Digitized by the Internet Archive in 2011 with funding from Boston Library Consortium IVIember Libraries http://www.archive.org/details/crosscountryineqOOacem 4 I DEWEY L.. Technology Department of Economics Working Paper Series Massachusetts Institute of CROSS COUNTRY INEQUALITY TRENDS Daron Acemogiu Working Paper 02-1 February 2002 Room E52-251 50 Memorial Drive Cambridge, MA 02142 This paper can be downloaded without charge from the Social Science Research Network Paper Collection at http://papers.ssm.com/abstract= xxxxx * 4 Massachusetts Institute of Technology Department of Economics Working Paper Series CROSS COUNTRY INEQUALITY TRENDS Daron Acemoglu Working Paper 02-1 February 2002 Room E52-251 50 Memorial Drive Cambridge, MA 021 42 This paper can be downloaded without charge from the Social Science Research Network Paper Collection http://papers.ssm.com/abstract^ xxxxx at Cross- Country Inequality Trends' Daron Acemoglu, February 7, MIT 2002 Abstract The economics increase in profession has wage inequahty in made but currently lacks a consensus on much less, in (continental) most popular explanations skills considerable progress in understanding the UK over the past several decades, why inequahty did not increase, or increased the U.S. and the Europe over the same time period. for these difTerential trends: increased faster in Europe, and that European labor market institutions pre- vented inequality from increasing. way towards accounting I argue that these two explanations go some for the differential cross-country inequality trends, do not provide an entirely satisfactory explanation. In addition, relative demand this reasoning, ing review the two I that relative supply of I it but appears that Motivated by for skills increased differentially across countries. develop a simple theory where labor market institutions creat- wage compression in Europe also encourage more investment in technologies increasing the productivity of less-skilled workers, thus implying less skill-biased technical change in Keywords: Europe than in the U.S. Relative Supply of Skills, Returns to Education, Skill-Biased Technical Change, Technology Adoption, JEL *I Classification Numbers: ,130. Wage Compression, Wage Inequality. J31. thank Larry Katz and Timothy Smeeding, and seminar participants at Cornell and MIT for Alexis Leon for outstanding research assistance, and the NSF and the Russell Sage comments and foundation for financial support. . Introduction 1 While over the 1980s wage inequaUty and the U.S. and the UK, Eirropean economies there (see, for was less of an education increased sharply in retiu'ns to increase, or even no change, in continental example, Freeman and Katz, 1995, Nickell and Bell, 1996, Blanchflower, Loveman, and Katz, 1995). Table and the for a 1 illustrates these trends log differences of the 90th number of countries by showing an estimate of the education premium and the 10th percentiles of the wage distribution from the Lirxembourg Income Studies dataset (and The U.S. from the CPS, see Section 4 and the Appendix for data details). replicates the estimates of the 90-10 differential for a 2 of the survey by Freeman and Katz (1995) U.S. inequality and skill premia were higher and from there countries, with the exception of for comparison. at the table also of countries from Table The nimrbers show end of the 1970s than in that most other on, they increased faster than in these other cases (perhaps Israel) These cross-country trends are well known. There sus that U.S. wage inequality rose because the relative than the number for the relative supply (e.g., skill also a fairly demand widespread consen- for skills increased faster Katz and Murphy, 1992, Berman, Bound and 1994, Katz and Autor, 2000, Acemoglu, 2002). So developments increased is premia in much why haven't the of continental same Griliches, technological Europe? There are three broad types of answers: 1. The 2. European wage-setting 3. For exogenous or endogenous reasons, technical change has been in The to them relative supply of skiUs increased faster in Europe. institutions prevented wage inequality from increasing. less skill biased Europe. first two explanations have been the most popular among economists, and as the "traditional explanations" . Both of these explanations fit I refer nicely into the relative-supply-demand framework, which has been a very useful tool in understanding ^ the U.S. wage inequality patterns. Europe equality in shoiild They both employment than in the U.S. Is this I 1 also shows that during the 1980s the of skilled .workers increased faster in a for the stabihty of the more rapid increase number European countries of employment in skilled account sufficient to European wage structure? suggests that these explanations account for a significant component of the differential inequality trends (in fact, for substantially originally conjectured!). Nevertheless, did not increase as UK. More relative much specifically, demand I in more than the author had also appears that the relative it many European economies as it demand fmd that the relative-supply-demand framework, with the same does a reasonable job of explaining some of the shifts across countries, number Nevertheless, there are a some degree