(Winter): 965–96.

Bierman, Leonard, and Raphael Gely. 2004. “‘Love, Sex and Politics? Sure. Salary? No Way’: Workplace Social Norms and the Law.” Berkeley Journal of Employment and Labor Law 25(1): 167–91.

Bronfenbrenner, Kate. 1994. “Employer Behavior in Certification Elections and First-Contract Campaigns: Implications for Labor Law Reform.” In Restoring the Promise of American Labor Law, edited by Sheldon Friedman, Richard W. Hurd, Rudolph A. Oswald, and Ronald L. Seeber, pp. 75–89. Ithaca, NY: Cornell University (ILR) Press.

Budig, Michelle J., and Paula England. 2001. “The Wage Penalty for Motherhood.” American Sociological Review 66(4): 204–55.

Chay, Kenneth Y., and David S. Lee. 2000. “Changes in Relative Wages in the 1980s: Returns to Observed and Unobserved Skills and Black-White Wage Differentials.” Journal of Econometrics 99(11): 1–38.

Colella, Adrienne, Ramona L. Paetzold, Ashghar Zardkoohi, and Michael J. Wesson. 2007. “Exposing Pay Secrecy.” Academy of Management Review 32(1): 55–71.

Correll, Shelly J., Stephen Benard, and In Paik. 2007. “Getting a Job: Is There a Motherhood Penalty?” American Journal of Sociology 112(3): 1297–338.

Danziger, Leif, and Eliakim Katz. 1997. “Wage Secrecy as a Social Convention.” Economic Inquiry 35(1): 59–69.

Edwards, Matthew A.. 2005. “The Law and Social Norms of Pay Secrecy.” Berkeley Journal of Employment and Labor Law 26(1): 41–63.

Eisenberg, Deborah Thomson. n.d. “Money, Sex and Sunshine: A Market-Based Approach to Pay Discrimi- nation.” Mimeo. University of Maryland School of Law.

Federal Reserve Board of St. Louis. 2013. “Labor Force Participation Rate–Women.” http://research.stlouis- fed.org/fred2/series/LNS11300002 (accessed September 17, 2013).

Freeman, Richard, and J. Medoff. 1984. What Do Unions Do?. New York: Basic Books. Gely, Raphael, and Leonard Bierman. 2003. “Pay Secrecy/Confidentiality Rules and the National Labor

Relations Act.” University of Pennsylvania Journal of Employment and Labor Law 6(1): 121–56. Green, Carole A., and Marianne A. Ferber. 2005. “Do Detailed Work Histories Help to Explain Gender and

Race/Ethnic Wage Differentials?” Review of Social Economy 77(3): 55–85. Greenhouse, Linda. 2007. “Justices’ Ruling Limits Lawsuits on Pay Disparity,” New York Times, May 30. Hegewisch, Ariane, and Hanna Liepmann. 2013. “Occupational Segregation and the Gender Wage Gap in

the U.S.” In Handbook on Research on Gender and Economic Life, edited by Deborah M. Figart, and Tonia L. Warnecke, pp. 200–17. Northampton, MA: Edward Elgar.

Imbens, Guido M., and Jeffrey M. Wooldridge. 2007. “What’s New in Econometrics? Lecture 10: Differ- ence-in-Differences Estimation.” http://www.nber.org/WNE/lect_10_diffindiffs.pdf (accessed June 18, 2015).

Institute for Women’s Policy Research. 2010. Pay Secrecy and Paycheck Fairness. Washington, DC: Author.

Kim, Marlene. 1999. “Inertia and Discrimination in the California State Civil Service.” Industrial Relations 38(1): 46–68.

Pay Secrecy and the Gender Wage Gap / 665

———. 2009. “Race and Gender Differences in the Earnings of Black Workers.” Industrial Relations 48(7): 466–88.

———. 2013. “Race and Ethnicity in the Workplace.” In Handbook on Research on Gender and Economic Life, edited by Deborah Figart and Tonia Warneke, pp. 218–35. Northampton, MA: Edward Elgar.

Lemieux, Thomas. 2006. “Increasing Residual Wage Inequality: Composition Effects, Noisy Data, or Rising Demand for Skills?” American Economic Review 96(6): 461–98.

Mar, Don. 2000. “Four Decades of Asian American Women’s Earnings: Japanese, Chinese and Filipino American Women’s Earnings 1960–1990.” Contemporary Economic Policy 18(4): 228–37.

McGuire, Gail M., and Barbara F. Reskin. 1993. “Authority Hierarchies at Work: The Impacts of Race and Sex.” Gender and Society 7(12): 487–506.

Mikulski, Barbara. 2015. “Mikulski, DeLauro Reintroduce Paycheck Fairness Act to End Wage Discrimina- tion.” http://www.mikulski.senate.gov/newsroom/press-releases/mikulski-delauro-reintroduce-paycheck- fairness-act-to-end-wage-discrimination (accessed June 24, 2015).

National Women’s Law Center. 2013. “Lilly Ledbetter Fair Pay Act.” http://nwlc.org/resource/lilly-ledbetter- fair-pay-act-0 (accessed June 10, 2015).

O’Neill, June. 2004. “The Gender Gap in Wages, Circa 2000.” American Economic Review 93(5): 309–14. Pische, Jorn-Steffen. 2005. “Empirical Methods in Applied Economics.” Lecture Notes. London School of

Economics. http://econ.lse.ac.uk/staff/spischke/ec524/evaluation3.pdf (accessed June 10, 2015). Smith, Ralph E.. 1979. “The Movement of Women into the Labor Force.” In The Subtle Revolution: Women

at Work, edited by Ralph E. Smith, pp. 1–35. Washington DC: The Urban Institute. U.S. Bureau of Labor Statistics. 2013a. “Economic News Release. Table 1. Median usual weekly earnings

of full-time wage and salary workers by sex, quarterly averages, seasonally adjusted.” http://www.bls.- gov/news.release/wkyeng.t01.htm (accessed December 16, 2013).

U.S. Bureau of Labor Statistics. 2013b. “Table 2, Employment Status of the Civilian Noninstitutional Popu- lation 16 years and over by Sex, 1972 to Date.” http://www.bls.gov/cps/cpsaat01.pdf (accessed September 17, 2013).

Verdugo, Richard R.. 1992. “On the Cost of Being a Minority Worker, 1972–1987.” Social Science Quar- terly 75(9): 663–73.

Yamane, Linus. 2002. “Native-Born Filipina/o Americans and Labor Market Discrimination.” Feminist Eco- nomics 8(7): 125–44.

APPENDIX

TABLE A1

VARIABLE DEFINITIONS AND MEANS

All Workers Full-Time, Year-Round

Men Women Men Women

Variable Definition Mean (S.D.) Mean (S.D.) Mean (S.D.) Mean (S.D.)

Ln real wage

ln of (2012) real wage

3.0599 (0.67)

2.7554 (0.651)

3.1196 (0.616)

2.836 (0.574)

Number of children

Number of own children

1.0053 (1.222)

1.0445 (1.162)

1.0529 (1.223)

0.9255 (1.100)

Metro area 1 if metro area 0.7962 (0.403)

0.7961 (0.403)

0.8026 (0.398)

0.8106 (0.392)

Central city

1 if central city 0.2535 (0.435)

0.2565 (0.437)

0.2471 (0.431)

0.2677 (0.443)

666 / MARLENE KIM

TABLE A1 (cont.)

All Workers Full-Time, Year-Round

Men Women Men Women

Hispanic 1 if Hispanic 0.1104 (0.313)

0.0859 (0.28)

0.1048 (0.306)

0.0857 (0.2799)

Female 1 if female n/a .4685a

(.499) n/a 0.41711

(0.493) Black 1 if black 0.1013

(0.302) 0.1306 (0.337)

0.0955 (0.294)

0.1407 (0.348)

Other race

other race 0.0476 (0.213)

0.0482 (0.214)

0.0469 (0.211)

0.0506 (0.219)

High school

1 if high school ed

0.2125 (0.409)

0.2136 (0.41)

0.2117 (0.409)

0.2193 (0.414)

Some college

1 if some college

0.2315 (0.422)

0.264 (0.441)

0.2352 (0.424)

0.2702 (0.444)

College 1 if college degree

0.2207 (0.415)

0.2141 (0.41)

0.2362 (0.425)

0.2258 (0.418)

Advanced degree

1 if advanced degree

0.0683 (0.252)

0.0704 (0.256)

0.0733 (0.261)

0.0777 (0.268)

Never married

1 if never married

0.192 (0.394)

0.163 (0.369)

0.1704 (0.376)

0.1807 (0.385)

Married 1 if married 0.6877 (0.463)

0.6202 (0.485)

0.7171 (0.451)

0.5856 (0.493)

Sep/div/ wid

1 if separated, divorced, or widowed

0.1203 (0.325)

0.2168 (0.412)

0.1124 (0.316)

0.2337 (0.423)

Pay secrecy

1 if in state where pay secrecy is outlawed

0.1399 (0.347)

0.1345 (0.341)

0.137 (0.344)

0.1337 (0.340)

Work experience

potential work experience

23.422 (10.79)

23.671 (10.68)

23.664 (10.572)

23.8562 (10.557)

Work experience2

above squared 664.94 (558.5)

674.47 (549.6)

671.75 (547.04)

680.569 (539.73)

Child under 5

1 if presence of child under 5

0.1706 (0.376)

0.1337 (0.34)

0.1762 (0.381)

0.1074 (0.310)

aMean and standard deviation for female is computed across both gender groups.

Pay Secrecy and the Gender Wage Gap / 667

Copyright of Industrial Relations is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

that women tend to outperform men in other corporate roles—including CFO—yet are paid less. For example, among the 870-plus customers of Xact- ly, which provides sales-performance and employee-performance software, in 2016 the average female salesperson outperformed the average male by 2% in sales-quota achievement, the company says. Yet the women’s total compensation—base pay plus variable pay—registered a rather shocking 21% less than the men’s.

“We find almost routinely that women on average have higher perfor- mance ratings, but their compensation doesn’t reflect that,” says Christine Hendrickson, an employment attorney with Seyfarth Shaw.

Nor do women fare well in the promotion department. According to 2016 research by McKinsey, which surveyed 132 companies employing

The Gender Pay Gap Persists Compensation for women doesn’t reflect their performance levels or their impact on business results. By David McCann

Despite significant evidence that companies with more women leaders experience greater profitability and

stock returns, men continue to enjoy more advancement opportunities at every stage of career development. ¶ Discus- sions about that disparity are frequently framed in terms of pay inequity. Indeed, the size and causes of the compensation

››

tional median. A study of 3,000 companies across

40 countries by Credit Suisse Re- search Institute found that, from 2009 through 2014, those with an approxi- mate three-to-one male-female man- agement mix had average annualized stock returns of nearly 23%. Where the ratio was two to one, average returns increased to more than 25%. And when the numbers were balanced (although that sample size was small), returns exceeded 28%.

