Supply and demand factors in understanding the educational earnings differentials: West Germany and the United States.

• Author: Ozturk, Gulgun Bayaz
• Position: Report

1. Introduction

   There is a vast literature that has documented the trends and differences in earnings inequality in the United States and the European countries. Widespread consensus indicates that the United States and the United Kingdom stood out from rest of the developed countries because of significant widening in their wage distributions during 1980s. There have been many studies that tried to explain the wage differentials across countries and some of those held institutional constraints responsible. They tried to show that those countries with slow growth in wage inequality were usually the ones with more centralized wage setting institutions.

   In this paper, we try to explain the earnings differentials of two countries, the United States and West Germany during 1980s and 1990s, by focusing on educational earnings differentials. Since the evolution of wages and inequality are substantially different in West and East Germany, we restrict our analysis to West Germany (Gernandt and Pfeiffer, 2007). Our goal is to understand the importance of institutional factors and the market forces in explaining the earnings differentials across the two countries in the 1980s and how they have changed in the following decade. An advantage of this pair wise comparison is less compromise on data quality compared to studies that analyze many countries.

   This study improves upon the existing literature by examining the trends in the 1990s and early 2000s which is a period shaped by major policy changes (i.e., trends towards decentralization in German labor markets). Further, it uses a different skill grouping to achieve better comparison between the two countries. Since the way skill groupings are formed might disguise the recent phenomenon of job polarization, we run an additional test to investigate whether our sample reproduces evidence that is consistent with the recent literature or not.

   The paper is organized as follows. The next section reviews the literature. Section 3 analyzes the trends in earnings inequality in both countries. Section 4 uses data from the March CPS files and the German Socio-Economic Panel (GSOEP) along with Cross National Equivalent Files (CNEF) and implements the relative demand and supply framework to investigate the sources of earnings differentials between the two countries. Section 5 provides evidence for a recent phenomenon in the labor markets; i. e. polarization of work in both countries. And the last section concludes.

2. Literature Review

   From the late 1960s to the beginning of 1980s, most of the OECD nations including the United States and West Germany experienced rapidly declining educational wage differentials. The primary reason underlying this common pattern among countries was the rapid increase in supply of college graduates despite the shifts in demand favoring highly educated workers. However, experience of the United States diverged from that of other OECD nations in the 1980s. Rapidly increasing educational attainment in most OECD countries muted changes in their wage structures. On the other hand, slower growth in the supply of highly educated workers in the United States, combined with shifts in relative demand favoring skilled workers, resulted in rising wage differentials by skill and, hence increasing overall inequality (Freeman and Katz, 1994). A survey by Levy and Murnane (1992) concluded that the shifts in demand and supply provide the major explanation for the wage dispersion between skill groups.

   Most of the developed nations operate in similar economic environment. They adopt similar technologies in their production processes that lead to comparable industry and occupation mixes. That's why, shifts in relative demand for skilled workers will not radically differ among these countries. Most economists argue that relative demand shifts occurred due to skill-biased technological change. In other words, a change in the production process led to increasing marginal product of skilled workers relative to the unskilled workers, which resulted in increased relative employment of highly educated workers in most sectors of the economy (Katz and Murphy 1992). However, starting with the beginning of 1990s, the implied monotonic changes in skill demand were replaced with non-monotonic rises in skill demand. Research showed that this new pattern is due to the emergence of job polarization in the labor market and so required a nuanced version of skill-biased technological change (Autor, Katz and Kearney 2008, Spitz-Oener 2006, Dustmann, Ludsteck and Schonberg, 2007)

   Given that developed nations faced similar relative demand shifts, economists have come up with two major answers to explain the cross-national wage differentials. The first one is a market-based explanation which emphasized the importance of market forces (in particular, rising relative skill supplies) underlying the small increases in overall earnings inequality in the European countries. And the second one was the role of institutional constraints and its effects on wage determination. Studies showed that countries with centralized labor market institutions (Sweden, France, Germany, the Netherlands) experienced smaller increases in overall inequality, while the countries with decentralized labor markets experienced larger increases (United States, United Kingdom). This led to the hypothesis that institutional factors were responsible for the relatively small increases in the overall inequality in most of the European countries (Freeman and Katz 1994, Katz et al. 1995, Gottschalk and Joyce, 1998, Acemoglu 2003).