also to this approach. I for skills did in the U.S. and the and Norway. differences in the cross-country inequality trends, for example, for Finland by rel- use a version of the relative-supply-demand framework to investigate this question. The evidence and in- have been associated with a relatively faster increase in the employiBent of more skilled workers. Table ative suggest that the lack of increase in of cases, in particular, Belgium, where Denmark, and Sweden, premia increased much less than predicted therefore conclude that the traditional explanations do not seem to Israel, skill provide an entirely satisfactory explanation for the differential inequahty trends across countries. Instead, not shift in appears that the U.S. and European relative it the same way over the it curves did past 20 years. Given data quality and compatibility caution. Nevertheless, demand suggests that issues, tliis evidence has to be interpreted with we should be tliinldng of explanations featuring differences in technical change or technology adoption across these coimtries. last part of the paper, skills may I propose a possible explanation not have increased as institutional much in Europe as for in the U.S. wage compression in Europe makes firms more 'Berman, Bound and Machin (1998) look relative The willing to demand basic idea is shifts all of the adopt technologies OECD countries and in their sample. have occurred at the same rate or with the same change from industry intensity across countries. In fact, to infer the overall rate of skill-biased technical data would require a variety of strong assumptions. for that at patterns of skill-upgrading in various industries conclude that there has been skill-biased technical change in However, their results do not imply that these why In the complementary making to unskilled workers, The paper proceeds as follows. In the next section, suggested above. In Section whether cross-comrtry 3, I 4, ineciuality trends use data from the I investigation. I find that Section 5, I demand demand in relative for skills across coimtries. In Luxembourg Income Studies and rmdertake such an although the differential behavior of relative supphes between much of the differences, there for skills did not increase as make a preliminary attempt demands change review the three explanations can be explained by the relative-supply-demand the U.S. and continental Europe explains that the relative I less skill bitised there. develop a simple framework to quantitatively investigate framework assuming similar changes Section change teclinical much is also evidence in continental at developing a simple theory Europe. In where relative differentially across com^itries. Section 6 concludes. Review of the Arguments 2 Traditional Explanations of Differential Inequality Trends 2.1 The traditional approaches explain the differential cross-country inequality trends by differences in the behavior of relative supplies. more rapid increase The first in the relative supply of skills in increase in inequality there. The second explanation claims that the Europe accoimts for the lack of explanation, on the other hand, emphasizes the role of European wage-setting institutions. According to this explanation, it is not the differential growth of skilled workers in the population, but the differential behavior of skilled employment that across countries. their demand More is responsible for differences specifically, firms for unskilled workers, m inequahty trends respond to wage compression by reducing and the employment of that of unskilled workers) increases in Europe compared skilled workers (relative to to the U.S. As a result, the market equihbriates with a lower employment of imskilled workers compensating their relatively liigher Figure 1 wages for in Europe. illustrates the first explanation using a standard relative-supply-demand diagram, with relative supply on the horizontal axis and the relative wage of skilled workers, a;, on the vertical a^s. For simplicity, I drew the relative supply of skills The diagram shows as vertical. supply of skills, will economy of this relative demand lead to higher wage for skills grew faster u;P'-^ to a;US-Post_ first level, is we can think that because of less skill-abunclant countries, the is for skills, the skill also affected skills also increases. premium rises Then the "European" equilibrium exliibit greater inequality no change by the same relative demand This captmes the essence of the in the skill than before. In premium will be at a point Figure fact, 1 like E wloich depicts the case in may wMch second one above this story, (e.g., Krugman, — minimum wages for 1994, OECD, 1994, Blau and Kalm, 1996). is the To captm-e example, because of union bargaining, unemployment benefits, or that keep the earnings of low-skill workers in line with those of high- workers. This can be represented as an institutional wage-setting line different from the relative supply curve as institutional wage-setting drawn unemployment. Now, even in premium might not skill in Figure relative as the of the story, there ^See, among I drew the increase; instead there will this relative supply curve, causing the absence of an increase in the relative supply of gap between the demand and (To make the story stark, and consquently ojf the In the figure, relative miemployiBent caused shown 2. hne as horizontal). The equilibrium now has to be along institutional wage-setting line, is there imagine that wage-setting institutions in Europe prevent wage inequaUty from increasing the not in Erurope.