Much research has also shown

gap between genders is a matter of unending debate.

Most experts put little credence in the oft-cited U.S. Census Bureau sta- tistics comparing earnings by gender. The bureau’s most recent report on the topic shows women earning 83 cents on the dollar paid to men. That’s up from 79 cents a year ago, but it’s still a raw figure that doesn’t compare men and women doing similar work or take into account factors like the time women spend out of the workforce focused on raising families.

That’s not to say the pay gap is fictional. “There’s no question there’s an income disparity, and probably in no case does more than half of that [79-cents-on-the-dollar] gap go away when you control for other factors,” says Barry Gerhart, a professor at the University of Wisconsin School of Business. “The question is, what causes that? That’s harder to answer.”

Cockeyed Data A host of studies have shown a link between gender diversity and cor- porate performance. For example, a 2015 report by McKinsey, based on data from 366 companies, found that those companies in the top quartile of gender-diversity metrics were 15% more likely to have financial returns that were above their industry’s na-

24 CFO | January/February 2017 | cfo.com

HUMAN CAPITAL

0%

20

40

60

80

100%

Women

Men

% of employees, 2016

■ Men ■ Women

Leadership Inequity The proportion of women in corporate America decreases at each higher level of leadership.

Source: Women in the Workplace 2016, McKinsey & Co.

Entry level

54 46

63

37

67

33

71

29

76

24

81

19

Manager Sr. Manager/ Director

VP SVP C-Suite

4.6 million people and separately surveyed 34,000 employees, women are under-represented at every level within corporate leadership pipe- lines—and more prominently so at each succeeding, higher level. (See “Leadership Inequity.”)

For every 100 women promoted, 130 men are promoted, McKinsey notes.

As for CFO representation, women make up just 14.1% of finance chiefs globally, though they’re heavily skewed toward Asia, and in particular China, where they account for 22% of finance chiefs, according to a Septem- ber 2016 report by Credit Suisse Re- search Institute. In the United States, among the 1,000 largest companies by revenue, as of July 2016 only 12% had a female CFO, according to Korn Ferry.

The higher up in the corporate hierarchy you look, the fewer women there are. But why is that?

Subtleties Abound Hendrickson suggests that women are, on average, more reticent to apply for jobs or promotions unless they meet all of the stated requirements. Appli- cants who seek promotions when they have 70% or 80% of the qualifications needed are less likely to get them. But “if you put yourself in the hat more often, you’re more likely to be selected for a promotion,” she says.

Also, women are 20% less likely than men to say that their manager often gives them difficult feedback that helps improve their performance, according to McKinsey’s research.

“Men may be more comfortable giving feedback to men,” says Janice Ellig, an executive recruiter and past president of the Women’s Forum of New York. “Sometimes they’re afraid of legal ramifications.” For its part, McKinsey reports that managers who hesitate to give feedback are more likely to fear they will trigger “an emo- tional response” from women.

“Direct feedback is [crucial], because improved performance leads

to getting choice assignments, which impacts pay,” Ellig points out.

Indeed, fear is the root cause of not having more gender balance in leadership ranks, according to Melissa Greenwell, chief op- erating officer for The Finish Line, a specialty shoe retailer.

“Women fear taking risks and having so much responsibility that they’ll be over- extended,” Greenwell writes. “They fear not being completely prepared. They fear being wrong…. Many aren’t willing to take the next step to find out whether they’ll be successful.”

Men, meanwhile, “fear changing the rules,” according to Greenwell. “The new idea of work-life integration is perceived to be fraught with sticky policy issues and precedents that many leaders do not want to handle.”

But qualities that women innately possess are ones that companies may overlook until it’s too late, Greenwell charges: “When things go wrong, what excuses do boards … typically cite? ‘They didn’t communicate. They didn’t listen to others. They didn’t ask enough questions. They didn’t collabo- rate. They took too many risks.’ These are traits more likely to be missing if women aren’t involved.”

Meanwhile, the McKinsey research also indicates that fewer women than men feel they are able to participate meaningfully in meetings (67% vs. 74%), have recently gotten a challeng- ing assignment (62% vs. 68%), believe their contributions are appropriately valued (49% vs. 54%), and say they are turned to for input on important deci- sions (56% vs. 63%).

Further, more men lobby for a promotion or new assignment (39% to 36%), ask for an increase in compensa- tion (29% to 27%), have a substantive interaction with a senior leader at

least once a week (62% vs. 51% among senior managers, and 46% vs. 40% among middle managers), and say they believe they’ll one day be a top execu- tive (32% vs. 24%).

Pushing Accountability There are increasing ef- forts to make companies more accountable for gender-based pay inequity.

It has long been illegal to pay men and women differently for doing the same work, under two federal statutes: the Equal

Pay Act of 1963 and the Civil Rights Act of 1964. In 2009, President Obama signed the Lilly Ledbetter Fair Pay Act, restoring the protection that had been stripped away by a Supreme Court decision. This year new laws with more-specific requirements took effect in California, Maryland, Massa- chusetts, and New York.

Massachusetts, for example, made it illegal to ask a job candidate about his or her prior compensation. In Califor- nia, companies can still ask about that, but cannot use the information in set- ting compensation. “Over time that’s a significant factor in perpetuating pay inequity,” says Margaret Keane, an employment attorney with law firm DLA Piper.

Also, a series of shareholder propos- als were filed in advance of the 2016 proxy season, asking nine technology companies—Adobe, Amazon, Apple, eBay, Expedia, Facebook, Google, Intel, and Microsoft—to study their compen- sation practices and commit to closing the gender pay gap. Several of them publicly made such commitments. Amazon, Apple, and Intel found they were at near 100% pay parity.

However, Keane notes, “This was a limited group of technology com- panies. I would not say that most em- ployers are going to come out evenly the way those did.” CFO

Courtesy Melissa Greenwell 25cfo.com | January/February 2017 | CFO

Melissa Greenwell

Copyright of CFO is the property of CFO Publishing LLC and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

EEOC

Employers Must Step Up on Pay Equity, EEOC’s Burrows Says

Employers can play a proactive role in helping the Equal Employment Opportunity Commission to close a persistent gender gap in average pay for fe- male workers compared with male workers, EEOC Commissioner Charlotte Burrows (D) said July 30.

Speaking at the National Industry Liaison Group’s annual conference in New York, Burrows said the EEOC values its collaboration with employ- ers and the partnership between the federal agency and businesses that share an interest in advancing equal employment opportunity has contributed to the substantial progress made since Title VII of the 1964 Civil Rights Act was enacted and the EEOC opened its doors in 1965.

But Burrows said although women’s average wages rose from 41 cents for every dollar earned by men in 1965 to about 80 cents for every dollar earned by men in the early 1980s, progress has stalled since then. Currently, women earn only 77 cents for every dollar earned by men, and the gen- der pay gap is even more pronounced for black fe- males and Hispanic women in the workforce, she said. Black women earn only about 66 cents for ev- ery dollar earned by white male workers and His- panic women earn only 54 cents for every dollar earned by white men, Burrows said.

What employers can do. Employers can help by making clear to their managers that pay equity is corporate policy and will be enforced internally, said Burrows, who joined the EEOC last December after serving as a Justice Department official and top civil rights aide to former Sen. Edward Kennedy (D- Mass.).

Employers also can advance pay equity by con- ducting internal analyses of their compensation sys- tems and fixing any sex-based disparities that are revealed, Burrows said. Businesses also should adopt ‘‘transparent’’ pay and promotion systems in which workers aren’t penalized for disclosing or dis- cussing their pay or asking about the compensation of co-workers, she said.

Burrows lauded President Barack Obama’s Ex- ecutive Order 13,665, issued last year, that bars fed- eral contractors from enforcing pay secrecy policies and retaliating against workers who disclose or dis- cuss their pay. The Labor Department’s Office of Federal Contract Compliance Programs has issued a proposed rule to enforce Obama’s executive order, and a final rule is expected by the end of 2015.

Employer policies prohibiting pay discussions can ‘‘act as a significant obstacle’’ to ensuring equal pay for women, Burrows said. Such policies also ‘‘may make litigation more likely’’ because workers sus- pecting discriminatory pay practices must sue to get discovery of compensation information, she said.

Once finalized, the Labor Department regulations implementing EO 13,665 will reach federal contrac- tors that employ about 26 million workers, Burrows said.

If current workplace conditions continue, the gen- der pay gap wouldn’t be closed for another 50 years, Burrows said. That’s unacceptable because the shortfall in women’s pay affects families as well as the individual workers and has a ‘‘ripple effect’’ that reduces women’s Social Security and private retire- ment pay because their lifetime earnings are circum- scribed, she said.

It also hampers employers’ ability to attract and retain the best talent available, Burrows said.

‘Stalled’ progress on race. Other areas in which more progress is needed are the employment of per- sons with disabilities; eradicating continued race dis- crimination in some workplaces; securing employ- ment protections for lesbian, gay, bisexual and trans- gender individuals; and preventing retaliation against workers who lodge discrimination com- plaints, Burrows said.

In all these areas, the EEOC and employers can collaborate to prevent discrimination or retaliation before it occurs, she said.

14 (No. 9)

September 2015 COPYRIGHT � 2015 BY THE BUREAU OF NATIONAL AFFAIRS, INC. HRF ISSN 1059-6038

Regarding employment opportunities for black workers, Burrows said much has been achieved since the 1960s. But she added that the data show that after a period of ‘‘strong and remarkable prog- ress’’ for blacks in the workplace from 1965 to 1980, things have ‘‘stalled a bit.’’ The EEOC even has seen evidence of ‘‘re-segregation’’ in some workplaces by race and by sex, Burrows said.

The EEOC has made a priority of continued ra- cial and sexual integration of the workforce at all levels, Burrows said. When employers commit to adopting transparent policies for hiring and promo- tion, then progress can be made, she said. In con- trast, a ‘‘word of mouth’’ system for identifying can- didates for advancement can exclude whole groups of people, Burrows said.

Avoiding liability for discrimination isn’t the only reason employers should adopt transparent policies, Burrows said. There’s a ‘‘compelling business rea- son’’ for policies that promote a diverse workplace, as Census Bureau projections indicate that by 2043, the U.S. will be a ‘‘majority minority’’ nation, she said.

Employers that ‘‘know how to recruit, hire and retain persons of color’’ will increase their competi- tive advantage in the marketplace, Burrows said.

