Application 3 : Annotated Bibliography

TheQualityofIndustrialRelationsandUnemploymentinDevelopingCountries.pdf

Home >Information Systems homework help >Application 3 : Annotated Bibliography

The Quality of Industrial Relations and Unemployment in Developing Countries

Horst Feldmann*

Abstract Using data from 45 countries for six years in the period 1995 to 2003, the author analyses whether the quality of industrial relations affects unemployment in developing countries. To measure the quality of industrial relations, the author uses the results of surveys in which senior business executives characterized the industrial relations of their countries of residence. According to the regression results, cooperative (confron- tational) industrial relations are likely to lower (increase) unemployment. In general, the magnitude of the effect appears to be moderate. It seems to be small among women. The results are robust to variations in specification.

1. Introduction

Cooperative relations between employers and employees, and between their associa- tions, have numerous advantages over confrontational industrial relations (Freeman and Medoff, 1984). For example, if industrial relations at an enterprise are of a coop- erative nature, there is less labor turnover. Consequently, such an enterprise not only has to look for and train new personnel less often, it is also more inclined to invest in the human capital of its employees. The employees themselves are also more willing to invest in their own human capital, in particular the company-specific type, if a greater length of service is to be expected. Furthermore, employees are more likely to make proposals for the improvement of working methods and production techniques if industrial relations are cooperative. Being on-site, employees are especially well- positioned to do so as they frequently have more detailed knowledge of the relevant critical points than their employer. In addition, if industrial relations at an enterprise are cooperative, staff members are more likely to make concessions in the case of an adverse development of the enterprise to help overcome the critical situation (e.g. temporary wage reductions). Similarly, in collective bargaining trade unions are prob- ably more willing to take into account the difficult economic environment enterprises may face, resulting from, for example, a recession, intense competitive pressure from foreign companies, or necessary structural adjustments in the industry. Cooperative industrial relations are thus likely to lead to general wage moderation in difficult times. Because of all these advantages, they may be expected to result in a comparatively lower level of unemployment, for example via more labor-intensive modes of produc- tion, higher domestic investment, and/or higher foreign direct investment inflows.

Recently, several studies have empirically analyzed the effect of industrial relations quality on unemployment. Blanchard and Philippon (2004) and Feldmann (2006a) use data from a cross section of industrial countries. Whereas Feldmann’s data for his industrial relations quality variable come from the period 1985 to 2002, Blanchard and

* Feldmann: Department of Economics & International Development, University of Bath, Bath BA2 7AY, United Kingdom. Tel: +44-1225-386853; Fax: +44-1225-383423; E-mail: [email protected].

Review of Development Economics, 13(1), 56–69, 2009 DOI:10.1111/j.1467-9361.2008.00459.x

mailto:[email protected]

Philippon’s come from the year 1999. Feldmann (2003) uses data from the years 2000 and 2001 for the OECD countries and those countries from central and eastern Europe that were EU accession countries at that time. In a more recent paper, he uses data from 12 transition countries for the years 1996 to 2001 (Feldmann, 2005a). All of these studies find that more cooperative industrial relations are likely to lower unemployment.

Empirical studies on the determinants of foreign direct investment decisions are instructive too. For example, in surveys of US and UK multinational companies, man- agers viewed industrial relations as one of the most important considerations in deciding where to invest (Karnani and Talbot, 1992; Marginson et al., 1995). Similarly, Cooke (1997) and Cooke and Noble (1998) find that the amount of US foreign direct investment a country receives is significantly affected by the characteristics of its industrial relations system.

Given the important role foreign direct investment inflows play in many developing countries, the quality of their industrial relations may have even larger labor market effects than in industrial countries. On the other hand, one may argue that unemploy- ment in developing countries is likely to be determined primarily by the level of economic development and by geographical and demographic conditions. Given the ambiguity of these considerations, the impact of industrial relations quality on unem- ployment in developing countries has to be resolved empirically. This paper is the first to analyze this issue. Section 2 describes the data and the empirical strategy. Section 3 presents and discusses the regression results. Section 4 concludes.

2. Data

To measure the quality of industrial relations, we use the results of surveys among senior business executives. The results come from the World Economic Forum’s Execu- tive Opinion Surveys (EOS), which are carried out annually in a large number of countries to determine the international competitiveness of the relevant economies. The respondents are a company’s CEO or a member of its senior management. In each country approximately 66 managers are interviewed. The industry structure of the companies questioned corresponds largely to the industry structure of the relevant economy (excluding the agricultural sector). Also, care is taken to question companies of various size categories and types (e.g. private and state-owned, domestically oriented and internationally active enterprises).1