^ Probably the more popular explanation among economists and commentators skill from explanation, where the supply of skills increases faster in continental Europe than in the U.S. is demand imagine that continental Europe but the relative supply of At a simple a given than the relative supply during the recent decades. result of the increase in the relative shifts, for skills, for where the consensus change or increased trade with As a Now inequality. as corresponding to the U.S., skill-biased technical demand that an increase in the skills, be equilibrium unemployment. by the increase relative supply of skills in the demand for skills and the intersection between institutional wage-setting hne. Notice that in the simplest version is full others, Katz, employment of skilled workers, Loveman and Blanchflower and Card and Lemieux (2001). (1995), and the indicated gap simply Murphy, Riddell and Romer (1998) unemployment. reflects unskilled relative to the U.S. is The unemployment increased fact that demand ciu-ves, and This relative- supply-demand framework is demand an empirical investigation of whether these two approaches can provide premia skill in continental in in Section 4, I 3, I present we can employment of develop a framework for undertaking such an investigation, some empirical Changes Differential 2.2 An If workers are large enough to accovmt for the differential behavior of wage inequality Europe. In Section and Europe. the elasticity of substitution between skilled and unskilled workers, investigate whether the observed differences in the growth of relative skilled curves. not only a parsimonious specification, but a satisfactory explanation for the behavior of we know Firms are along their different supply behavior or institutional characteristics of the European economies pick different points along the relative also enables Europe often interpreted as evidence in favor of this explanation.-^ Notice that both of these explanations are "supply-side". relative in results using tliis in the Relative framework. Demand for Skills alternative to the traditional explanations involves differential changes in the relative demand for skills across countries. These differential changes could reflect four distinct forces:'* 1. Different countries could develop their own technologies, with different degrees of unemployment Europe increased for all and Bell (1994), Card, Kramartz and Lemieux (1996) and Krueger and Pischke (1998). Nevertheless, some of the increase in unemployment among the high-education workers in Europe may reflect the effect of wage compression within education groups on job creation (e.g., if firms are forced to pay the same wages to low-skill •'But in contrast to the prediction of this simple story, groups, not simply for the low-education workers. college graduates as the high-skill college graduates, they increasing ''Yet unemployment among another alternative function off their "relative is in See, for example, Nickell may stop hiring the low-skill college graduates, college graduates). to introduce labor demand market imperfections that would also force firms to curves" (the model of Section 5 features labor market imperfections, still along their relative demand curves). An example would be efficient-bargaining between and unions. Nevertheless, even if firms are off their relative demand curves, they will be located along some equilibrium locus, and in this case we have to explain why this equilibrium locus is shifting differentially. I am not aware of any analysis of cross-country inequality trends using a model where firms are off their relative demand curves. Here I will develop an explanation that keeps firms along their relative demand curves, but suggest why the relative demand curves may have changed differentially but firms are firms across countries. skill bias. Some 2. countries could be lagging behind the world technology frontier, and may not have adopted the most recent skill-biased teclmologies. Wliile all countries face the 3. skill-biased technologies same technology from frontier, some may have adopted more this frontier. Different countries have experienced different degrees of trade opening, affecting 4. the demand for skills differentially. Although the fourth explanation the simplest, the current consensus is that in- creased international trade played a limited role in the increase in the relative demand is by Katz and Autor, 2000, and Acemoglu, for skills of the recent decades (see the surveys 2002). Therefore, relative demand own I for skills. technologies. In conditions make this context, I focus on the technological reasons for differential changes in the it Plausibly, many advanced economies Acemoglu (1999b), I develop some of their analyzed a model where differences in local profitable for countries to develop showed that greater trade with some skill-scarce of their ov,ti teclmologies. In LDCs may technical change in the U.S., while inducing the development of technologies in Europe. Nevertheless, it gests that the most likely differently in continental why Europe not differential development of the relative skiU- replacing OECD set of technologies. reason for is more appears plausible that most have access, and even relatively rapid access, to the same cause skiU-biased demand for skills may new economies This sug- ha\-e beha\-ed teclmologies or slow technology diffusion, but different incentives to adopt available technologies. In the last section of the paper, I make a preliminary attempt theory of differential adoption of available teclmologies.