LGBT rights on rise. Although the recent U.S. Supreme Court decision recognizing a constitutional right to same-sex marriage doesn’t directly affect employment, Burrows said, the court articulated a ‘‘basic principle of equality’’ that will have workplace repercussions.

Most large U.S. companies, including federal con- tractors, already prohibit workplace discrimination against LGBT individuals, Burrows said. Recent surveys show about 89 percent of Fortune 500 com- panies bar discrimination based on sexual orienta- tion and about 60 percent of those large employers also prohibit bias based on gender identity, she said. Some of the strongest amicus support for LGBT rights in the Supreme Court marriage case came from business organizations, she noted.

The EEOC in a federal sector case against the Department of Transportation for the first time said alleged that bias based on sexual orientation is ‘‘nec- essarily sex discrimination’’ under Title VII. The agency in 2012 previously had decided in a federal sector case that Title VII protects transgender per- sons from sex discrimination based on failure to con- form with gender stereotypes.

The EEOC takes the same position under Title VII with respect to private employers, Burrows said.

Addressing the contractors’ representatives, Bur- rows said she hopes ‘‘we can develop best practices’’ to ensure LGBT persons are treated with respect and equality in the workplace.

Retaliation hinders enforcement. Retaliation charges under Title VII and the other laws the EEOC enforces have become the most frequently cited allegation in both private sector and federal cases, Burrows said. The EEOC recently held a pub- lic meeting on the issue, seeking input on ways to curb retaliation against employees who lodge bias complaints.

Retaliation ‘‘creates an atmosphere of intimida- tion’’ that affects the entire workplace, not just the alleged victim, Burrows said. It also impedes EEOC enforcement, as the anti-discrimination laws largely depend on individual charges to trigger agency in- vestigations, lawsuits and remedies, she said.

‘‘If people are afraid to tell us of a problem, we can’t help,’’ Burrows said. The key to curbing retali- ation appears to be ‘‘creating a culture of respect’’ in the workplace, she said.

The EEOC at age 50 looks back on enormous progress in civil rights, but ‘‘there’s no question we still have some challenges,’’ Burrows said.

The agency is ‘‘eager’’ to continue working with its ‘‘partners and colleagues’’ to increase equal op- portunity for all in the workplace, she said.

By Kevin McGowan

To contact the reporter on this story: Kevin McGowan in New York at kmcgowan@bna.com

To contact the editor responsible for this story: Susan J. McGolrick at smcgolrick@bna.com

(No. 9) 15

HRFOCUS ISSN 1059-6038 BNA September 2015

Copyright of HR Focus is the property of Bloomberg BNA and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Journal of Work and Organizational Psychology

Journal of Work and Organizational Psychology (2018) 34(1) 46-55

Cite this article as: Chernyak-Hai, L., Kim, S. K., & Tziner, A. (2018). Gender profiles of workplace individual and organizational deviance. Journal of Work and Organizational Psychology, 34, 46-55. https://doi.org/10.5093/jwop2018a6

ISSN:1576-5962/© 2018 Colegio Oficial de Psicólogos de Madrid. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Gender Profiles of Workplace Individual and Organizational Deviance Lily Chernyak-Haia, Se-Kang Kimb, and Aharon Tzinera

aNetanya Academic College, Israel; bFordham University, USA

Workplace deviance (Berry, Ones, & Sackett, 2007; Bodankin & Tziner, 2009; Cohen-Charash & Mueller, 2007; Dilchert, Ones, Davis, & Rostow, 2007; Levy & Tziner, 2011) is defined as “voluntary behavior that violates significant organizational norms and, in so doing, threatens the well-being of the organization, or its members or both” (Robinson & Bennett, 1995, p. 556). Examples include theft, sabotage, vandalism, embezzlement, withdrawal, harassment, and drug use (Bennett & Robinson, 2000; Gruys & Sackett, 2003; Robinson & Bennett, 1995; Sackett & DeVore, 2001; Spector et al., 2006). These behaviors are defined as dysfunctional because they

harm organizations in many respects, including the thwarting of goal achievement, inhibition of fellow employees, and disruption of procedures, productivity, and profitability (Aubé, Rousseau, Mama, & Morin, 2009; Dalal, 2005; Lanyon & Goodstein, 2004; Pearson, Andersson, & Porath, 2005; Robinson, 2008; Spector & Fox, 2005; Spector et al., 2006; Vardi & Weitz, 2004). For instance, in 2010 alone, a loss of 15.9 billion dollars was attributed to employee theft in the U.S.A. (Hollinger & Adams, 2010).

Past research has indicated that individual variables may account for personal differences in work deviance, such as employees’

h t t p : / / j o u r n a l s. c o p m a d r i d. o r g / j wo p

A R T I C L E I N F O

Article history: Received 3 May 2017 Accepted 11 January 2018

Keywords: Workplace interpersonal and organizational deviance Gender Profile Analysis via Multidimensional Scaling (PAMS)

A B S T R A C T

Employees’ workplace deviant behaviors have a harmful potential for organizations in many respects. Past research has indicated that individual variables may account for personal differences in work deviance. One of the prevalent findings is that men display direct aggression more frequently than women. Yet, most of the past studies have reported results providing information on the magnitude of a general behavioral tendency of each gender, leading to rough distinctions. Unlike the previous studies, we focused on examining profiles of the role of gender in interpersonal and organizational deviance, utilizing Profile Analysis via Multidimensional Scaling that allowed us to compare specific deviance behavior indicators between males and females included in the profiles. The current exploratory study reveals that gender differences in aggressive workplace behavior are not only those apparent in inter-personal relations but also when directed towards the organization. Moreover, the reported results point to specific behavioral profiles of men and women that could not be revealed using the mean difference analyses.

Perfiles de género de desviación individual y organizacional en el trabajo

R E S U M E N

Las conductas inapropiadas de los empleados en el lugar de trabajo tienen un potencial perjudicial para las organizacio- nes en muchos aspectos. Investigaciones anteriores indican que las variables individuales pueden explicar las diferencias personales en el comportamiento laboral inadecuado. Uno de los hallazgos prevalentes es que los hombres muestran agresión directa con más frecuencia que las mujeres. Sin embargo, la mayoría de los estudios presentan resultados con información sobre la magnitud de una tendencia de comportamiento general de cada género, lo que lleva a distinciones muy generales. A diferencia de los estudios anteriores, nos centramos en examinar el papel del género en la inadecuación interpersonal y organizacional utilizando el Análisis de Perfil por Escalamiento Multidimensional, que permite comparar los indicadores específicos de comportamiento inadecuado entre hombres y mujeres incluidos en los perfiles. El estudio actual revela que las diferencias de género en el comportamiento agresivo en el lugar de trabajo no sólo son evidentes en las relaciones interpersonales, sino también cuando se dirigen hacia la organización. Además, los resultados apuntan a perfiles de conducta específicos de hombres y mujeres que no aparecían en los análisis de diferencia de medias.

Palabras clave: Comportamiento interpersonal y organizacional inadecuado en el trabajo Género Perfil Análisis mediante escalamiento multidimensional

Correspondence: lilycher@netanya.ac.il (L. Chernyak-Hai).

Journal of Work and Organizational Psychology

Revista de Psicología del Trabajo y de las Organizaciones

Editor Jesús F. Salgado

Associate Editors Antonio García-Izquierdo Francisco J. Medina Silvia Moscoso Ramón Rico Carmen Tabernero

Vol. 34, No. 1, April 2018

ISSN: 1576-5962

47 L. Chernyak-Hai et al. / Journal of Work and Organizational Psychology (2018) 34(1) 46-55

personal traits and abilities (e.g., Berry et al., 2007; Dalal, 2005; Dilchert et al., 2007; Salgado, 2002; Salgado, Moscoso, & Anderson, 2013), job experiences (e.g., Hollinger & Clark, 1982; Kulas, McInnerney, DeMuth, & Jadwinski, 2007; Zhang, Mayer, & Hwang, 2017), and work stressors, including difficult work conditions and interpersonal conflicts (Bruk-Lee & Spector, 2006; Chen & Spector, 1992; Diefendorff & Mehta, 2007; Mitchell & Ambrose, 2007; Spector & Fox, 2005). Furthermore, one of the prevalent findings is that workplace deviance may be also related to gender. In general, past studies examining the mean differences in aggressiveness have shown that men display and report aggression more frequently than women (e.g., Geen 2001; Griskevicius et al., 2009; Harris, 1996; Hershcovis et al., 2007; Kogut, Langley, & O’Neal, 1992; Martinko, Douglas, & Harvey, 2006; Tavris, 1984). These differences are said to reflect a stable personality disposition, which persists throughout the lifespan and in various life areas (e.g., Walker, Richardson, & Green, 2000). Specifically, men tend more than women to display aggression towards others without provocation and to display their hostility directly. Men are also strongly motivated to retaliate aggressively against self-invalidating events more than women (see Bjorkqvist, Osterman, & Lagerspetz, 1994).

One of the explanations of these findings traces gender-based differences in aggression to physiological dispositions, such as higher levels of testosterone among men (e.g., Archer 2006; Book, Starzyk, & Quinsey, 2001). Another explanation emphasizes internalized gender roles based on culturally prevalent gender stereotypes. While men in most cultures are stereotyped as cold, competitive, self-relying and authoritative, women are traditionally believed to be warm, nurturing, caring, and dependent (Fiske, Cuddy, Glick, & Xu, 2002; Kawakami, White, & Langer, 2000). Accordingly, men are expected to display aggressiveness and anti-social behaviors more than women. Specific to the organizational context, previous research has indicated that these beliefs lead to a double standard that gives men greater freedom than women to express negative feelings such as anger (e.g., Black, 1990). Further, interactions between gender and personality traits were assessed in predicting counterproductive work behaviors directed at individuals. Specifically, agreeableness and pleasantness were found to be negative predictors among men (but not women), whereas emotional stability was a negative predictor among women (but not men) (Gonzalez-Mulé, DeGeest, Kiersch, & Mount, 2013). On the other hand, Hershcovis et al.’s (2007) meta-analysis of workplace aggression indicated that there is little gender difference with respect to indirect forms of aggression. Although this meta-analysis does not address the definitions of “direct” versus “indirect” aggression, we may look at several definitions suggested in the literature. For example, Richardson (2014) defines “direct” aggression as more obvious form of aggression, such as direct physical or verbal attack (e.g., yelling, hitting). In contrast, “indirect” aggression described less obvious acts such as those where a person does not confront the target directly (e.g., spreading rumors, damaging property). Further, according to Warren, Richardson, and McQuillin (2011), direct aggressive behaviors involve confronting another person, whereas indirect aggressive behaviors involve attempting to hurt someone by going through another person or object. Indirect aggression was also conceptualized as behaviors bringing harm by rejection or exclusion, including acts of verbal and physical aggression that are unrelated to relationships (e.g., giving “dirty looks”) (see Marshall, Arnold, Rolon- Arroyo, & Griffith, 2015).