The typical EOS question asks participants to indicate on a numerical scale which of the two statements specified in each case they favor. After the interviews, arithmetic means for each question are calculated from the answers for each country. The Box contains the questions on industrial relations quality used in the following analysis. We use the answers from the years 1995 and 2000 to 2002. For two reasons, we do not use survey results from more recent years. First, unemployment data compiled according to a uniform methodology are unavailable for the years after 2003 (see below). Second, for the reasons given below, our industrial relations quality variable was lagged by one year. Furthermore, we neither use survey results from years prior to 1995 nor from 1996 to 1999 mainly because data for some control variables are lacking for those years (see below).2 The World Economic Forum used different scales in the years before 1997. For the purpose of our analysis, we converted the 1995 answers to the 1-to-7 scale used in the more recent surveys.3 The industrial relations questions are phrased in a similar way (see Box). There are some slight variations but these are only refinements of style to

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 57

make the questions more precise. Thus the answers to all questions can be used simultaneously.

Questions on Industrial Relations Quality from the Executive Opinion Surveys (EOS)

• EOS 1995: “Relations between managers and employees are: (0 = generally fragile, 10 = generally productive)”

• EOS 2000: “Management/worker relations are generally cooperative. (1 = strongly disagree; 7 = strongly agree)”

• EOS 2001 and 2002: “Labor-employer relations in your country are: (1 = gener- ally confrontational; 7 = generally cooperative)”

Source: World Economic Forum (2000, 2002, 2003), World Economic Forum and Institute for Management Development (1995).

There are two main reasons to assume that the answers to the Executive Opinion Surveys correctly reflect the quality of industrial relations:

• Not only is the selection of respondents largely representative; what is even more important is the fact that the respondents have comprehensive knowledge of and practical experience with the industrial relations of their countries.

• The questions are phrased objectively and, at the same time, permit a better coverage of the various facets of industrial relations quality than conceivable “objective” indicators. For example, legal standards stipulate the relations between management and labor only to a small extent. Even if they could be translated into a numerical scale without difficulty, these “objective” data would represent only a few aspects of the industrial relations. Informal standards and traditions are important as well. Therefore, the answers to the EOS questions appear to be better suited to capture the quality of the relations between employers and employees and their associations.

Of course, potential drawbacks also have to be considered in connection with the use of the Executive Opinion Surveys. One potential drawback is that each respondent could use his own yardstick when answering the questions. For example, on the 1-to-7 scale, an item marked 7 by one person may be marked only 5 by another. However, in the planning, implementation, and analysis of the surveys, care was taken to ensure the use of a uniform yardstick. For one, the respondents were provided with a written explanation of the answering scale. Also, the answers were examined for robustness and consistency using various methods. In one of these checks, half of the responses from each country were randomly dropped from the sample. As the national EOS scores remained stable in the process, they have obviously not been distorted by individual peculiarities in responding (Blanke et al., 2003).

Another cause of concern is that there may be a systematic bias among respondents at the national level. For example, respondents in a country might have a similar pessimistic assessment of the quality of industrial relations if this topic had recently been discussed extensively and with a negative flavor in the press.Also, the questions may be interpreted differently in different countries. For these reasons, the survey results may not accurately reflect objective differences in the national industrial relations climate. The authors of the Executive Opinion Surveys tried to avoid this problem by providing all respondents with a written explanation of the answering scale and by asking them to think in world

58 Horst Feldmann

terms rather than in national terms. In order to check whether a “perception bias” exists, we compare the national EOS scores with related hard data. As an objective indicator, we use log severity rate of strikes and lockouts, that is the natural logarithm of annual number of days not worked due to strikes and lockouts per 1000 employees. As strike activity has traditionally been very high in some countries and very low in others, we use the natural logarithm of the severity rate. This diminishes the impact of outliers. Data both for the severity rate and from the EOS are available for a subgroup of 28 developing countries.4 Using averages over the years 1995 and 2000 to 2002, the correlation coefficient between log severity rate and national EOS scores is –0.32 (significant at the 1% level). Thus with an increase in strike and lockout activity, fewer people think that industrial relations are productive or cooperative.Although the correlation coefficient is not very high, the survey data are supported by the objective data.

In our regression analysis, the country group consists of 45 developing countries (see Appendix). Developing countries are defined as low and middle income countries according to World Bank classification. Greece, Slovenia, and South Korea have suc- ceeded in becoming high income countries in recent years. However, as they were below the relevant income threshold for some of the years and only slightly above it afterwards, they are included in this paper’s group of countries for the whole estimation period.5

To measure the effects on unemployment, we use not only the overall unemployment rate but also unemployment rates relating to female and young workers. This enables us to determine not only whether the quality of industrial relations affects the overall level of unemployment. It also enables us to analyze to what extent it affects two important demographic groups that usually have above-average unemployment rates.