^ Here main idea, leaving the details to that section. ^An briefly summarize the basic idea of the theory I propose is developed by Beaudry and more harmful to unskilled workers. capital, and hurts unskilled workers. alternative theory of differential cross-country inequality trends Green (2000). They construct a model The The I to develop such a in which capital scarcity introduction of a new technology increases the In Europe, capital is relatively workers do not decline as much. more abundant than demand for is in the U.S., so the relative earnings of unskilled is to link the incentives to adopt wage structure, wliich in part wage compression institutional complementary technologies to the degree of compression in the determined by labor market institutions. In particular, in Eirrope makes firms more wilhng to adopt teclmologies to unskilled workers, inducing less skill-biased technical change there. based on three premises: This theory is There is 1. is new some degree of rent-sharing between firms and workers, example, for because of bargaining over quasi-rents. 2. The 3. A skill bias of technologies variety of labor is determined by firms' teclmology choices. market institutions tend to increase the wages of low-skill workers wages of comparable workers in Europe, especially relative to the All three premises are plausible. firms' profitabihty or investments The both in the U.S. finding of high correlation between wages in union and non-union sectors (e.g., and Katz and Summers, 1989, Groshen, 1991, Blanchflower, Oswald, and Sanfey, 1997) support the view that there is is some amount of rent-sharing. That teclmology adoption close to the heart of and many especially, 1999a, 2000), I economists, and in previous work wage compression in minimum demand for skills it it can help in the U.S. wages, unions and social insurance programs create Europe and increase the pay of workers low-skill OECD, is widely shared 1994, Blau and that firms in Europe may (e.g., 1996). The new find how and wage inequality by most economists, and supported by existing evidence Kahn, endogenous Acemoglu, 1998, developed this theme in detail and showed us understand the behavior of the Finally, the view that (e.g., is more implication of combining these three premises profitable to adopt American counterparts. This is new is technologies with unskilled workers than their because with wage compression, firms are forced to pay higher wages to unskilled workers than they would otherwise do (that is, greater than the "bargained" wage). This creates an additional incentive for these firms to increase the productivity of unskilled workers: they are already paying high wages, and additional investments will not necessarily translate into higher wages. Put differently, the labor market institutions that push the wages of these workers up make residual claimant of the increase in productivity their employers the due to technology adoption, encouraging the adoption of technologies complementary to unskilled workers in Europe.^ A simple numerical example illustrates worker's productivity ogy adopted. is tliis also that solely adoption, the firm's profits are equal to 1/2 x 10 6 = 4. The by the = 5, to 10 —9= 1, since trast, its profits it above 18. adopt the new technology minimum wage will firm's profits without technology legislation be paid 9 now change has to pay 9 to the worker because of the minimrun wage. In con- with technology adoption are laws, the firm without technology This implies that the worker 9. The adopt the new technology'. The reason wage technol- while with technology adoption, firm, therefore, prefers not to requires the worker to be paid at least is when the new Now firm). because of the subsequent rent-sharing. Next suppose that a unless his productivity Suppose that a wages are equal to half of the worker's productivity, and technology adoption costs 6 (incurred — clearly. 