Yet, in relation to both types of aggressive and anti-social behavior, including in the organizational context, findings of studies that examined gender differences in terms of mean scores of behaviors are still in question. These scores are mainly obtained from self- report measures or descriptions of external observers, such as colleagues or supervisors. However, while these comparisons provide information on the magnitude of a general tendency of each gender to behave in certain way (and indicate whether any differences in

these tendencies are statistically significant), they led only to rough distinctions but did not provide data regarding the prevalence of more specific behavioral manifestations (see Bjorkqvist et al., 1994; Kogut et al., 1992; Meyers-Levy & Loken, 2015). Since the previous gender differences studies were based on overall or average measure of verbal and physical aggression (e.g., Marshall, et al., 2015) or of direct and indirect aggressive behaviors (e.g., Warren et al., 2011), not a specific measure of them, it would not be possible to identify which subscales or items (included in the mean score) are responsible for gender differences.

With respect to workplace deviance, in particular, Bennett and Robinson (2000) asserted that two conceptual dimensions comprise interpersonal deviance and organizational deviance. Interpersonal deviance is targeted at its members (e.g., managers, supervisors, colleagues), while organizational deviance is targeted at the organization, per se. Although both types of behavior vary along the dimension of severity, this distinction is important since, following Bennett and Robinson (2000), there is a qualitative difference between them and each category of deviant behavior is indicated to be motivated by different factors. Accordingly, past research suggested separate targets of workplace deviance in order to demonstrate different patterns of such behavior (e.g., Hershcovis et al., 2007).

The Present Research

A review of empirical ethical decision making literature, in the scope of eight years, conducted by O’Fallon and Butterfield (2005) showed that relatively small amount of studies has examined moderators to the ethical decision-making process. Yet, they argue that examining interaction effects has a potential to broaden the understanding of the decision making addressing ethical issues. Exploring individual factors relevant to ethical decisions as well as using various statistical methods are said to be the strengths of organizational behavior research (O’Fallon & Butterfield, 2005). Accordingly, in the present study we aimed to explore the moderating effect of employee’s gender, implementing a profile analysis paradigm. Further, although there is an empirical support for high correlation between interpersonal and organizational deviance, the two types were found to have different relationships with personal characteristics such as Big Five variables, lending support for assessing these types of deviance separately (see Berry et al., 2007; Robinson & Bennett, 1995). Therefore, Berry et al. (2007) propose that organizations concerned with instances of employees’ deviance should examine individual traits correlated with such behavior.

We focused on examining the role of gender in the two categories of workplace deviance recorded above (i.e., interpersonal and organizational) through the employment of a statistical method, namely, Profile Analysis via Multidimensional Scaling (PAMS; Kim, Annunziato, & Olatunji, 2017; Kim, Frisby, & Davison, 2004). Unlike the previous studies that examined mainly mean differences in gender, we sought to examine which counterproductive work behaviors measuring variables (relevant to interpersonal and organizational deviance) made gender differences. This novel profile approach augments the mean difference analyses, making it possible to compare specific deviance behavior indicators between males and females included in the profiles. Therefore, the current profile analysis (PAMS) includes much richer information to assess deviance behaviors displayed by men and women more specifically.

Although the statistical method of multidimensional scaling has been used to study organizational deviance in the past (indicating different dimensions of workplace deviant behavior; see Robinson & Bennett, 1995), the PAMS technique utilizes nonmetric multidimensional scaling (MDS) estimating scale values for input variables (in the present project, workplace deviant behaviors) and “interprets” an array of scale values in a given dimension as their

48Gender Profiles of Workplace Individual and Organizational Deviance

profile pattern, which is considered to represent a core profile for individual person response profiles. Interpreting dimensions as core profiles has been supported and validated by numerous studies (e.g., Frisby & Kim, 2008; Kim, 2013; Kim et al., 2017; Kim, Davison, & Frisby, 2007; McKay et al., 2014; Olatunji, Kim, & Wall, 2015; Sosinsky & Kim, 2013). Specifically, person p’s response profile can be replicated with PAMS parameter estimates, indicating a level index for person p (mean of input variable scores) + person p’s weights on core profiles + a residual.

The benefits of the PAMS method can be summarized as follows: (1) PAMS summarizes numerous person profiles with a few core profiles, to make it easier to understand the individual response profiles; (2) accordingly, PAMS allows us to explain participants’ response profiles in terms of both observed mean scores and latent dimension scores; (3) further, PAMS interprets dimensions not as single construct factors but as core profiles that include all input variables as a constellation of individual persons’ profiles.

In sum, our research sought to contribute to the literature addres- sing individual differences in workplace deviant behavior through the discovering of interpersonal and organizational behavior patter- ns. Specifically, given the inconsistent evidence on gender differences in workplace deviance, in the present exploratory research we aimed to investigate further manifestations of two particular types of wor- kplace deviance implementing an analysis technique which is capa- ble to unveil profiles of the deviant behavior rather than comparing between mean scores of the behaviors in question.

Method

Participants

The participants, all volunteers, were 122 employees (66 men, 56 women; mean age = 42.20, SD = 7.82) employed at large electricity

supplier company in Israel, in the departments of operations and logistics (41.2%), finance and economics (35.3%), engineering (18.6%), accounting (3.9%), and sustainability (1%). The participants were sampled individually upon invitation by the experimenter; all employees approached agreed to participate in the study. Forty- five percent of employees stated that they were married, 34% were divorced, and 21% indicated that they were unmarried but had stable relations.

Informed consent was obtained from all individual participants included in the study.

Procedure and Measures

The participants signed up for a study examining “Issues regarding workplaces”. An experimenter explained that the study would involve answering paper-and-pencil questionnaires and that the participants were expected to give honest answers representing their actual feelings and thoughts. After completing the measures, all participants were debriefed.

Workplace deviance measure. The participants were asked to complete a 24-item measure of Workplace Deviance Behavior - WDB (Bennett & Robinson, 2000). This measure was reported to have considerable construct validity for its two scales – interpersonal and organizational deviance. Moreover, the scales have also showed discriminant validity as they were not highly correlated with other concepts of organizational behavior (see Bennett & Robinson, 2000 for detailed description of validation analyses). The present research engaged a Likert scale ranging from 1 (very untypical) to 6 (very typical), reflecting participants’ judgment of each behavior as typical for the employees in their organization. Specifically, the two dimensions of WDB were assessed such that 7 items were incorporated into an interpersonal deviance scale (Cronbach’s alpha = .82, M = 2.21, SD = 0.71) and 17 items to an organizational

Table 1. Inter-correlational Matrix for Both Genders 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

1. org1 2. org2 .44** 3. org3 .26** .39** 4. org4 .59** .43** .34** 5. org5 .54** .48** .38** .71** 6. org6 .11 .26** .55** .27** .26** 7. org7 .14 .19* .47** .21* .22* .59** 8. org8 .32** .38** .55** .42** .45** .61** .65** 9. org9 .46** .45** .53** .57** .50** .41** .27** .51**

10. org10 .13 .31** .52** .20* .29** .56** .54** .62** .36** 11. org11 .36** .50** .51** .35** .44** .44** .39** .43** .58** .50** 12. org12 .42** .43** .49** .49** .46** .44** .38** .48** .63** .46** .73** 13. org13 .32** .40** .53** .52** .46** .55** .46** .56** .69** .46** .64** .78** 14. org14 .43** .54** .47** .51** .57** .36** .32** .47** .63** .37** .63** .64** .67** 15. org15 .40** .22* .18* .50** .50** .16 -.01 .12 .38** .07 .27** .37** .37** .50** 16. org16 .33** .28** .38** .45** .41** .42** .37** .34** .34** .37** .42** .48** .40** .41** .53** 17. org17 .25** .33** .32** .34** .53** .31** .23* .25** .35** .29** .40** .47** .42** .43** .51** .77** 18. org18 .28** .29** .51** .24** .32** .47** .37** .41** .37** .54** .56** .46** .46** .45** .13 .39** .35** 19. org19 .36** .31** .37** .45** .50** .34** .25** .27** .40** .34** .42** .49** .47** .48** .58** .67** .68** .49** 20. org20 .30** .27** .54** .21* .33** .60** .46** .50** .41** .56** .54** .50** .44** .39** .09 .39** .31** .66** .39** 21. org21 .13 .15 .48** .10 .16 .59** .60** .47** .31** .59** .47** .43** .38** .32** .01 .41** .28** .59** .34** .77** 22. org22 .01 .17 .48** .12 .11 .52** .34** .33** .23** .41** .31** .32** .34** .16 .02 .21* .20* .41** .20* .61** .58** 23. org23 .17 .14 .39** .25** .28** .26** .30** .27** .32** .31** .45** .49** .50** .37** .20* .29** .28** .35** .28** .37** .38** .45** 24. org24 .52** .44** .34** .67** .98** .23** .21* .44** .48** .27** .42** .46** .43** .54** .49** .40** .52* .29** .48** .30** .14 .10 .29** 25. ID .45** .46** .67** .62** .59** .67** .57** .70** .66** .62** .77** .75** .75** .66** .33** .56** .46** .74** .56** .73** .64** .52** .62** .57** 26.OD .57** .58** .64** .67** .75** .54** .46** .61** .74** .55** .70** .77** .76** .77** .59** .70** .70** .57** .73** .57** .48** .38** .46** .72** .85**

*p < .05, **p < .01.