Almost all of the unemployment data used in this paper come from the latest edition of the ILO’s (2005) Key Indicators of the Labour Market, which includes data up to 2003.6

All variables are exclusively based on labor force surveys. Thus the data do not refer to registered unemployment. Instead they are based on an international standard which defines the unemployed as all persons above a specific age who, during the reference period, were without work, currently available for work, and seeking work. Although national coverage of unemployment can vary with regard to factors such as age limits and criteria for seeking work, in the latest edition of its Key Indicators of the Labour Market the ILO has undertaken great efforts to produce series that are comparable across countries. With regard to age limits, for example, most national series presented in this publication refer to the age group 15 years and older. Furthermore, for the latest edition the ILO has “cleaned” the national time series to eliminate breaks in series.Thus the data are comparable over time. Although the ILO’s unemployment data are not completely harmonized across countries, they are harmonized to a large extent.

We control for the impact of labor market and business regulations by using the ratings for the respective components from the Economic Freedom of the World (EFW) index (Gwartney and Lawson, 2005).7 Several recent studies have found both stricter labor market regulation and stricter business regulation to affect adversely labor market performance.8 The EFW component “labor market regulations” covers five types of regulation: statutory minimum wage, hiring and firing regulations, centraliza- tion of collective bargaining, unemployment benefits, and military conscription. The EFW component “business regulations” consists of five indicators: price controls, administrative conditions for starting a new business, the time senior management needs to spend dealing with government bureaucracy, ease of starting a new business, and irregular payments.9 The rating scale of the EFW index ranges from 0 to 10.10 As higher marks on the scale indicate more flexible regulation, we label these controls “flexible labor market regulations” and “flexible business regulations.”

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 59

We control for the impact of the tax burden by using the EFW indicator “top marginal tax rate.” It is based on the top marginal income and payroll tax rate and on the income threshold at which the top marginal income tax rate applies. As higher values on the scale represent lower marginal tax rates and higher income thresholds, we label this variable “low top marginal tax rate.” Previous studies have found that a heavy tax burden is likely to lower employment and to increase unemployment in industrial countries (Prescott, 2004; Feldmann, 2006b).11

The EFW index has been calculated for every fifth year from 1970, and annually since 2000. The component “business regulations” was added to the index only in 1995. For these reasons we are unable to use the EOS data on industrial relations quality from years prior to 1995 and from the period 1996 to 1999.

To control for business cycle conditions, we normalized each country’s GDP growth rate for its average growth rate over the previous ten years and divided the values by ten. We label this variable “GDP growth gap.”12 In addition, we use year dummies to control for year-specific effects. Furthermore, we use GDP per capita to account for the effects of differences in the level of economic development.13 This is particularly impor- tant with respect to the generosity of the social safety net. Social benefits usually become more generous as GDP per capita rises so that workers can more easily afford to be unemployed. The variable “GDP per capita” controls for the possible impact this may have on unemployment as it proxies for the generosity of the social safety net (as well as for other factors related to the level of economic development). Additionally, we use a dummy variable for those countries that are in transition from planned to market economy because this process has a major impact on their labor markets.

Furthermore, we use two variables to control for the impact of geographical condi- tions. In a series of papers, Sachs and co-authors have demonstrated that levels and growth rates of GDP per capita are strongly correlated with key geographical variables (Gallup et al., 1999; Sachs, 2001). Our first geographical control is the share of land area in geographical tropics (Center for International Development, 1999, 2001). Sachs and co-authors argue that tropical climates hinder production and development. One may thus hypothesize that they may also increase unemployment. Our second geographical control is the mean distance to the nearest ice-free coastline, measured in thousands of kilometers (Center for International Development, 2001). A long distance is likely to increase transport costs for international trade, thus possibly increasing unemployment.

We also control for ethnic fractionalization.14 In ethnically heterogeneous societies, the group that comes to power may implement policies that lower output and increase unemployment. Previous studies have found ethnic fractionalization to be inversely related to GDP per capita growth (Easterly and Levine, 1997; Alesina et al., 2003). We hypothesize that ethnic fractionalization is likely to increase unemployment. For example, members of ethnic minorities may be discriminated against when it comes to hiring and firing. Furthermore, the ruling group may implement policies that discrimi- nate against industries dominated by the losing groups. They may also pursue policies, such as financial repression and overvalued exchange rates, that create rents for them- selves at the expense of society at large. All of this is likely to increase unemployment. Finally, we employ a dummy variable for interstate and internal wars because they may severely disrupt the labor markets of the countries in which they take place (Centre for the Study of Civil War, 2005).

The variables “industrial relations quality,” “flexible labor market regulations,” “flex- ible business regulations,” and “low top marginal tax rate” were lagged by one year to allow for slow adjustment and to avoid simultaneity problems. Changes in the quality

60 Horst Feldmann

of industrial relations are likely to affect unemployment only after some time. The same can be expected from changes in labor market and business regulations and tax policies.