   Krugman (1994) argues that "in a strong welfare state the increase in underlying pressures toward inequality may not be clearly visible in the actual distribution of earned wages, since those workers whose relative wages would have fallen the most are instead priced out of the labor market" (Krugman, 1994, p. 31) Hence, increasing inequality in the United States and rising structural unemployment rates in the European countries are the "two sides of the same coin" that occurred as a result of technological change favoring skilled workers (see Figure 1 for the evolution of unemployment rates in both countries).

   Since then there have been several attempts both to understand the role of labor market institutions in explaining the international differences in wage inequality and to test the Krugman hypothesis. The empirical evidence is sparse. Some economists emphasized the importance of market forces without abandoning the impact of institutional constraints in determining wages (Gottschalk and Joyce 1998). Others suggested that wage-setting institutions were an important determinant of cross-national differences in wage distributions (Blau and Kahn 1996, Dinardo, Fortin and Lemiuex 1996, Dinardo and Lemieux 1997). However, despite recognizing the role of institutional constraints and their impact on wage structures, some economists showed that Krugman hypothesis did not hold (Card, Kramarz and Lemieux 1999 and Nickel and Layard 1999).

   [FIGURE 1 OMITTED]

   Most of the existing research studied the wage structures in the 1980s and before. And also, they focused on many countries with a few years of data points. Hence, in order to achieve comparability across countries, they had to compromise on data quality. In this study, we investigate the relative importance of market forces and institutional factors to explain the educational earnings differentials in the United States and West Germany during the 1980s and 1990s. We use the simple supply and demand framework described by Katz and Murphy (1992) and Acemoglu (2003) to differentiate the influence of institutional factors on the changes in relative wages of skilled workers.

3. Trends in Overall Inequality

   Table 1 summarizes overall inequality in each country using three different measures. The first two panels for each country document the coefficient of variation and Gini coefficient of the distribution of earnings. We adjust earnings in the United States using CPI-UX1 inflation index. In West Germany, earnings are adjusted using CPI obtained from the CNEF. (2) Moreover both data sets are weighted to represent their respective populations.

   For confidentiality purposes, earnings are top coded in the United States. However, there is no top coding in the German data set. To be able to compare the labor market outcomes across countries and time, we use the procedure that is proposed by Burkhauser, Couch, Houtenville and Rovba (2004). They impose the most restrictive top code on earnings in all years so that the same percentile of the distribution is affected in every year. (3)

   We find that the most restrictive top code in the sample for the United States, which is composed of males aged 25 to 55, is 1.59%. We impose the same restriction on German data set. Moreover, to be consistent with other studies and to eliminate the effects of outliers on the inequality measures, we trim the bottom 1% of earnings distribution in each country. (4) Coefficient of variation (CV) and Gini coefficients are based on trimmed data. Since another solution to top coding is to use inequality measures like percentile points, we also provide estimates for log deviations at different percentile points based on untrimmed data. These measures also allow us to observe trends in inequality across the earnings distribution. The last three panels for each country summarize log deviation between the 90th and 10th percentiles, and their deviations from median earnings.

   Earnings inequality increased in both countries during the last two decades. Average annual growth of both CV and Gini was approximately 0.9% in Germany, while it was approximately 1% (CV) and 0.9% (Gini) for the United States over the whole period. Declining trend in overall inequality continued until 1992 in Germany. This trend was then replaced with a sharp rise in inequality at an annual growth rate of 1.64% between...