49 L. Chernyak-Hai et al. / Journal of Work and Organizational Psychology (2018) 34(1) 46-55

deviance scale (Cronbach’s alpha = .92, M = 2.90, SD = 0.79). Items reflecting interpersonal workplace deviance included items that related to a colleague at work who “Said something hurtful to someone at work” or “Made [an] ethnic/religious/racial remark or

joke at work” or “Publicly embarrassed someone at work”. Similarly, representative items focusing on organizational workplace deviance concerned colleagues who “Falsified a receipt to gain more money” or “Intentionally worked slower” or who “Discussed confidential

Table 3. Inter-correlational Matrix for Women 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

1. org1 2. org2 .41** 3. org3 .12 .38** 4. org4 .55** .43** .19 5. org5 .43** .49** .34** .62** 6. org6 .00 .28* .54** .22 .29* 7. org7 .06 .22 .52** .07 .22 .70** 8. org8 .25 .43** .57** .32* .44** .68** .64** 9. org9 .41** .48** .30* .45** .47** .30* .15 .42** 10. org10 .07 .33* .61** .14 .27* .56** .51** .67** .30* 11. org11 .27* .57** .50** .17 .48** .33* .30* .41** .49** .56** 12. org12 .39** .47** .29* .25 .46** .26 .21 .46** .48** .47** .68** 13. org13 .21 .38** .39** .33* .37** .47** .42** .51** .51** .49** .54** .70** 14. org14 .28* .55** .38** .28* .47** .26* .28* .36** .51** .37** .65** .58** .58** 15. org15 .19 .09 -.01 .30* .38** -.01 -.18 -.14 .21 -.12 .14 .29* .23 .38** 16. org16 .18 .26* .39** .27* .35** .38** .30* .29* .23 .34* .39** .43** .32* .34* .40** 17. org17 .09 .32* .35** .13 .44** .29* .21 .25 .29* .29* .42** .53** .42** .29* .40** .78** 18. org18 .14 .22 .44** .02 .21 .42** .34** .33* .21 .68** .53** .40** .36** .34** -.19 .32* .32* 19. org19 .26 .28* .32* .28* .36** .32* .24 .18 .33* .37** .48** .47** .46** .34** .35** .74** .74** .46** 20. org20 .31* .27* .49** .11 .34* .52** .48** .55** .32* .58** .48** .44** .24 .29* -.20 .28* .23 .64** .37** 21. org21 .14 .21 .51** .02 .18 .58** .56** .49** .24 .59** .42** .39** .24 .26* -.10 .45** .37** .59** .43** .78** 22. org22 .11 .23 .45** .15 .21 .61** .45** .39** .09 .44** .25 .26 .20 .08 .00 .33* .38** .48** .42** .58** .66** 23. org23 .20 .19 .37** .26 .38** .33* .20 .25 .21 .35** .51** .50** .49** .43** .30* .38** .43** .46** .46** .30* .31* .37** 24. org24 .43** .46** .32* .59** .99** .27* .20 .42** .45** .24 .48** .47** .36** .46** .40** .35** .45** .18 .36** .30* .15 .20 .39** 25. ID .36** .52** .64** .46** .56** .68** .50** .67** .51** .69** .74** .63** .60** .55** .07 .51** .45** .71** .56** .69** .63** .57** .70** .54** 26. OD .46** .65** .56** .48** .68** .49** .43** .56** .67** .56** .71** .75** .70** .67** .37** .68** .72** .50** .72** .51** .51** .45** .53** .66** .80**

*p < .05, **p < .01.

Table 2. Inter-correlational Matrix for Men

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 1. org1 2. org2 .49** 3. org3 .36** .43** 4. org4 .63** .45** .44** 5. org5 .63** .49** .39** .75** 6. org6 .21 .25* .56** .30* .24* 7. org7 .21 .16 .43** .33** .23 .49** 8. org8 .40** .34** .54** .52** .48** .53** .64** 9. org9 .49** .44** .68** .65** .52** .50** .37** .59** 10. org10 .19 .30* .45** .25* .31* .55** .58** .57** .41** 11. org11 .45** .47** .53** .50** .42** .54** .48** .45** .65** .45** 12. org12 .43** .43** .63** .65** .47** .58** .52** .51** .73** .45** .78** 13. org13 .42** .42** .66** .67** .52** .61** .49** .60** .82** .45** .73** .85** 14. org14 .56** .54** .55** .66** .63** .44** .35** .55** .71** .38** .62** .69** .74** 15. org15 .57** .33** .31* .63** .59** .29* .12 .33** .48** .23 .38** .42** .47** .57** 16. org16 .44** .31* .38** .57** .47** .46** .43** .37** .41** .40** .43** .50** .46** .48** .61** 17. org17 .39** .35** .29* .49** .59** .33** .25* .25* .38** .28* .37** .41** .42** .53** .59** .77** 18. org18 .43** .35** .57** .43** .42** .51** .39** .47** .51** .42** .61** .52** .55** .55** .38** .46** .39** 19. org19 .45** .35** .41** .59** .60** .36** .26* .35** .46** .30* .38** .49** .48** .60** .75** .63** .63** .53** 20. org20 .29* .28* .59** .30* .33** .67** .44** .45** .50** .56** .61** .55** .64** .48** .32** .49** .38** .69** .41** 21. org21 .13 .09 .47** .18 .14 .59** .64** .43** .38** .60** .53** .48** .53** .36** .11 .39** .21 .58** .26* .75** 22. org22 -.06 .11 .52** .09 .05 .44** .25* .27* .32** .38** .37** .37** .44** .22 .03 .14 .07 .35** .02 .64** .51** 23. org23 .12 .12 .40** .23 .20 .20 .40** .31* .40** .26* .39** .47** .51** .32** .13 .21 .13 .27* .09 .45** .45** .54** 24. org24 .60** .43** .35** .72** .98** .21 .22 .47** .50** .30* .37** .45** .49** .60** .56** .44** .57** .39** .58** .31* .13 .02 .21 25. ID .54** .44** .69** .73** .61** .65** .61** .71** .77** .54** .80** .82** .86** .73** .52** .61** .49** .78** .57** .76** .65** .47** .55** .58** 26.OD .66** .56** .68** .80** .79** .56** .48** .64** .79** .52** .69** .78** .82** .83** .73** .71** .70** .65** .75** .63** .47** .33** .39** .76** .88**

*p < .05, **p < .01.

50Gender Profiles of Workplace Individual and Organizational Deviance

organizational information with an unauthorized person” (see Tables 1-3 for correlation matrices).

Since the employees answered both interpersonal and organizational deviant behavior items, it could be argued that common method variance might be a limitation in the present study. Yet, note that the two dimensions of workplace deviance have been reported to be highly correlated in past research (e.g., r = .86 in Bennett & Robinson, 2000; r = .96 in Lee & Allen, 2002). Moreover, in order to address this point directly, we employed the Harman’s Single-Factor Test to assess the degree to which inter-correlations among the items of the two dimensions might indicate a common method bias (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). The single-factor that emerged from the analysis accounted only for 45.11% of the expected variance. While the result does not rule out completely the possibility of same-source bias, according to Podsakoff et al. (2003) less than 50% of the explained variance accounted for by the first emerging factor indicates that common method bias is an unlikely explanation of our investigation findings.

Function of coordinates (or scale values) in PAMS. By conven- tion, scale values along a dimension are calibrated such that they can be considered a set of contrast coefficients describing score response patterns in the data. Individual differences in score response patterns can be described in terms of the scale value patterns derived from the scaling, and a set of scale values in each dimension is considered a core profile for person response profiles of observed scores. To iden- tify core profiles, we utilized the multidimensional scaling package, SMACOF, embedded in the R domain (http://cran.r-project.org) with an “ordinal” (or nonmetric scaling) option. Since SMACOF or PAMS does not estimate standard errors for its coordinates (or scale values), and when we interpret dimensions as core profiles, our interpreta- tion could be misleading without any statistical test for the coordi- nates. For this reason, we will, utilizing the bootstrapping method, estimate bootstrap standard errors for core profile coordinates. If the (z) ratios between coordinates and their standard errors were equal to or larger than |2|, we would consider those coordinates to be sta- tistically significant.

Results

In the present report, we label interpersonal deviance as ID and or- ganizational deviance as OD. We start with presenting PAMS’ results on: 1) relations between gender and inter-personal deviance versus organizational deviance and 2) profiles of OD & ID separately for men and women.

Identifying ID Core Profiles for Males and Females

We identified two core profiles of interpersonal workplace deviance for males and females, respectively. The STRESS values for the two-dimensional solutions for males (STRESS =.04) and females (STRESS =.01) indicated a good model fit. Note that STRESS less than .05 is usually considered a good fit (Kruskal, 1964). The two ID core profile accounted for 81% of total variance occurring in male responses on ID items and accounted for 74% of total variance occurring in female responses on ID items.

The first ID core profiles. The patterns of the first core profiles were virtually identical for males and females, and the correlation between them was .99. The first core profile usually represents an array of item mean scores (or a mean profile) and the correlations between them for both male and female were 1.00 (see Table 4). Figure 1 depicts the mean profiles for males and females along with the first dimension (core) profile. As shown in the figure, the mean and the first dimension profiles were identical and there was no gender difference detected in any individual deviance variables.

Table 4. Means of Interpersonal Deviance Items on the First Core Profile for Males and Females

Male MFemale MItem

4.274.34 int4 made fun of someone at work

1.491.48 int6 said something hurtful to someone at work

1.701.67 int8 made ethnic/religious/racial/remark or joke at work

2.052.27 int11 cursed at someone at work

1.951.88 int18 played a mean prank on someone at work

1.611.66 int20 acted rudely toward someone at work

2.232.38 int23 publicly embarrassed someone at work

5

4

3

2

1

0 int4 int6 int8

M_Mean F_Mean

int11 int18 int20 int23

3,0

2,5

2,0

1,5

1,0

0,5

0,0

-0,5

-1,0

int4 int6 int8

M_Dim 1 F_Dim 1

int11 int18 int20 int23

Figure 1. The ID Mean Profiles Juxtaposed with the ID First Dimension (Core) Profiles.

The second ID core profiles. As shown in Figure 2, there were significant differences in the second core profiles between males and females. The male profile had a plateau from Int4 to int20 and then a sharp drop to int23, whereas the female profile had a zigzag pattern. Considering the content of the items where the gender differences appeared – “Made ethnic/religious/racial remark or joke at work” (int8), “Cursed at someone at work” (int11), and “Publicly embarrassed someone at work” (int23) – we labeled the second core profiles as “Rudeness”.

Table 5. Standardized Profile Coordinates, Bootstrap Standard Errors, and Effect Sizes for Interpersonal Deviance Items Obtained from the Second Core Profile

Effect sizeFemaleMaleItem

0.90 1.67: -0.57

(0.05) 1.70: 0.34

(0.17)

int8 made ethnic/religious/racial remark or joke at work

2.22 2.27: -1.66

(0.06) 2.05: 0.56

(0.17) int11 cursed at someone at work

-3.74 2.38: 1.20

(0.06) 2.23: -2.54

(0.11) int23 publicly embarrassed someone at work

Note. The values are observed means that are used as core profile coordinates and the values in parentheses are coordinate standard errors estimated by bootstrapping. The z-ratios between coordinates and standard errors are equal to or larger than |2.00|, which indicates statistical significance for the coordinates in terms of z-test.

51 L. Chernyak-Hai et al. / Journal of Work and Organizational Psychology (2018) 34(1) 46-55

M_Dim 2

int4 int6 int8 int11 int18 int20 int23

F_Dim 2

1,5

1,0

0,5

-0,5

-1,0

-1,5

-2,0

-2,5

-3,0

Figure 2. The ID Second Dimension (Core) Profile: Rudeness.