The regression coefficients are estimated using the random effects, feasible general- ized least-squares (FGLS) procedure that incorporates time-invariant country effects (Swamy-Arora method). This enables us to exploit both the cross-country and the time-series variation included in the sample while simultaneously controlling for un- observed country effects. Allowing for cross-country differences in labor market performance that reflect the influence of omitted variables is highly desirable, but the random effects method for doing so produces biased estimates if variables included as controls are correlated with country-specific error terms. Therefore, a Hausman test for misspecification of the random effects model is shown for each regression. As the results from this test indicate, none of our estimates is biased (Tables 1 to 3). Thus in our case the random effects FGLS model is the appropriate choice. Finally, to correct for heteroskedasticity, we estimate robust t-statistics using the technique developed by White.

3. Results

Before we discuss the results from the multivariate regressions, let us briefly take a look at the bivariate associations between our industrial relations quality variable, on the one hand, and the unemployment rates, on the other hand (Figures 1 to 3).15 Using country averages, the figures indicate that countries with more cooperative (confron- tational) industrial relations generally have lower (higher) unemployment rates, both among the total labor force as well as among women and youths. Of course, these bivariate associations do not prove that there is a causal impact of industrial relations quality on unemployment. Therefore, the next step is to see whether there is a statis- tically significant effect once we control for the impact of labor market regulations and the other factors mentioned in the previous section.

Tables 1 to 3 present our multivariate regressions. In each table, column 1 reports our basic specification. According to these results, more cooperative industrial relations appear to lower unemployment among the total labor force as well as among female and young workers. Our results thus corroborate the hypothesis mentioned in section 1. They are also in line with previous empirical studies.

However, in contrast to the results for industrial countries (Feldmann, 2006a), cooperative industrial relations in developing countries appear to have only a modest effect on unemployment. For example, take Uruguay and Costa Rica. According to the EOS, Uruguay had some of the most confrontational industrial relations in our sample of 45 developing countries. On average over the years 1995 and 2000 to 2002, its score was 3.6. By contrast, Costa Rica had the most cooperative industrial rela- tions. Its score averaged 5.5. Costa Rica also had substantially lower unemployment rates, both among the total labor force and among the two demographic groups. According to our estimates, if industrial relations in Uruguay had been as coopera- tive as in Costa Rica, Uruguay’s unemployment rate would have been 1.9 percentage points lower among the total labor force, 0.9 percentage points lower among women, and 2.0 percentage points lower among youths, ceteris paribus. Of course, these figures should be interpreted with some caution. However, they illustrate that the magnitude of the effect, although modest compared to industrial countries, is still likely to be noticeable.16

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 61

We also checked the robustness of our regression results (Tables 1 to 3, columns 2 to 5). In a first check, we dropped the two geographical controls. In a second check, we dropped the variables “ethnic fractionalization” and “war.” In a third check, we addi- tionally controlled for the percentage share of children in the population.17 This share varies widely across countries. Although large variations in this share are more likely to affect employment rates, they may also affect unemployment rates. In a final check, we added the EFW variable “secure property rights and strong rule of law,” lagged by one year. This variable indicates the extent to which countries protect their citizens’ prop-

Table 1. Regressions to Explain the Unemployment Ratea

(1) (2) (3) (4) (5)

Industrial relations quality -9.89*** -11.14*** -9.84*** -9.18*** -10.59*** (-5.61) (-4.93) (-4.97) (-6.62) (-7.15)

Flexible labor market regulations -3.15 -3.36 -3.11 -3.95 -6.61** (-0.95) (-1.14) (-1.10) (-1.02) (-2.29)

Flexible business regulations -1.92* -2.48*** -1.22 -1.20 -4.50*** (-1.91) (-3.26) (-1.39) (-0.93) (-4.41)

Low top marginal tax rate 3.49 1.55 2.57 3.44 3.80 (0.86) (0.40) (0.67) (0.90) (0.87)

GDP growth gap -0.49 -0.48 -0.41 -0.52 -0.63 (-0.83) (-0.81) (-0.68) (-0.91) (-1.13)

GDP per capita -0.15 0.03 -0.23 -0.29*** -0.23 (-1.37) (0.41) (-1.53) (-2.82) (-1.48)

Transition country 0.13 2.34*** 0.06 -2.13* 0.11 (0.17) (4.02) (0.10) (-1.84) (0.19)

Tropical area -7.12*** -5.40*** -6.22*** -6.40*** (-3.45) (-4.95) (-2.68) (-2.77)

Distance to coastline -2.68* -2.17 -2.41 -2.28 (-1.67) (-1.42) (-1.33) (-1.52)

Ethnic fractionalization 8.84*** 4.17 8.47*** 9.01*** (2.65) (1.28) (2.72) (2.82)

War 1.07 0.73 1.04 1.18* (1.48) (1.29) (1.39) (1.69)

Population aged 0–14 -0.24*** (-3.27)

Secure property rights and strong rule of law

8.60*** (4.35)

