Week 4 - Assignment: Assess Distributive Policy Administration at the Agency Level and Week 5 - Assignment: Evaluate the Political Backlash of Failed Policies

ILRReview, 67(Supplement) 2014. © by Cornell University. Print 0019- 7939/Online 2162- 271X/00/6703 $05.00

THE DEPARTMENT OF LABOR AT THE

INTERSECTION OF RESEARCH AND POLICY

ALAN B. KRUEGER*

I want to begin by thanking Secretary Solis for her service as secretary of labor during a time when American workers are recovering from the worst recession since the Great Depression. Secretary Solis’s commitment to unemployment insurance, job training, green jobs, and enforcement of reg- ulations has helped the economy to recover, and will provide for a stronger foundation for economic growth going forward.

My first experience in government was when I served as the chief econo- mist for the Department of Labor (DOL) when Bob Reich was secretary of labor in 1994–95. I recall the close working relationship that Bob had with Laura Tyson, then Council of Economic Advisers (CEA) chair, and I am pleased that we have been able to continue that tradition. Moreover, my time working at the labor department made it much easier for me to be able to make a transition to serve as chief economist of the Department of the Treasury in the midst of the financial crisis in February 2009, as well as in my current job. As a professor before joining DOL, I had never heard about Exec Sec (then headed by Michael Kerr) or the other essential administra- tive processes that enable a major government department to function.

I want to use my time today to discuss the role that the Labor Department has played in developing and using research to inform public policy. My experience at DOL was part of a long line in this tradition.

Channeling Research into Policy

Since its creation, the labor department has played a critical role, often at the forefront of the government, in using research to inform policy mak- ing, and, equally important, in developing policy- related research. It has often been a leader in the government in investing in research—and, just as

*Alan B. Krueger is Bendheim Professor of Economics and Public Affairs, Princeton University. He was chairman of the President’s Council of Economic Advisers when these remarks were presented, and he served as Chief Economist of the Department of Labor in 1994–95.

DOL AT THE INTERSECTION OF RESEARCH AND POLICY 585

important, in developing the institutional and human infrastructure to con- duct research—as well as in creating an institutional structure to incorpo- rate research into public policy making. This history has served the government and the research community well.

As an economist, I think that DOL has helped to overcome two important market failures when it comes to the production of actionable, policy- related research. First, government agencies and the public benefit from policy- related research on topics such as the benefits and costs of job training or the most effective strategies for enforcing workplace safety regulation, but researchers have little private incentive to conduct such research, especially in fields that highly prize theoretical work over applied policy research. Sec- ond, academic researchers do not always understand the research needs or priorities of policy offices. Over the years, DOL has overcome these market failures by setting up internal and external institutions and by building re- search capacity.

Developing Institutional Capacity

Three offices within DOL have been central in the department’s efforts to develop policy- related research and in the department’s ability to incorpo- rate research findings in policymaking and program implementation: The Office of the Assistant Secretary for Policy Evaluation and Research (ASPER, now ASP), the Chief Economist’s (CHECO) Office, and the Chief Eval- uation Office (CEO). The Bureau of Labor Statistics (BLS), of course, has long played a critical role in providing unimpeachable data for policymak- ing across the government, but I will focus on the roles of ASPER, CHECO and CEO, which have received less attention.

ASPER was founded in 1972, when James Hodgson was secretary of labor. His goal was for ASPER “to focus on general economic policy, jobs and train- ing programs, income maintenance efforts, health and disability issues, and labor- management concerns.” These are still central issues today. ASPER pioneered research and program evaluation methods during the Nixon and Ford administrations. In the Carter years, ASPER’s focus moved to address- ing problems of inflation, unemployment, and low productivity growth. During the Reagan years, ASPER became ASP.

In preparing my remarks, I asked Dan Hamermesh, former director of research at ASPER, to recall his time at ASPER. He e- mailed the following response and granted permission to reproduce it here:

First, Orley Ashenfelter was in charge of Evaluation (the E in ASPER) at ASPER 1972–73; George Johnson did it 1973–74, and Ernie Stromsdorfer did it 1974–77. A career bureaucrat, non- economist, ran Research (the R in ASPER) for 2 years, then I was asked to be the first outsider, did it 1974–75 (thus getting to be in D.C. at the denouement of Watergate, and having the pleasure of remov- ing the Nixon picture from my wall on August 9, 1974). Frank Stafford took over from me, did 1975–76, and Al Gustman did 1976–77. Under Carter, they re- organized, and there were no academic temporaries in similar positions until the Clinton Administration.