Table 6. Means of Organizational Deviance Items on the First Core Profile for Males and Females

Male MFemale MItem

5.055.11 org1 worked on a personal matter instead of work for your employer

3.623.18 org2 taken property

2.122.18 org3 spent too much time fantasizing or daydreaming instead of working

3.883.82 org5 falsified a receipt to gain more money

1.641.68 org7 taken an additional or a longer break than is acceptable at your workplace

3.383.57 org9 come in late to work without permission

1.411.54 org10 littered work environment

2.292.38 org12 told someone about the lousy place where you work

2.422.38 org13 lost temper

2.882.77 org14 neglected to follow your boss’s instructions

4.064.16 org15 intentionally worked slower

3.113.20 org16 discussed confidential organizational information with an unauthorized person

3.123.23 org17 left work early without permission

3.033.05 org19 left work for someone else

1.581.68 org21 used an illegal drug or consumed alcohol on the job

1.731.65 org22 put little effort into your work

3.883.80 org24 dragged out work in order to get overtime

Note. M = Mean.

We compared the scale values between genders in terms of their standard deviation (SD) units and included only 0.8 or above SD units for further investigation (see Table 5). This rationale was based on the large effect size (0.8 or above) of Cohen’s d, which represents SD unit differences (between two group means). Moreover, to test statistical significance of the core profile coordinates (or scale values), we estimated bootstrap standard errors for the coordinates and if z-ratios between coordinates and their standard errors were less than |2|, those coordinates would be (statistically) insignificant, and we would consider them to be zero. The male coordinates of the item “Made ethnic/religious/racial remark or joke at work” (int8) had 0.9 SD units above the female scale value, and also the male scale value

of the item “Cursed at someone at work” (int11) had 2.22 standard deviation units above the female scale value. However, the female scale value of “Publicly embarrassed someone at work” (int23) had 3.74 standard deviation units above the male scale value. The results imply that men reported more “Ethnic/religious/racial remarks” or “Jokes at work” and “Cursing someone at work” than women; in contrast, women reported more “Publicly embarrassed someone” at work. The z-ratios for all these coordinates were lager than |2| and considered to be statistically significant.

Identifying OD Core Profiles for Males and Females

We identified three core profiles of organizational workplace deviance for males and females, respectively. The STRESS values for the three-dimensional solutions for males (STRESS = .03) and females (STRESS = .02) indicated a good model fit. The three OD core profile accounted for 73% of total variance occurring in male responses on OD items and also accounted for 73% of total variance occurring in female responses on OD items.

The first OD core profiles. The patterns of the first core profiles were virtually identical for males and females, and the correlation between them was .99. The first core profile usually represents an array of item mean scores (or a mean profile) and the correlations between them for both male and female were 1.00 (see Table 6). Figure 3 depicts the mean profiles for males and females along with the first dimension (core) profile. As shown in the figure, the mean and the first-dimension (core) profiles were identical and there was no gender difference detected in any organizational deviance variables.

M_Mean F_Mean

6,0

5,0

4,0

3,0

2,0

1,0

0,0

or g1

or g2

or g3

or g5

or g7

or g9

or g1

0 or

g1 2 or

g1 3

or g1

4 or

g1 5 or

g1 6 or

g1 7 or

g1 9 or

g2 1 or

g2 2 or

g2 4

M_Dim 1 F_Dim 1

2,5 2,0 1,5 1,0 0,5 0,0

-0,5 -1,0 -1,5 -2,0

or g1

or g2

or g3

or g5

or g7

or g9

or g1

0 or

g1 2 or

g1 3

or g1

4 or

g1 5 or

g1 6 or

g1 7 or

g1 9 or

g2 1 or

g2 2 or

g2 4

Figure 3. The OD Mean Profiles Juxtaposed with the OD First Dimension (Core) Profiles.

The second OD core profiles. As shown in Figure 4, there were significant differences in the second core profiles between males and females. According to the content of these items, we labeled this dimension “Free-rider phenomenon”. In the “Falsified a receipt to gain more money” item (org5), females had 0.74 SD units larger than males. It is important to note that the original effect size was 0.90, but its size was adjusted to 0.74 because z-ratio between the

52Gender Profiles of Workplace Individual and Organizational Deviance

male org5 coordinate and its bootstrap standard error was less than |2| and we considered the coordinate to be zero. However, all other coordinates’ z-ratios were larger than |2|, implying their statistical significance. For “Left work for someone else” (org19), males had 0.87 SD units larger than females; and for “Dragged out work in order to get overtime” (org24), females had 0.99 SD units higher than males. In sum, men reported more “Leaving work for someone else” than women, whereas, women reported more “Falsified a receipt to gain more money” and “Dragged out work in order to get overtime”.

Table 7. Standardized Profile Coordinates, Bootstrap Standard Errors, and Ef- fect Sizes for Organizational Items Obtained from the Second and Third Core Profiles

Effect sizeFemaleMaleItem

-0.74 (recalculated)

3.82: 0.74 (0.09)3.88: -0.16 (0.09) org5 (2nd Profile) falsified a receipt to gain more money

0.873.05: -1.98 (0.13)3.03: -1.12 (0.12) org19 (2nd Profile) left work for someone else

-0.993:80: 0.75 (0.10)3.88: -0.23 (0.10) org24 (2nd Profile) dragged out work in order to get overtime

1.573.18: 0.80 (0.20)3.62: 2.37 (0.18) org2 (3rd Profile) taken property

0.901.54: -0.36 (0.07)1.41: 0.54 (0.06) org10 (3rd Profile) littered work environment

1.192.38: -1.95 (0.13)2.42: -0.77 (0.09) org13 (3rd Profile) lost temper

-1.31 (recalculated)

4.16: 0.12 (0.17)4.06: -1.31 (0.16) org15 (3rd Profile) intentionally worked slower

-0.923.20: 0.28 (0.11)3.11: -0.64 (0.16)

org16 (3rd Profiel) discussed confidential org. information with an unautho- rized person

Note. The values are observed means that are used as core profile coordinates and the values in parentheses are coordinate standard errors estimated by bootstrapping. Except the male org5 and the female org15 z-ratios, all other z-ratios between coordinates and standard errors are equal to or larger than |2.00|, which indicates statistical significance for the coordinates. Since the male org5 and the female org15 z-ratios are less than |2.00|, we may consider their coordinates statistically to be insignificant (considered to be 0), and those effect sizes are recalculated.

M_Dim 2 F_Dim 2

2,5 2,0 1,5 1,0 0,5 0,0 -0,5 -1,0 -1,5 -2,0 -2,5

or g1

or g2

or g3

or g5

or g7

or g9

or g1

0 or

g1 2 or

g1 3

or g1

4 or

g1 5 or

g1 6 or

g1 7 or

g1 9 or

g2 1 or

g2 2 or

g2 4

Figure 4. The OD Second Dimension (Core) Profiles: Free-rider Phenomenon.

The third OD core profiles. According to the content of these items we named this dimension “Bringing damage to work environment.” In the coordinate for the “Taken property” item (org2), males had 1.57 SD units higher than females; for the “Littered work environment” item (org10), males had 0.90 SD units higher than females; and for “Lost temper” (org13), males had 1.19 SD units higher than females. All these coordinates’ z-ratios were larger than |2|. However, for the “Intentionally worked slower” item (org15), females had 1.43 SD units higher than males, and for “Discussed confidential org. information with an unauthorized person” (org16), females had 0.92 SD units higher than males. In sum, men reported more “Taken property”, “Littered work environment”, and “Lost temper” than women,

whereas women reported more “Intentionally worked slower” and “Discussed confidential org. information with an unauthorized person” (see Figure 5).

or g1

or g2

or g3

or g5

or g7

or g9

or g1

0 or

g1 2 or

g1 3

or g1

4 or

g1 5 or

g1 6 or

g1 7 or

g1 9 or

g2 1 or

g2 2 or

g2 4

M_Dim 3 F_Dim 3

3,0 2,5 2,0 1,5 1,0 0,5 0,0

-0,5 -1,0 -1,5 -2,0 -2,5

Figure 5. The OD third dimension (core) profile: Brining damage to work en- vironment.

Discussion

The present investigation sought to examine gender differences in two types of workplace deviance – interpersonal and organizational – by adopting a profile analysis approach that allows us to discover specific behavioral profiles and, accordingly, to compare gender differences in their behavioral patterns. Unlike comparing mean differences between genders, we compared the effect sizes for all interpersonal deviance (ID) and organizational deviance (OD) measuring variables included in male and female profiles. Since all measuring variables were included in a single profile, we could easily inspect gender differences appearing in specific variables simultaneously across latent dimensions. Hence, the profile analysis adopted here carries gender difference information not only in observed mean scores of individual input variables (represented in the first dimension or core profile) but also in latent scores of them (represented in the subsequent dimension or core profile coordinates). The unique aspect of the present study consists in casting a further “luminous” light on gender differences in organizational and interpersonal deviance behaviors relative to the conventional mean score-comparison approach. As shown in Tables 5 and 7, there was no virtual mean difference of ID or OD items between males and females but substantial gender differences were found in profile coordinates.

The gender differences found in the core profiles are consistent with the past research findings pointing to lesser direct aggression among women compared to men (e.g., Arnocky, Sunderani, Miller, & Vaillancourt, 2012; Bjorkqvist et al., 1994; Card, Stucky, Sawalani, & Little, 2008; Cross, Copping, & Campbell, 2011). However, none of the previous studies has reported all these gender differences appearing in ID and OD measurement altogether as shown in the current study. Accordingly, of interest are present findings indicating that the gender differences were apparent on both inter-personal and organizational levels. Specifically, direct aggression was found in males on inter-personal level items, such as “Ethnic/religious/racial remarks” and “Cursing” and on organizational level items, such as “Leaving work for someone else”, “Taking property”, “Littering work environment”, and “Losing temper”.

In contrast, women reported more indirect aggression on inter- personal level item, “Publicly embarrassing someone”, and on organizational level items, “Falsifying receipts”, “Dragging out work”, “Working slower”, and “Discussing confidential organizational information” (as shown in Tables 3 and 5). Therefore, the current study reveals that gender differences in aggressive workplace behavior are not only those apparent in inter-personal relations, as studied by the previous research, but also when directed towards the organization, its goals, or values.

53 L. Chernyak-Hai et al. / Journal of Work and Organizational Psychology (2018) 34(1) 46-55

Interestingly, no gender difference was found in mean score profiles which were in fact the first core profiles and this finding complements the conventional mean difference methods (e.g., t-test or ANOVA). Hence, if we relied on the conventional mean difference approach, we would not find any gender differences with our current data.