Number of observations 128 128 128 128 128 Number of countries 45 45 45 45 45 R2 0.34 0.28 0.32 0.36 0.37 Standard error of regression 1.54 1.59 1.52 1.50 1.49 F-statistic 4.25*** 3.73*** 4.41*** 4.19*** 4.44*** Hausman test 10.88 20.94 11.38 13.86 9.84

Notes: a Feasible generalized least-squares estimates with country-specific random effects (Swamy-Arora method). All regressions are based on data for the years 1995, 1996, and 2000 to 2003. The following variables were lagged by one year: “industrial relations quality,” “flexible labor market regulations,” “flexible busi- ness regulations,” “low top marginal tax rate,” “secure property rights and strong rule of law.” Heteroskedasticity-consistent t-statistics in parentheses (White method). * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions also contain year dummies and a constant term.

62 Horst Feldmann

erty rights and adhere to the rule of law. In order to earn high marks on the rating scale, countries must effectively protect private property rights and ensure a strong rule of law. Unprotected property rights and a lacking rule of law undermine both people’s incentive to take up a gainful employment and enterprises’ incentive to hire workers and innovate; they may thus result in relatively high unemployment. Using data from large groups of countries, Feldmann (2004, 2007) found evidence supporting this hypothesis. In all of these checks, the coefficient on our industrial relations quality variable is statistically significant and very similar to the results from our basic speci- fication (Tables 1 to 3).18

Table 2. Regressions to Explain the Female Unemployment Ratea

(1) (2) (3) (4) (5)

Industrial relations quality -4.72*** -6.08*** -5.01*** -3.76** -5.90*** (-3.29) (-2.77) (-3.23) (-2.55) (-3.24)

Flexible labor market regulations -7.46*** -7.79*** -7.29*** -8.62*** -11.72*** (-3.95) (-3.71) (-4.33) (-2.88) (-9.43)

Flexible business regulations -0.75 -1.34 0.12 -0.05 -3.93 (-0.27) (-0.45) (0.04) (-0.02) (-1.48)

Low top marginal tax rate 3.78 2.22 2.86 3.56 4.08 (0.84) (0.52) (0.68) (0.87) (0.84)

GDP growth gap -0.49 -0.49 -0.45 -0.51 -0.67 (-0.74) (-0.82) (-0.65) (-0.79) (-1.17)

GDP per capita -0.20* -0.02 -0.27* -0.37** -0.28 (-1.94) (-0.38) (-1.87) (-2.62) (-1.63)

Transition country -2.03 1.04* -2.01 -5.94*** -2.06 (-1.44) (1.76) (-1.50) (-2.72) (-1.32)

Tropical area -7.98** -6.32** -6.31* -6.93* (-2.36) (-2.49) (-1.86) (-1.91)

Distance to coastline -2.19 -1.44 -1.76 -1.56 (-1.41) (-1.30) (-0.92) (-1.08)

Ethnic fractionalization 8.29* 2.78 8.10** 8.41** (1.97) (0.60) (2.34) (2.27)

War 1.10 0.86 1.09 1.22 (1.35) (1.65) (1.30) (1.53)

Population aged 0–14 -0.39*** (-3.19)

Secure property rights and strong rule of law

10.81*** (3.08)

Number of observations 121 121 121 121 121 Number of countries 44 44 44 44 44 R2 0.27 0.21 0.25 0.29 0.31 Standard error of regression 1.69 1.73 1.68 1.62 1.63 F-statistic 2.74*** 2.42*** 2.92*** 2.89*** 3.07*** Hausman test 6.03 12.82 7.35 11.15 5.27

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 63

4. Conclusion

Our regression results indicate that more cooperative industrial relations are likely to lower unemployment in developing countries. The magnitude of the effect appears to be moderate, both among the overall labor force and among youths. It seems to be small among women. Our results are robust to variations in specification. They cor- roborate the hypothesis that cooperative industrial relations improve labor market performance. Furthermore, they are in line with previous empirical studies that have

Table 3. Regressions to Explain the Youth Unemployment Ratea

(1) (2) (3) (4) (5)

Industrial relations quality -10.75* -14.34* -10.42* -9.30* -13.18** (-1.87) (-1.89) (-1.81) (-1.67) (-2.28)

Flexible labor market regulations -2.70 -4.33 -1.81 -3.75 -13.19*** (-0.38) (-0.60) (-0.26) (-0.49) (-4.49)

Flexible business regulations -14.81*** -15.52*** -13.68*** -13.71*** -22.63*** (-2.76) (-4.64) (-2.98) (-2.66) (-3.08)

Low top marginal tax rate 12.10 8.84 11.13 11.63 12.01 (1.52) (1.16) (1.39) (1.52) (1.39)