586 ILRREVIEW

For me, ASPER was a career- changing experience. My dissertation (finished Fall 1968) was on labor demand, but I hadn’t done anything on it thereafter. During the recession of the time I was asked by Paul O’Neill, then Deputy Bud- get Director, later Secretary of the Treasury under W, about justifications in terms of labor- demand elasticities for ideas about wage subsidies and their ef- fects. That resulted in my survey papers, JHR 1976, 1979, my 1986 Handbook paper, and ultimately my 1993 magnum opus, Labor Demand (in the Preface of which I thank Paul O’Neill).

I believe Orley’s work on the “Ashenfelter dip” was stimulated by questions that came up while he was at ASPER; and I know George Johnson’s interests were also re- directed. Same for Al Gustman—he hadn’t done anything on retirement before then, but it has been most of his career since then. Looking at the records of the 5 of us, I would say that only Frank Stafford’s career was not greatly af- fected by our time in ASPER.

I also asked Dan about a story I heard from Orley Ashenfelter about his use of unemployment insurance data to predict the unemployment rate, and being very successful in that effort. Here is Dan’s response:

In terms of your question about predicting the unemployment rate (CPS rate), the correct story is as follows: I had developed an equation relating the monthly insured unemployment rate (X) to the CPS rate (Y). Since the IUR came out a week ahead of the CPS rate, this allowed me to get a prediction of the CPS rate. Apparently the then- Secretary of Labor, John Dunlop, heard about it, probably from my Assistant Secretary, and got involved in a betting pool [on the unemployment rate] with the then Secretary of Treasury (William Simon), the Director of what is now OMB, and, I think, a few others. So for my last 4 months there (through August 1975), Dunlop would ask me what my model predicted. Indeed, I was lying on the couch one Saturday morning in our rented house, and my wife answered the phone to hear this gruff voice asking for me. Dr. John had one question, of course, which I answered. Apparently my forecasts were pretty good, because he won the pool more often than the other bettors!!!

Anyway, an absolutely delightful question, great reminiscing for this old fogey.

We still use UI data to predict the unemployment rate, but instead of the IUR, initial UI claims have proved to be the best real- time, high- frequency predictor of the unemployment rate and job gains.

Developing Talent

Even before ASPER was started, the labor department was actively involved in recruiting and developing the human capital necessary to conduct re- search on labor markets and labor- management relations. Under Howard Rosen’s leadership, the Manpower Development and Training Administra- tion (MDTA), the forerunner of ETA, founded a highly successful disserta- tion grant program. From 1966 through the 1970s, the department gave out more than 300 dissertation fellowships. The average grant in 1972 was $10,000; in today’s dollars that is $55,000. Students were required to work on policy- related topics and to present their findings in person at DOL. Half of the grants went to economics PhD students, about 20% went to sociology students, and the remaining 30% were spread over scholars in industrial and

DOL AT THE INTERSECTION OF RESEARCH AND POLICY 587

labor relations, business administration, education, psychology, and politi- cal science.

Table 1 reports several of the recipients of these grants. Many of these scholars are influential in academia and policy. Bob Groves became the di- rector of the Census Bureau and is now the provost at Georgetown. Richard Freeman and Jim Heckman need no description. Other notable recipients include Fran Blau, Mike Boskin, Barry Hirsch, Marjorie Honig, Larry Kahn, Harry Katz, Tom MaCurdy, Jim Medoff, Robert Moffitt, Paul Osterman, Harvey Rosen, Jim Smith, and Kip Viscusi. These scholars and many other recipients of the doctoral dissertation grant program have made major con- tributions to policy- related research and profoundly influenced their fields.

While I have not conducted a rigorous cost- benefit analysis of the disser- tation program, I think it is a safe bet that taxpayers received a good return from this investment. When I served as the chief economist at the labor de- partment, Mary Eccles and I tried to resuscitate this program and made a fair amount of progress. Despite getting commitments from agencies at DOL to use some of their budget to restart the dissertation program, we were unable to get it started because Secretary Reich feared that it would look like pork to Congressional appropriators and be at risk of being cut. But I do think that the dissertation grant program served as a model for the government, and one which will likely be followed by other fields, such as international development.