Based on gender differences appearing in core profile patterns, the present findings may imply further that the organizational context, particularly, brings diverse suggestions for interpersonal deviance for males and females, beyond the differentiation of directness of aggressive behavior. This notion, which could be the subject of further research, is consistent with the findings of Santos and Eger (2014) indicating that organizational context has specific implications for aggressive behavior, as males exhibiting higher organizational workplace deviance, in contrast to interpersonal workplace deviance, were those with over five years working experience in the company.

There may be psychological factors, such as stress related problems (e.g., O’Leary-Kelly, Griffin, & Glew, 1996), experiencing low self-esteem, increased lack of confidence at work, and physical and psychological pains (e.g., Griffin, O’Leary, & Collins, 1998), that account for these gender differences. One plausible explanation is that mirroring the picture of aggressive behavior in broader social contexts, direct deviant behaviors in the workplace characterize men more than women. Another possibility is that, given gender differences in the content of deviant behaviors shown in the present work, men and women interpret differently the plausibility of certain behavior in organizational context. For example, status differences of men and women in the workplace may be a possible reason. Past research has shown the relevance of gender to different aspects of career and employment. Although much has changed through the years, women still enjoy lesser career advancement and employment prestige compared to men (Timberlake, 2005) and perceived as more “applicable” for support roles than for leadership roles (Eagly & Karau, 2002; Eagly & Sczesny, 2009). Accordingly, women’s reports of indirect aggression on interpersonal deviance (e.g., publicly embarrassing someone) and organizational deviance items (e.g., falsifying receipts, dragging out work, working slower, and discussing confidential organizational information) may be attributed to perceived threat on employment status. Women, being aware of their initially lower status compared to men as well as of social stereotypes allowing men to exhibit aggression more than to women (e.g. Heliman & Chen, 2005; Kark, Waismel-Manor, & Shamir, 2012), may forecast high costs in manifesting direct aggression and choose indirect acts instead.

Limitations

There are a couple of limitations in the current investigation. We used a self-reported measure of workplace deviance, as done in past studies, and yet these responses may not represent participants’ true deviance behaviors. Since the reported acts are clearly negative, possibly some employees refrained from reporting the actual frequency with which they exhibited these acts. Yet, the assumption here is that employees are aware of the range of their deviant workplace acts more than anyone else and, since not every behavior is easily observed, self-reports are primary source of relevant information which specifically useful if the respondents are guaranteed anonymity (Bennett & Robinson, 2000). Relatedly, a concern may be raised regarding common method bias. However, research has pointed to misconceptions in arguing for common method bias due to the mere usage of self-report measures, supporting their implementation when reasonable (see Conway & Lance, 2010). A relatively recent meta-analysis which assessed the explained variance in reports of organizational deviance did not find significant increase by adding other-reported (e.g., peers or supervisors) measures to self-reported workplace deviance measures (see Berry, Carpenter, & Barratt, 2012).

Further, we report the results of Harman’s single-factor test used to assess the potential for common method bias empirically, pointing that this may not be considered as a problem in the present study. A different limitation of the present study is a relatively small sample size that could affect the statistical power of attained results. Yet, we estimated standard errors of core profile coordinates, utilizing the bootstrap method, to assess their statistical meaningfulness. Finally, being an exploratory investigation of the patterns of workplace defiance, the present research did not examine specific predictions on the variables which may explain the explored differences, mediate, or moderate the relations. Future research may explore male and female employees’ perceptions of status differences followed by evaluations of possible costs versus merits if certain behavior will be implemented. Such examination will raise employers’ awareness of different forms of organizational deviance as influenced by socio- demographic variables, in this case, gender.

In sum, the present findings bring both empirical and practical implications to the research of workplace deviant behavior. While comparing mean reports of men and women has pointed to inconsistent results (e.g., Black, 1990; Gonzalez-Mulé, et al., 2013; Hershcovis et al., 2007), the present findings reveal that employees’ gender predicts differential forms of interpersonal and organizational deviance and demonstrate the unique value in employing the profile analysis capable of discovering such patterns. Theorists and practitioners should be aware of the relevance of gender not to the mere question who is more prone to workplace deviance – men or women –, but rather what types of negative acts we may expect from the two genders and what are the organizational experiences that encourage them to display these acts.

Ethical Standards

All procedures performed in this study were in accordance with the ethical standards of the authors A and C institutional committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants.

Conflict of Interest

The authors of this article declare no conflict of interest.

References

Archer, J. (2006). Testosterone and human aggression: an evaluation of the challenge hypothesis. Neuroscience and Biobehavioral Reviews, 30, 319-345. https://doi.org/10.1016/j.neubiorev.2004.12.007

Arnocky, S., Sunderani, S., Miller, J. L., & Vaillancourt, T. (2012). Jealousy mediates the relationship between attractiveness comparison and females’ indirect aggression. Personal Relationships, 19, 290-303. https://doi.org/10.1111/j.1475-6811.2011.01362.x

Aubé, C., Rousseau, V., Mama, C., & Morin, E. (2009). Counterproductive behaviors and psychological well-being: The moderating effect of task interdependence. Journal of Business and Psychology, 24, 351-361. https://doi.org/10.1007/s10869-009-9113-5

Bennett, R. J., & Robinson, S. L. (2000). Development of a measure of workplace deviance. Journal of Applied Psychology, 85, 349-360. https://doi.org/10.1037/0021-9010.85.3.349

Berry, C. M., Carpenter, N. C., & Barratt, C. L. (2012). Do other-reports of counterproductive work behavior provide an incremental contribution over self-reports? A meta-analytic comparison. Journal of Applied Psycogy, 97, 613-636. http://doi.org/10.1037/a0026739

Berry, C. M., Ones, D. S., & Sackett, P. R. (2007). Interpersonal deviance, organizational deviance, and their common correlates: A review and meta-analysis. Journal of Applied Psychology, 92, 410-424. https://doi. org/10.1037/0021-9010.92.2.410

Bjorkqvist, K., Osterman, K., & Lagerspetz, K. M. (1994). Sex differences in covert aggression among adults. Aggressive behavior, 20, 27-33. https://doi. org/10.1002/1098-2337(1994)20:1<27::AID-AB2480200105>3.0.CO;2-Q

Black, K. S. (1990). Can getting mad get the job done? Working Woman, 15, 86-90.

54Gender Profiles of Workplace Individual and Organizational Deviance

Bodankin, M., & Tziner, A. (2009). Constructive deviance, destructive deviance and personality: how do they interrelate? Amphiteatru Economic Journal, 11, 549-564. https://doi.org/EconPapers.repec.org/ RePEc:aes:amfeco:v:11:y:2009:i:26:p:549-565

Book, A. S., Starzyk, K. B., & Quinsey, V. L. (2001). The relationship between testosterone and aggression: A meta-analysis. Aggression and Violent Behavior, 6, 579-599. https://doi.org/10.1016/S1359-1789(00)00032-X

Bruk-Lee, V., & Spector, P. E. (2006). The social stressors counterproductivity work behaviors link: Are conflicts with supervisors and coworkers the same? Journal of Occupational Health Psychology, 11, 145-156. https:// doi.org/10.1037/1076-8998.11.2.145

Card, N. A., Stucky, B. D., Sawalani, G. M., & Little, T. D. (2008). Direct and indirect aggression during childhood and adolescence: A meta- analytic review of gender differences, inter-correlations, and relations to maladjustment. Child Development, 79, 1185-1229. https://doi. org/10.1111/j.1467-8624.2008.01184.x

Chen, P. Y., & Spector, P. E. (1992). Relationships of work stressors with aggression, withdrawal, theft and substance use: An exploratory study. Journal of Occupational and Organizational Psychology, 65, 177-184. https://doi.org/10.1111/j.2044-8325.1992.tb00495.x

Cohen-Charash, Y., & Mueller, J. S. (2007). Does perceived unfairness exacerbate or mitigate interpersonal counterproductive work behaviors related to envy? Journal of Applied Psychology, 92, 666-680. https://doi.org/10.1037/0021-9010.92.3.666

Conway, J. M., & Lance, C. E. (2010). What reviewers should expect from authors regarding common method bias in organizational research. Journal of Business and Psychology, 25, 325-334. https://doi. org/10.1007/s10869-010-9181-6

Cross, C. P., Copping, L. T., & Campbell, A. (2011). Sex differences in impulsivity: a meta-analysis. Psychological Bulletin, 137, 97-130. https://doi.org/10.1037/a0021591

Dalal, R. S. (2005). A meta-analysis of the relationship between organizational citizenship behavior and counterproductive work behavior. Journal of Applied Psychology, 90, 1241-1255. https://doi. org/10.1037/0021-9010.90.6.1241

Diefendorff, J. M., & Mehta, K. (2007). The relations of motivational traits with workplace deviance. Journal of Applied Psychology, 92, 967-977. https://doi.org/10.1037/0021-9010.92.4.967

Dilchert, S., Ones, D. S,, Davis, R. D., & Rostow, C. D. (2007). Cognitive ability predicts objectively measured counterproductive work behaviors. Journal of Applied Psychology, 92, 616-627. https://doi. org/10.1037/0021-9010.92.3.616

Eagly, A. H., & Karau, S. J. (2002). Role congruity theory of prejudice toward female leaders. Psychological Review, 109, 573-598. https://doi. org/10.1037/0033-295X.109.3.573

Eagly, A. H., & Sczesny, S. (2009). Stereotypes about women, men, and leaders: Have times changed? In M. Barreto, M. K. Ryan, & M. T. Schmitt (Eds.), The glass ceiling in the 21st century: Understanding barriers to gender equality (pp. 21-47). Washington, DC: American Psychological Association Books.

Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878-902. https://doi.org/10.1037/0022-3514.82.6.878

Frisby, C. L., & Kim, S.-K. (2008). Using Profile Analysis via Multidimensional Scaling (PAMS) to identify core profiles from the WMS-III. Psychological Assessment, 20, 1-9. https://doi.org/10.1037/1040-3590.20.1.1

Geen, R. G. (2001). Human aggression (2nd ed.). Philadelphia, PA: Open University Press.

Gonzalez-Mulé, E., DeGeest, D. S., Kiersch, C. E., & Mount, M. K. (2013). Gender differences in personality predictors of counterproductive behavior. Journal of Managerial Psychology, 28, 333-353. https://doi. org/10.1108/JMP-12-2012-0397

Griffin, R. W., O’Leary, A. M., & Collins, J. (1998). Dysfunctional Work Behaviors in Organizations. In C. L. Cooper & D. M. Rousseau (Eds.), Trends in Organizational Behaviors (pp. 65-82). New York, NY: John Wiley & Sons.