10 without technology adoption, and 20 is Assume they are 1/2 x 20 more point still 4. for this Therefore, the change is clear: finxi now prefers to because of the iTiinimum was already forced to pay high wages to the worker, even when marginal product was low, so it became the effective residual Iris claimant of the increase in productivity due to technology adoption. may be This reasoning imphes that there greater incentives to invest in technologies complementing workers whose wages are being pushed up by labor market institutions. Since European labor market institutions increase the pay of low-skill workers, teclinol- ogy may be endogenously An less skiU biased in Europe than in the U.S. additional implicationn of this make job is that institutional wage compression wall creation less profitable in Eiirope, thus as in the second explanation above, we expect unemployment this story, the rise in tionately model in Europe imemployment can be on the rmskiUed, wliich ^This reasoning is to increase relative to the U.S. Interestingly, in is across the board rather than fall dispropor- consistent with the evidence (see the references in similar to the intuition for why firms find it profitable to invest in the training of market imperfections in Acemoglu and Pischke (1999). It also relies on the notion that firms obtain some "rents" from the employment relationship. In the absence of such rents, firms would simply lay off workers when their wages are pushed up. their employees in the presence of labor footnote 3). Therefore, the overall macro predictions of county trends. However, as yet there is approach are consistent with cross- tliis no detailed evidence supporting an empirical investigation of these ideas may be an interesting area now and for futm^e research. Shifts develop the relative-supply-demand framework in more detail for a quantitative Consider the following simple model evaluation of the traditional explanations. e.g., theory, The Relative-Supply-Demand Prame^vork 3.1 I Demand Cross-Country Relative 3 tliis (see, Wekii, 1970, Katz and Mvnphv, 1992, or Acemoglu, 2002): there are two types of workers, unskilled (low-education) workers and skilled (liigh-education) workers. denote the employment of unskilled and skilled workers in country j at time and H^ These employment {t). levels because of differences in the education may The aggregate production function for by L^ {t) vary across countries and over time both levels of the population, presence of labor market distortions, there t I may be and also because, in the imemploynient. economy j takes the constant elasticity of substitution (CES) form Y^{t) where p for < 1, and A^, [t) = [{Ai{t)u{t)r + {Ai{t)w{t)rf', and A^ {t) are factor-augmenting technology terms, which are, now, allowed to vary across countries. The and unskilled workers MPl [t) = elasticity of substitution in this production function is The marginal product (1) cr = 1/ (1 — between skilled p). of the two factors can be written as {Al {t)Y [{Af {t)Y + {Ai {t)Y [W ^'~'^" {t) IV [t)y] , (2) and MPh = {K (0)' [{^i {t)Y {H^ Suppose that wages are related and ufj^{t) = pMPl {t). The it) lu {t))-" + K (i)) linearly to marginal product: case where /? = 1 w-'^ (t) = PMPj^ {t) corresponds to workers being paid their full marginal product, with no rent sharing. Irrespective of the value of wiit) That this specification firms will is, 'in Throughout the paper, as returns to schooling) I will be along their relative demand curves. think of o'-' [i] both as a measure of skill premium (such and as a measure of inequality. This earn higher wages commensurate with their will skills as reflected by uj^ (t) (see practice, of course, there are Juhn, many is motivated by a reasoning Murphy and skills other factors, in addition Alternatively, as long as firms are along their relative will Equation be more "skilled", and the market price of Acemoglu, 2002). In to, or instead of. the skill identical workers. demand curve, the skill pre- be (3) can be rewritten in a more convenient form by taking cr This equation shows that the workers, will Pierce, 1993, premium, determining wage dispersion among observationally mium we have MPiity whereby even among observationally equivalent workers some and they /3, H^ {t) / U {t), skill \A',{t)J premium is logs, \V{t)j a decreasing in the relative supply of skilled except in the special case where a —> oo (where skilled and unskilled workers are perfect substitutes). Another important point to note from this equation is that as long as > 1, i.e., as /A] (t) premiimi and wage in- cr long as skilled and unskilled workers are gross substitutes, an increase in Aj^ corresponds to skill-biased technical change and raises the equahty. Interestingly, skill- biased when cr < 1, it is a decline in technical change (see Acemoglu, 2002). A-'^^ {t) skill /Af But the (i) (t) that corresponds to case with cr < 1 is not of great empirical relevance in the context of skilled and imskilled workers, since ahnost existing estimates suggest that cr > 1 (e.g., Freeman, 1986). 10 all . . Let us start with a relatively weak form of the common technology assumption. In particular, suppose that Ai it) = viO' (0 Ah (0 and Aj = {t) This assuiBption can be interpreted as follows. There by Ah tries (t) and Ai (f), wliich potentially may differ in their ability by the term 6-' {t). advantages ti\'e assumed to be time J c- = a [t] = ^^ or less skill-biased over time. sectors. In addition, comitries (4), and rfl tliis is captured time varying, different it is compara- (though these are /?/ In {A^ {t) = /Ai c'+\na{t) - 1 [t)) is In (f^) (6) , the measure of skill-biased technical change, and ^izln^ ^vi/vi- A In a [t), where j = for the change using an estimate for the elasticity of substitution, a Alna(0 = refers to the U.S. imskilled workers, a, consensus that e.g.. is may have the ternis Then, using U.S. data we can construct an estimate by and Comi- we obtain Ino;^ (i) In a world technology represented world technology b\- (5) invariant). Substituting (5) into where ability to use two sectors as captured iu the is ' (i) to use the world technology efficiently, Although the symmetric between the two becomes more A^ r^^O^ [t] it is ou. Although the to pin is difficult greater than 1, , {t), denoted as: n . elasticity of substitution down a i..V^"(0 + -^1" Aln.