Institutional Capacity

The department also helped develop institutional research capacity outside of DOL. A good example is the founding of MDRC (Manpower Demonstra- tion Research Corporation). MDRC is a nonprofit research organization that has conducted invaluable research on job training, job search assis- tance, and welfare reform, among other topics. MDRC was founded in 1974 by the Ford Foundation and a consortium of federal departments, led by

Table 1. Sample of Alumni of Doctoral Dissertation Grant Program

Steven Allen Linda Datcher Joseph Hotz Craig Olson Thomas Barocci Gregory DeFreitas H. Allan Hunt Paul Osterman John Barron Thomas DePrete Michael Hurd Anthony Pellechio Andrea Beller Mary Eccles William Johnson Jeffrey Perloff Suzanne Bianchi Ronald Ehrenberg Lawrence Kahn Robert Reischauer Francine Blau David Featherman Harry Katz Myron Roomkin Richard Block Randall Filer Morris Kleiner Harvey Rosen Barry Bluestone Robert Flanagan Leonard Kreider Laurence Seidman Michael Boskin Richard Freeman Robert Lerman James Smith Steven Braun John Geweke Robert Lucas Charlotte Stiglitz Richard Butler Robert Groves Thomas MaCurdy Nancy Tuma George Cave Bennett Harrison Janice Madden Kip Viscusi Andrew Cherlin James Heckman James Medoff Michael Wachter Kim Clark Barry Hirsch Wesley Mellow Halbert White Sheldon Danziger Marjorie Honig Robert Moffitt Kenneth Wolpin

588 ILRREVIEW

DOL, and that also included the Department of Health, Education, and Welfare; the Department of Justice; and the Department of Housing and Urban Development.

MDRC conducted the first large- scale randomized evaluation of an em- ployment and training program, the National Supported Work Demonstra- tion, which played a central role in research in labor economics and was used, for example, by Robert LaLonde in his dissertation years later. To this day, DOL sponsors MDRC research on critical topics such as youth transi- tions to work and improving employment stability for low- wage workers.

In 1993, Bob Reich created a new office at DOL, the Chief Economist’s Office. Reich’s motivation was twofold: he wanted a top economist who would have cachet with the National Economic Council, another organiza- tion that he proposed, and he wanted to give academic economists an op- portunity to work in the labor department, which he thought would enrich their research when they returned to academia.

This tradition continues today. Larry Katz was the first chief economist, and Secretary Reich could not have found anyone better to interact with the National Economic Council and to translate economic research that was rel- evant to the department. One of Larry’s major achievements was during the Reemployment Act, when he was able to apply evaluation results to incorpo- rate job search assistance in One Stop Job Centers, which were also part of the reemployment initiative under Secretary Reich. Perhaps most important, he was able to persuade the Congressional Budget Office that job search as- sistance would pay for itself, because even though there was an initial layout of money, those who went through the program would find employment more quickly and hence require less support from unemployment insurance benefits. I put the finishing touches on an outstanding report that Larry had started called “What’s Working and What’s Not in Employment and Training.”

Let me lastly turn to the Chief Evaluation Officer’s position. This is a new office that was created under Secretary Hilda Solis’s leadership. Alex Mas and the Office of Management and Budget provided critical support to es- tablish this office, and it has become a gold standard for other departments. The Chief Evaluation Officer’s role is to support evaluation research within various departments and to help the department understand findings from these evaluations.

At the labor department, one of the evaluation projects underway looks at a very important part of the American Recovery and Reinvestment Act of 2009: a subsidy to help workers pay for health insurance under COBRA if they lost their jobs during the recession. This was a generous subsidy of about 65%.

The Office of Management and Budget has been building on this exam- ple and acting director Jeffrey Zients issued a memorandum advising de- partments to create a position like the Chief Evaluation Officer or another high- rank position that incorporates the same functions. Key elements of OMB’s initiative to strengthen agency evaluation capacity are described in the part of Zients’s 2012 memo reproduced below:

DOL AT THE INTERSECTION OF RESEARCH AND POLICY 589

Memorandum to Heads of Executive Departments and Agencies

Office of Management and Budget, May 18, 2012

Strengthening agency evaluation capacity.