GDP growth gap -1.26 -0.84 -1.10 -1.25 -1.82 (-0.74) (-0.68) (-0.64) (-0.74) (-0.88)

GDP per capita -0.20 0.19 -0.30 -0.40 -0.48 (-0.92) (0.66) (-1.47) (-1.38) (-1.44)

Transition country -1.15 4.37 -1.15 -5.02 -1.40 (-0.31) (1.30) (-0.30) (-1.45) (-0.40)

Tropical area -14.34*** -12.25*** -12.96*** -11.49*** (-3.35) (-3.97) (-4.03) (-3.28)

Distance to coastline -1.64 -0.62 -1.24 -0.55 (-0.64) (-0.27) (-0.50) (-0.26)

Ethnic fractionalization 12.82*** 4.85 11.84** 12.77*** (3.24) (0.56) (2.55) (2.97)

War 1.17 0.55 1.21 3.83** (0.95) (0.60) (0.96) (2.21)

Population aged 0–14 -0.38** (-2.12)

Secure property rights and strong rule of law

26.10** (2.25)

Number of observations 93 93 93 93 93 Number of countries 38 38 38 38 38 R2 0.35 0.28 0.34 0.36 0.39 Standard error of regression 3.36 3.52 3.34 3.27 3.18 F-statistic 3.01*** 2.66*** 3.36*** 2.85*** 3.32*** Hausman test 5.03 11.28 5.25 8.10 4.06

64 Horst Feldmann

tested this hypothesis for industrial and transition countries. They are also in line with previous empirical studies indicating that more cooperative industrial relations are likely to increase foreign direct investment inflows, which may be an important channel through which cooperative industrial relations lower unemployment. Having said this, the transmission channels from industrial relations quality to aggregate labor market performance need to be systematically analyzed in future research. Similarly, the reasons why industrial relations quality appears to have a larger impact on unemploy- ment in industrial countries than in developing countries also need to be examined.

Appendix: List of Countries

Argentina, Bolivia, Brazil, Bulgaria, Chile, China, Colombia, Costa Rica, Croatia, Czech Republic, Ecuador, Egypt, El Salvador, Estonia, Greece, Guatemala, Honduras, Hungary, India, Indonesia, Jamaica, Latvia, Lithuania, Malaysia, Mauritius, Mexico,

U n em

p lo

y m

en t

ra te

( %

)

Industrial relations quality

y = –5.23x + 32.43

R2 = 0.23

3.0 3.5 4.0 4.5 5.0 5.5 6.0

Figure 1. Industrial Relations Quality and Unemployment

F em

al e

u n em

p lo

y m

en t

ra te

( %

)

Industrial relations quality

y = –5.76x + 36.28

R2 = 0.18

3.0 3.5 4.0 4.5 5.0 5.5 6.0

Figure 2. Industrial Relations Quality and Female Unemployment

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 65

Morocco, Nicaragua, Panama, Peru, Philippines, Poland, Romania, Russia, Slovak Republic, Slovenia, South Africa, South Korea, Sri Lanka, Thailand, Trinidad and Tobago, Turkey, Ukraine, Uruguay, Venezuela.

References

Acemoglu, D., S. Johnson, and J.A. Robinson, “The Colonial Origins of Comparative Develop- ment: An Empirical Investigation,” American Economic Review 91 (2001):1369–401.

Alesina, A., A. Devleeschauwer, W. Easterly, S. Kurlat, and R. Wacziarg, “Fractionalization,” Journal of Economic Growth 8 (2003):155–94.

Bassanini, A. and R. Duval, “The Determinants of Unemployment across OECD Countries: Reassessing the Role of Policies and Institutions,” OECD Economic Studies 42 (2006):7–86.

Blanchard, O. and T. Philippon, “The Quality of Labor Relations and Unemployment,” NBER Working Paper 10590 (2004).

Blanke, J., F. Davier, and F. Paua, “The Executive Opinion Survey,” in World Economic Forum, The Global Competitiveness Report 2002–2003, Oxford: Oxford University Press (2003).

Center for International Development, “geodata.xls,” Cambridge, MA: Center for International Development at Harvard University (1999).

Center for International Development, “physfact_rev.xls,” Cambridge, MA: Center for Interna- tional Development at Harvard University (2001).

Centre for the Study of Civil War, “Armed Conflict Dataset, Version 3-2005b.” Available at http://www.prio.no (2005).

Cooke, W.N., “The Influence of Industrial Relations Factors on U.S. Foreign Direct Investment Abroad,” Industrial and Labor Relations Review 51 (1997):3–17.

Cooke, W.N. and D.S. Noble, “Industrial Relations Systems and US Foreign Direct Investment Abroad,” British Journal of Industrial Relations 36 (1998):581–609.

De Haan, J., S. Lundström, and J.-E. Sturm, “Market Oriented Institutions and Policies and Economic Growth: A Critical Survey,” Journal of Economic Surveys 20 (2006):157–91.