^^'(0 in In precisely, there is perhaps aroimd 1.4, a between skilled and fairly well-established but possibly as large as 2 (see, Freeman, 1986, Katz and Mirrphy, 1992, Angrist, 1995, Card and Lemieux, 2001). Hence, in the empirical exercise Now defme I will use cr = 1.4 and Afe as the /c-period difference operator, AfcX Then, predicted changes in the skill = X {t) —X premiirm 11 {t for — cr = 2 as two reference values. i.e., k) country j between between t — k and t are given by:^ AfclncI;^ {t) The new implicit assumption = Afclna(i) - ^A^ln (f^) in this exercise is that there is teclinologies across countries. Instead, it is (7) no delay some quite possible that adoption of in the new of the skill- biased technologies developed or adopted in the U.S. are only introduced in continental Europe with a lag. 4 (i) That = implying that there instead of is, viO' Ah {t) a delay of is - {t y (5), we would have and AJ k^) (t) = 77/^^ (t) A [t - k^) (8) , periods for country j in the adoption of frontier technologies. Motivated by the possibility of such delays, as an alternative method I use U.S. data from 1974 to 1997 to recover estimates of Alna(i), and calculate the average annual growth rate of Ina(i), denoted by predicted change in the skill g. I premimu in then construct an alternate estimate for the comrtry j between dates t — k and as: t A,ln^^-(t)=^fc-^A,ln(|^^ In tliis exercise, I use 1974 as the starting point, since earliest observation for is five any other country from the LIS data, and reasonable time lag for diffusion of technologies as the final year, since this my it (9) is among five OECD the the final year for wliich there is years prior to the years appears as a countries. LIS data for I use 1997 a comitry in sample. Whether the relative-supply-demand framework provides a for cross-coimtry inequality trends skill premium changes, the A^ actual changes, the ^Equivalently, At we can can then be investigated by comparing the predicted \\\Cj-' (f)'s from A In x^ (t) = A In a;^ fit and the Afclna)-' (i)'s from (9), to the write: = a provides a good (7) Inu;^ (i)'s. Alnc^(t) where satisfactory explanation (t) — A In x° (t) with j Aln = to the data. 12 ' ' \V ^ (t) as the U.S., and investigate whether this equation Can 4 Differences in Relative Supplies Explain In- equality Differences? Data 4.1 I now undertake a tive supplies preliminary investigation of whether differences in the behavior rela- can explain the differential inequality trends across countries. More cally, I investigate whether equations country inequality trends, for I (7) and use data from (9) specifi- provide a good description of the cross- Luxembourg Income Study (LIS) dataset (see, example, Gottschalk and Smeeding, 1997, for an im-estigation of cross-country come inequality using this dataset, of differences in ferent years, I wage inequality). and Gottschalk and Joyce, 1998, Because LIS data for in- an investigation for different countries refer to dif- combine these with the March Current Population Siurveys (GPS) for the corresponding years. There are a number of difficulties on years of schoohng or education there is in using the is LIS for this purpose. First, missing for a nmnber of countries. In particular, no consistent education information for the UK and Sweden, lands, the education categories change over the sample period in a difficult to for information and for the way Nether- that makes it construct comparable relative supply numbers. Second, for some countries, example, Belgium and Prance, incomes are reported after taxes. Third, because the LIS sample is limited to household heads, relative supplies have to be constructed for heads of households instead of the whole population. information skill is Finally, because weeks worked missing for some countries, annual earnings have to be used to construct (coUege) premia. These difficulties notwithstanding, the LIS provides a convenient data set for a preliminary look at whether differences in relative supplies could account for the differential inequality trends. The skill premia are obtained from a log-wage regression analysis using annual earn- ings of full-time- fuU- year male household heads aged of 18-64. The "college premium" is estimated as the coefficient on the diunmy for college education. Following other authors, in the case where there (this is the case for the is no education information, I use information on occupation UK and Sweden, see the Appendix). 13 The regression also contains other education dummies, a quartic in age (since experience cannot be constructed for every country), and a race or immigrant For the 1.5 CPS calculations, when dummy when apphcable (see the Appendix). earnings are top coded, they are assigned the value of times the tope code.