Agencies should have a high- level official who is responsible for program evaluation and can: Develop and manage the agency’s research agenda; Conduct or oversee rigorous and objective studies; Provide independent input to agency policymakers on resource allocation and to program leaders on program man- agement; Attract and retain talented staff and researchers, including through flexible hiring authorities such as the Intergovernmental Personnel Act; and Re- fine program performance measures, in collaboration with program managers and the Performance Improvement Officer.

These goals can be accomplished by different kinds of leaders, ranging from a chief evaluation officer who reports to the Secretary or Deputy Secretary to the head of an independent institute in the agency. An existing official could play the role, or a forceful new position could replace several less empowered ones. OMB invites agencies to propose in their budget submissions ways to strengthen the agency’s evaluation capacity, within tight resource constraints.

Support for Evidence- Based Initiatives

OMB invites your agency to participate in a number of forums to improve use of evidence: OMB and the Council of Economic Advisers will organize a series of topical discussions with senior policy officials and research experts in the agen- cies. The meeting agendas will focus on administrative and policy levers for driv- ing an increasing share of Federal investments into evidence- based practices. We will plan summer meetings in order to help inform agencies’ evaluation plans and budget submissions, and will also have follow- up meetings in the fall.

The Council of Economic Advisers has collaborated with OMB to work with the departments to emphasize the importance of evaluating govern- ment programs and to develop evaluation plans.

I don’t want to leave you with the impression that evaluation is easy or that it fits in naturally in government, however. There are often conflicts of interest. Government agencies have a tendency to want to control the flow of information, especially if the information can be used to criticize their programs. In the George W. Bush years, a first- rate Mathematica Policy Re- search study of the Job Corps program that cost millions of dollars to con- duct was suppressed for several years. But safeguards can be in put place to limit conflicts of interest and to maximize the value of government sup- ported policy research.

Challenges for the Research Community

I benefit from policy- related research in my current role. In my job as chair- man of CEA, I am more a consumer of research than a producer of re- search. As such, I would like to highlight a set of research questions that would, in my view, help the council to develop and implement effective pol- icy and to more effectively advise the president.

Broadly speaking, two types of research influence public policy. One in- volves developing models of how markets and the macroeconomy work and

590 ILRREVIEW

estimating the key parameters underlying those models; the other evaluates specific policies, such as COBRA subsidies.

An example of the former is the labor supply, or the job search, model. This research takes years to develop and is typically not specific to a particu- lar policy issue. The labor supply elasticity could therefore be applied to tax policy or to many other issues. In fact, the Congressional Budget Office has used econometric estimates of labor supply elasticities to forecast the effects of tax policy. Of course, researchers need to use the correct models and have the relevant parameters as well.

Another aspect of this type of work which I appreciate much more in my current role than I did previously is that it helps with economic forecasting. A critical part of developing a budget and doing economic planning is hav- ing a forecast. The Council of Economic Advisers, together with OMB and Treasury, form the Troika that develops the forecasts underlying the budget. Their forecast plays a very important role in economic policymaking.

The second type of research is specific to a particular policy. An example is research on how to best structure tiers for extended unemployment insur- ance benefits to target geographic areas. Another example involves studying how the incentives of training providers affect their behavior and partici- pant outcomes. This form of study requires researchers to delve into consid- erable detail; so much so, in fact, that they often may miss broader trends and policy implications. Such work is not necessarily well- received in aca- demic circles, yet it is critical for designing and implementing effective eco- nomic policies.

There are many research questions in both of the aforementioned cate- gories that are particularly important to address in the current economic environment. Let me take advantage of this distinguished audience to raise seven questions on which I believe research could be especially valuable for policymakers.

The first question has to do with labor force participation. The labor force participation rate peaked in 2000 and has declined since then (see Figure 1). This is a critical input for Troika forecasts, especially when consid- ering the effect of labor force participation on the Social Security program

1. Why did the labor force participation rate peak in 2000, and what does it imply for the future?

2. Trends in labor’s share of national income? 3. Can we extract even more useful data from weekly UI claims? 4. Which strategies help to improve the transition from school to work? 5. What are the best methods of targeting regulatory enforcement? 6. What is the best way of structuring incentives for job training providers? 7. When does the economy get a reprieve from Okun’s law?

Box 1. Under-Researched Economic Policy Questions

DOL AT THE INTERSECTION OF RESEARCH AND POLICY 591

in the near future. If researchers and policymakers understand why labor force participation peaked, they can have a better chance of modeling fu- ture outcomes.