Easterly, W. and R. Levine, “Africa’s Growth Tragedy: Policies and Ethnic Divisions,” Quarterly Journal of Economics 112 (1997):1203–50.

European Commission, Employment in Europe 2005: Recent Trends and Prospects, Luxembourg: Office for Official Publications of the European Communities (2005).

Feldmann, H., “Lohnfindungssysteme, Arbeitsbeziehungen und Arbeitsmarkt-Performance,” Jahrbuch für Wirtschaftswissenschaften 54 (2003):32–59.

Y o

u th

u n

em p

lo y

m en

t ra

te (

% )

Industrial relations quality

y = –13.49x + 80.31

R2 = 0.33

3.0 3.5 4.0 4.5 5.0 5.5 6.0

Figure 3. Industrial Relations Quality and Youth Unemployment

66 Horst Feldmann

http://www.prio.no

Feldmann, H., “Rechtssystem und Arbeitsmarkt-Performance,” Schmollers Jahrbuch 124 (2004):471–509.

Feldmann, H., “Labour Market Institutions and Labour Market Performance in Transition Countries,” Post-Communist Economies 17 (2005a):47–82.

Feldmann, H., “Hayek’s Theory of Cultural Evolution: A Critique of the Critiques,” in J.G. Backhaus (ed.), Entrepreneurship, Money and Coordination: Hayek’s Theory of Cultural Evo- lution, Cheltenham: Edward Elgar (2005b).

Feldmann, H., “The Quality of Industrial Relations and Labour Market Performance,” Labour 20 (2006a):559–79.

Feldmann, H., “Government Size and Unemployment: Evidence from Industrial Countries,” Public Choice 127 (2006b):451–67.

Feldmann, H., “Credit Market Regulation and Labor Market Performance around the World,” Kyklos 59 (2006c):497–525.

Feldmann, H., “Economic Freedom and Unemployment around the World,” Southern Economic Journal 74 (2007):158–76.

Freeman, R.B. and J.L. Medoff, What Do Unions Do? New York: Basic Books (1984). Gallup, J.L., J.D. Sachs, and A. Mellinger, “Geography and Economic Development,” CID

Working Paper 1, Center for International Development at Harvard University (1999). Gwartney, J.D. and R.A. Lawson, Economic Freedom of the World: 2005 Annual Report,

Vancouver: Fraser Institute (2005). International Labour Office, Key Indicators of the Labour Market, 4th edn, Geneva: ILO (2005). International Labour Office, “LABORSTA.” Available at http://laborsta.ilo.org (2006). International Monetary Fund, World Economic Outlook. April, Washington, DC: IMF (2003). Karnani, A. and B. Talbot, “Global Manufacturing Networks Survey: A Preliminary Assessment

of Strategic, Operational and Organizational Practices,” Working Paper 700, School of Busi- ness Administration at the University of Michigan (1992).

Marginson, P., P. Armstrong, P.K. Edwards, and J. Purcell, “Extending Beyond Borders: Multi- national Companies and the International Management of Labour,” International Journal of Human Resource Management 6 (1995):702–19.

Nicoletti, G. and S. Scarpetta, “Product Market Reform and Employment in OECD Countries,” Economics Department Working Paper 472, OECD (2005).

OECD, Employment Outlook, Paris: OECD (2002). Prescott, E.C., “Why Do Americans Work so much more than Europeans?” Federal Reserve

Bank of Minneapolis Quarterly Review 28 (2004):2–13. Sachs, J.D., “Tropical Underdevelopment,” NBER Working Paper 8119 (2001). World Bank, “WDI Online”. Available at http://www.worldbank.org (various issues). World Economic Forum, The Global Competitiveness Report 1996, Geneva: World Economic

Forum (1996). World Economic Forum, The Global Competitiveness Report 2000, Oxford: Oxford University

Press (2000). World Economic Forum, The Global Competitiveness Report 2001–2002, Oxford: Oxford

University Press (2002). World Economic Forum, The Global Competitiveness Report 2002–2003, Oxford: Oxford

University Press (2003). World Economic Forum and Institute for Management Development, The World Competitive-

ness Report 1995, Geneva: World Economic Forum, Lausanne: Institute for Management Development (1995).