Griskevicius, V., Tybur, J. M., Gangestad, S. W., Perea, E. F., Shapiro, J. R., & Kenrick, D. T. (2009). Aggress to impress: Hostility as an evolved context dependent strategy. Journal of Personality and Social Psychology, 96, 980-994. https://doi.org/10.1037/a0013907

Gruys, M. L., & Sackett, P. R. (2003). Investigating the dimensionality of counterproductive work behavior. International Journal of Selection and Assessment, 11, 30-41. https://doi.org/10.1111/1468-2389.00224

Harris, M. B. (1996). Aggression, gender, and ethnicity. Aggression and Violent Behavior, 1, 123-146. https://doi.org/10.1016/1359- 1789(95)00012-7

Heilman, M. E., & Chen, J. J. (2005). Same behavior, different consequences: reactions to men’s and women’s altruistic citizenship behavior. Journal of Applied Psychology, 90, 431-441. https://doi.org/10.1037/0021- 9010.90.3.431

Hershcovis, M. S., Turner, N., Barling, J., Arnold, K. A., Dupré, K. E., Inness, M., ... Sivanathan, N. (2007). Predicting workplace aggression: a meta- analysis. Journal of Applied Psychology, 92, 228-238. https://doi. org/10.1037/0021-9010.92.1.228

Hollinger, R. C., & Adams, A. (2010). National Retail Security Survey Final Report. Gainsville, FL: University of Florida.

Hollinger, R. C., & Clark, J. P. (1982). Formal and informal social controls of employee deviance. The Sociological Quarterly, 23, 333-343. https:// doi.org/10.1111/j.1533-8525.1982.tb01016.x

Kark, R., Waismel-Manor, R., & Shamir, B. (2012). Does valuing androgyny and femininity lead to a female advantage? The relationship between gender-role, transformational leadership and identification. The Leadership Quarterly, 23, 620-640. https://doi.org/10.1016/j. leaqua.2011.12.012

Kawakami, C., White, J. B., & Langer, E. J. (2000). Mindful and masculine: Freeing women leaders from the constraints of gender roles. Journal of Social Issues, 56, 49-63. https://doi.org/10.1111/0022-4537.00151

Kim, S.-K. (2013). Identifying between-person and within-person factors to enhance understanding observed score profiles. British Journal of Mathematical and Statistical Psychology, 66, 435-451. https://doi. org/10.1111/j.2044-8317.2012.02055.x

Kim, S.-K., Annunziato, R. A., & Olatunji, B. O. (2017). Profile analysis of the effects of treatment on change in eating disorder indicators. International Journal of Methods in Psychiatric Research, 23. First published online. https://doi.org/10.1002/mpr.1599

Kim, S.-K., Davison, M. L., & Frisby, C. L. (2007). Confirmatory factor analysis and profile analysis via multidimensional scaling. Multivariate Behavioral Research, 42, 1-32. https://doi. org/10.1080/00273170701328973

Kim, S. K., Frisby, C. L., & Davison, M. L. (2004). Estimating cognitive profiles using profile analysis via multidimensional scaling (PAMS). Multivariate Behavioral Research, 39, 595-624. https://doi.org/10.1207/ s15327906mbr3904_2

Kogut, D., Langley, T., & O’Neal, E. C. (1992). Gender role masculinity and angry aggression in women. Sex Roles, 26, 355-368. https://doi. org/10.1007/BF00291548

Kruskal, J. B. (1964). Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika, 29, 1-27. https://doi. org/10.1007/BF02289565

Kulas, J. T., McInnerney, J.E., Demuth, R.F., & Jadwinski, V. (2007). Employee satisfaction and theft: Testing climate perceptions as a mediator. The Journal of Psychology, 141, 389-402. https://doi.org/10.3200/ JRLP.141.4.389-402

Lanyon, R. I., & Goodstein, L. D. (2004). Validity and reliability of a pre- employment screening test: the counterproductive behavior index (CBI). Journal of Business and Psychology, 18, 533-553. https://doi. org/10.1023/B:JOBU.0000028450.43358.15

Lee, K., & Allen, N. J. (2002). Organizational citizenship behavior and workplace deviance: The role of affect and cognitions. Journal of Applied Psychology, 87, 131-142. https://doi.org/10.1037/0021- 9010.87.1.131

Levy, T., & Tziner, A. (2011). When destructive deviance in the workplace becomes liability: a decisional behavioral model. Quality and Quantity, 45, 233-239. https://doi.org/10.1007/s11135-009-9277-0

Marshall, N. A., Arnold, D. H., Rolon-Arroyo, B., & Griffith, S. F. (2015). The association between relational aggression and internalizing symptoms: A review and meta-analysis. Journal of Social and Clinical Psychology, 34, 135-160. https://doi.org/10.1521/jscp.2015.34.2.135

Martinko, M. J., Douglas, S. C., & Harvey, P. (2006). Understanding and managing workplace aggression. Organizational Dynamics, 35, 117- 130. https://doi.org/10.1016/j.orgdyn.2006.03.002

McKay, D., Kim, S.-K., Taylor, S., Abramowitz, J. S., Tolin, D. F., Coles, M. E., Timpano, K. R., & Olatunji, B. O. (2014). An examination of obsessive- compulsive symptoms and dimensions using profile analysis via multidimensional scaling (PAMS). Journal of Anxiety Disorders, 28, 352-357. https://doi.org/10.1016/j.janxdis.2014.03.004

Meyers-Levy, J., & Loken, B. (2015). Revisiting gender differences: What we know and what lies ahead. Journal of Consumer Psychology, 25, 129- 149. https://doi.org/10.1016/j.jcps.2014.06.003

Mitchell, M. S., & Ambrose, M. L. (2007). Abusive supervision and workplace deviance and the moderating effects of negative reciprocity beliefs. Journal of Applied Psychology, 92, 1159-1168. https://doi. org/10.1037/0021-9010.92.4.1159

O’Fallon, M. J., & Butterfield, K. D. (2005). A review of the empirical ethical decision-making literature: 1996-2003. Journal of Business Ethics, 59, 375-413. https://doi.org/10.1007/s10551-005-2929-7

O’Leary-Kelly, A. M., Griffin, R. W., & Glew, D. J. (1996). Organization- motivated aggression: A research framework. Academy of Management Review, 21, 225-253. https://doi.org/10.5465/AMR.1996.9602161571

Olatunji, B. O., Kim, S.-K., & Wall, D. (2015). Extracting body image symptom dimensions among eating disorder patients: The Profile Analysis via Multidimensional Scaling (PAMS) approach. Body Image: An International Journal of Research, 15, 16-23. https://doi.org/10.1016/j. bodyim.2015.04.008

Pearson, C. M., Andersson, L. M., & Porath, C. L. (2005). Workplace incivility. In S. Fox & P. E. Spector (Eds.), Counterproductive workplace behavior: Investigations of actors and targets (pp. 177-200). Washington, DC: APA.

Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879-903. https://doi.org/10.1037/0021-9010.88.5.879

55 L. Chernyak-Hai et al. / Journal of Work and Organizational Psychology (2018) 34(1) 46-55

Richardson, D. (2014). Everyday aggression takes many forms. Current Directions in Psychological Science, 23, 220-224. https://doi. org/10.1177/0963721414530143

Robinson, S. L. (2008). Dysfunctional workplace behavior. In J. Barling & C. L. Cooper (Eds.), The Sage handbook of organizational behavior (pp. 141-159). Thousand Oaks, CA: SAGE.

Robinson, S. L., & Bennett, R. J. (1995). A typology of deviant workplace behaviors: a multidimensional scaling study. Academy of Management Journal, 38, 555-572. https://doi.org/10.2307/256693

Sackett, P. R., & DeVore, C. J. (2001). Counterprooductive behaviors at work. In N. Anderson, D. S. Ones, H. K. Sinangil, & C. Viswesvaran (Eds.). Handbook of Industrial, Work, and Organizational Psychology. Vol 1 (pp. 145-164). London, UK: Sage.

Salgado, J. F. (2002). The Big Five personality dimensions and counterproductive behaviors. International. Journal of Selection and Assessment, 10, 117-125. https://doi.org/10.1111/1468-2389.00198

Salgado, J. F., Moscoso, S., & Anderson, N. (2013). Personality and counterproductive work behavior. In N. D. Christiansen & R. P. Tett (Eds.), Handbook of personality at work (pp. 606-632). New York, NY: Routledge.

Santos, A., & Eger, A. (2014). Gender differences and predictors of workplace deviance behavior: the role of job stress, job satisfaction and personality on interpersonal and organizational deviance. International Journal of Management Practice, 7, 19-38. https://doi. org/10.1504/IJMP.2014.060541

Sosinsky, L. S., & Kim, S.-K. (2013). A profile approach to child care quality, quantity, and type of setting: Parent selection of infant child care arrangements. Applied Developmental Science, 17, 39-56. https://doi. org/10.1080/10888691.2013.750196

Spector, P. E., & Fox, S. (2005). A model of counterproductive work behavior. In S. Fox & P. E. Spector (Eds.), Counterproductive workplace behavior: Investigations of actors and targets (pp. 151-174). Washington, DC: APA.

Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh, A., & Kessler, S. (2006). The dimensionality of counterproductivity: Are all counterproductive behaviors created equal? Journal of Vocational Behavior, 68, 446-460. https://doi.org/10.1016/j.jvb.2005.10.005

Tavris, C. (1984). On the wisdom of counting to 10: Personal and social dangers of anger expression. In P. Shaver (Ed.), Review of personality and social psychology: Emotions, relationships, and health (pp.170- 191). Beverly Hills, CA: Sage.

Timberlake, S. (2005). Social capital and gender in the workplace. Journal of Management Development, 24, 34-44. https://doi. org/10.1108/02621710510572335

Vardi, Y., & Weitz, E. (2004). Misbehavior in organizations: Theory, research, and management. Mahwah, NJ: Erlbaum.

Walker, S., Richardson, D. S., & Green, L. R. (2000). Aggression among older adults: The relationship of interaction networks and gender role to direct and indirect responses. Aggressive Behavior, 26, 145- 154. https://doi.org/10.1002/(SICI)1098-2337(2000)26:2<145::AID- AB1>3.0.CO;2-Q

Warren P., Richardson D. S., & McQuillin, S. (2011). Distinguishing among non-direct forms of aggression. Aggressive Behavior, 37, 291-301. https://doi.org/10.1002/ab.20394

Zhang, C., Mayer, D. M., & Hwang, E. (2017). More is less: Learning but not relaxing buffers deviance under job stressors.‏ Journal of Applied Psychology, 103, 123-136. https://doi.org/10.1037/apl0000264

Copyright of Revista de Psicologia del Trabajo y de Las Organizaciones is the property of Colegio Oficial de Psicologos de Madrid and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.