A second research topic is the labor share of income, shown in Figure 2. This is a very old topic, one which received considerable attention from Nicholas Kaldor and Bob Solow in the 1950s.1 Labor economists are gener- ally no longer interested in this question because they can actually look at microdata and can examine incomes of different categories of people. They can look at well- educated or poorly educated people, or people at various quantiles of the income distribution. For budgetary purposes, however, knowing the share of national income, both wages and salaries and benefits, is extremely important, because labor is taxed at a different rate than capi- tal. Having a sense of whether labor share will continue to decline or whether it will regress back toward its mean is crucial for projecting out the government budget years from now.

The third question is whether researchers can do a better job of predict- ing the unemployment rate from unemployment insurance claims. As I mentioned, the four- week moving average of initial claims is a better predic- tor than the insured unemployment rate, which Dan Hamermesh used dur- ing his tenure as director of research in the 1970s. One particularly striking fact is that over the last 40 years, the information that the unemployment insurance system is generating, although extremely valuable, has not changed very much. Given advances in data processing and computer tech- nology, we would expect to be able to learn more about the industries that experience a high level of layoffs, the demographics of the workers, the

1 See Kaldor (1955–56) and Solow (1958).

Figure 1. Labor Force Participation Rate, 1948–Present

Source: BLS

592 ILRREVIEW

education level of the workers, etc. Given how valuable the unemployment insurance claims are for predicting the current state of the economy—it is our best high- frequency indicator—it is certainly worth asking the question, “Can we extract more information from UI claims?” (In this regard, I am pleased that DOL recently has launched an initiative to report initial claims by industry and demographic group.)

Fourth, which strategies help to improve the transition from school to work? A long- standing problem in the U.S. jobs market is that the transition from school to work is very messy and seems to be extremely inefficient. Can we help to direct school leavers more quickly and effectively to job matches?

Fifth, which methods of targeting regulatory enforcement work best? This is an issue not only at the DOL, but also in financial regulation. In fact, in some respects it is an even bigger challenge in the financial regulations because there are billions of transactions every day. This creates an enor- mous amount of data to sift through. Nonetheless, we should be able to use modern methods from computer science and statistics to more effectively apply our limited regulatory dollars to accomplish our goals.

Number six, what is the best way to structure incentives for job- training providers? This applies not only to DOL programs, but also to the federal government’s expanding role in financing community college education. How should we structure funds for community colleges? How do we make sure that we do not have unintended consequences, such as skimming of the best workers who would do well even in absence of the training?

Finally, number seven. I couldn’t resist including something on Okun’s law, which we usually learn when we take introductory macroeconomics and then don’t think about much later on. It turns out that forecasting at CEA has relied on Okun’s law as a guide for years. Okun’s law doesn’t hold every

Figure 2. Labor Share of National Income, 1948–Present

Source: BLS

DOL AT THE INTERSECTION OF RESEARCH AND POLICY 593

year, however (Figure 3). When is it that we get deviations from Okun’s law, from the relationship between GDP growth and unemployment? That is a very old topic in economics yet one which I think will benefit from renewed attention.

I don’t mean to imply that these seven questions are the most important questions for research in economic policy, but I do think that these are questions on which further research would yield a high marginal benefit for society And I can think of no better audience, nor a more appropriate occa- sion than this one, to highlight these issues and to challenge the audience to conduct further research.

References

Kaldor, Nicholas. 1955–56. Alternative theories of distribution. Review of Economic Studies 23(2): 83–100.

Solow, Robert M. 1958. A skeptical note on the constancy of relative shares. American Economic Review 48(4): 618–31.

Zients, Jeffrey D. 2012. Memorandum to the heads of executive departments and agencies. Office of Management and Budget Memorandum M- 12- 14, May 18. Accessed at http:// www.whitehouse.gov/sites/default/files/omb/memoranda/2012/m- 12- 14.pdf.

Figure 3. Okun’s Law Relationship between Unemployment and GDP Growth

Source: BLS and BEA

F o

u r

Q u

a rt

e r

C h

a n

g e

in U

n e

m p

lo ym

e n

t R

a te

Four Quarter Percent Change in GDP

Copyright of ILR Review is the property of Sage Publications 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.