Notes

1. Over time, there have been several changes to the World Economic Forum’s surveys that are of minor importance to our analysis. First, in 1995 they were conducted in collaboration with the Institute for Management Development, Lausanne, and between 1996 and 2001 in collaboration with the Center for International Development at Harvard University. Second, the number of

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 67

http://laborsta.ilo.org

http://www.worldbank.org

countries covered (including industrial and transition countries) has increased steadily from 49 in 1995 to 80 in 2002. Third, the average number of respondents per country varied somewhat over the years. During the four years from which we use EOS data, the lowest figure was 59 (in 2002) while the highest was 68 (in 2000). 2. Prior to 1995, only a few developing countries were included in the surveys anyway. At that time, the surveys were almost exclusively carried out in industrial countries. There is one addi- tional reason why we had to exclude the results from the 1996 survey. In this year, the question was phrased in the following way: “Industrial relations in your country are becoming more cooperative (1 = strongly disagree; 6 = strongly agree)” (World Economic Forum, 1996). Thus, in contrast to all other years, in 1996 the World Economic Forum asked about changes in industrial relations quality, rather than about the state (or “level”) of industrial relations quality. 3. For our regression analysis, we additionally divided all EOS ratings by ten. 4. The severity rate was calculated using data from ILO (2006). 5. To check whether the three countries make a difference, we also ran regressions excluding them. As it turned out, the results were hardly affected at all. 6. Data on the youth unemployment rate are partly from European Commission (2005). “Youths” refers to the age group 15 to 24. 7. The EFW index has been widely used in the empirical literature. For a survey, see De Haan et al. (2006). 8. See, e.g., OECD (2002), International Monetary Fund (2003), Nicoletti and Scarpetta (2005), Bassanini and Duval (2006). 9. All five indicators carry equal weights. This is also true for the EFW component “labor market regulations.” 10. For the purpose of our regression analysis, we divided all EFW ratings by ten. 11. The tax burden on labor (“tax wedge”) would have been a preferable indicator. However, data on this indicator are available for industrial countries only. The EFW indicator “top mar- ginal tax rate” can be regarded as a proxy for the tax burden on labor, because countries with a large (small) tax wedge usually also have a high (low) top marginal income and payroll tax rate and a low (high) income threshold at which the top marginal income tax rate applies. Besides, Feldmann (2006b) found that high top marginal income and payroll tax rates, and low income thresholds at which top marginal income tax rates apply, exert a detrimental impact of their own on unemployment in industrial countries. 12. The output gap would have been the best indicator to control for the impact of business cycle fluctuations. However, data on this variable are also available for industrial countries only. 13. GDP per capita is in thousands of constant 2000 international dollars, converted using purchasing power parities. The source for both GDP per capita data and data to calculate the GDP growth gap is World Bank (various issues). 14. Ethnic fractionalization is defined as one minus the Herfindahl index of ethnic group shares, reflecting the probability that two randomly selected individuals from a population belong to different groups (Alesina et al., 2003). 15. In Figures 1 to 3, each data point represents one developing country. Due to data availability, there are 45 countries in Figure 1, 44 in Figure 2, and 38 in Figure 3. The EOS scores, plotted on the horizontal axes, are averages over the years 1995 and 2000 to 2002. The unemployment rates, plotted on the vertical axes, are averages over the years 1996 and 2001 to 2003. 16. A few remarks on the estimates for the controls may be appropriate. While neither the top marginal tax rate nor the GDP growth gap have a statistically significant effect, flexible business regulations appear to benefit the overall labor force and particularly young people. Women seem to benefit from flexible labor market regulations. Female unemployment also appears to fall as GDP per capita rises. By contrast, ethnic fractionalization is likely to have a strong adverse impact on the total labor force as well as on both demographic groups. Finally, our regression results do not support the hypothesis that adverse geographical conditions increase unemploy- ment. This result, although counterintuitive, is not surprising. The geographical view of economic development held by Sachs and co-authors has been seriously challenged in the recent empirical literature. For instance, Acemoglu et al. (2001) found that latitude (a variable often used because

68 Horst Feldmann

countries that are closer to the equator tend to have more tropical climates) does not adversely affect GDP per capita once they control for endogeneity of institutions. According to Feldmann (2005b), in panel regressions both of income per capita and economic growth the estimated coefficient on latitude is statistically insignificant once unobserved country effects are controlled for using the random effects method. Feldmann (2006c) also found that more adverse geographi- cal conditions are correlated with lower unemployment in panel regressions based on data from 74 industrial, developing, and transition countries. 17. Source: World Bank (various issues). “Children” refers to the age group 0 to 14. 18. For two reasons, we do not include the variables “population aged 0–14” and “secure property rights and strong rule of law” in our basic specification. First, each of these variables is fairly closely correlated with some of our basic controls. The former is highly correlated with both “GDP per capita” (correlation coefficient -0.69) and “transition country” (-0.73), and moder- ately correlated with “low top marginal tax rate” (0.56) and “tropical area” (0.59). The latter additional control is moderately correlated with both “flexible business regulations” (0.55) and “GDP per capita” (0.52). The theoretical arguments for including “population aged 0–14” and “secure property rights and strong rule of law” are not as strong as those for including the other five control variables. The second reason for excluding the two additional controls from our basic specification is the moderate number of observations, which bars us from employing too many independent variables jointly.

INDUSTRIAL RELATIONS AND UNEMPLOYMENT 69