^ The 90-10 wage differential is simply the log difference between the earnings of the household head at the 90th percentile and those of the household head at the 10th percentile (again the sample is limited to full- time- full- year Relative supplies are constructed adapting the (1998) to the data available here. equivalents divided by the number In particular, method it is male heads of households). of Autor, Katz and Krueger the ratio of niuuber of college- of noncollege-equivalents. Here, I define college and noncollege equivalents as follows: college equivalents=college graduates-|-0.5x workers with some college, and noncollege equivalents=high school dropouts and graduates-|-0. 5 x workers with some col- lege. When (see the college information is Appendix). Moreover, not available, occupations are used for this calculation for the Netherlands and Germany, there category for those with some college.^ is no separate Presumably, workers with some college are al- ready included either with those with secondary education or among those with college education. Tliis countries. For may create problems with the relative supply estimates from these two aU these calculations, I use a sample of male heads of households rather than the whole population, since the LIS only has information To for heads of households. the extent that there are differential trends over the years across different countries in female labor force participation, this hmitation of the LIS may create biases in the analysis here. ^Because the sample is narrower than usual (in particular, it excludes women), is not weighted by weeks worked, and includes age instead of experience controls, the college premium for the U.S. is estimated to be lower than usual. ^For Germany, we only use data from 1989 and 1994. The survey also includes data for 19S1 and 1984, but the 1981 survey is the first LIS survey, and data quality is low and LIS advises against the use of this survey. The 1984 survey has a very different classification of education, and implausibh' few people outside the categories of high school and high school and less. .Although I report overall wage inequality numbers from the survey in Table 1, I do not use it for calculating returns to schooling or relative supplies. 14 Because there are a number of problems with the French data are reported after taxes, there is is no education variable for the first not possible to limit the sample to full-time- full year workers), the analysis. I also exclude (for LiLxembourg since the sample I example, incomes two surveys, and it exclude France from size is small. Empirical Results 4.2 Figures 3a-c and 4a-c show the results of the empirical exercise. Figures 3a-c depict the estimates obtained from (7) for two values of the elasticity of substitution, cr = 2, while Figures 4a-c show the estimates from of the elasticity of substitution. In all A/elnu)-' (i), are translated into level cases, the of patterns are apparent and which estimate of the and Canada, the predicted for the UK, the college skill change estimates, Aklnuj-' premium from the and and the (t) year estimate for which method and First, for Australia premia estimates are close to the actual estimates, while premium increases more than predicting too large an increase in the skill premia is first figures, irrespective of predicted. Therefore, there evidence that the relative-supply-demand framework with This conclusion 1.4 for that country. elasticity of substitution are used. skill = again for the same two values estimates by choosing the each country to be the same as the actual A number (9), cr consistent with the results of Katz, common among is no technology trends is the Anglo-Saxon countries. Loveman, and Blanchflower (1995), Murphy, Riddell and Romer (1998) and Card and Lemieux (2001). The predicted skiU premia estimates are also close to the actual estimates for Finland and Norway. Second, for the Netherlands, the predicted skill premia decline, whereas the actual premia remain approximately constant. This reflects substantial increase in the skill relative supplies of skills in this coimtry. Given the shortcomings of the LIS data for the Netherlands noted above, increases in the relative supply of skills are likely to be overstated, and these results need to be interpreted with caution. Next, Norway and Finland the predicted and actual skill premia move more or for Germany, less in tandem. Hence, the relative-supply-demand framework with coimnon technology changes appears to perform well for these countries, relative supply changes may be though it has to be borne in mind that overstated in the LIS data. 15 German Finally, for Belgiiun, Denmark, and Sweden, the predicted premia estimates skill increase substantially while actual skill premia are approximately constant or actually declined. For the Israel, same result is obtained using the first method, but not the second. Given the quality of the LIS data, these results have to be viewed as preUminary, and more work using detailed microdata from these countries is necessary to reach firmer conclusions. Nevertheless, the results suggest that for a significant fraction of the continental European coimtries in the sample, the relative-supply-demand frajBework with common technology trends does not provide an entirely satisfactory explanation for the differential skills behavior of skill premia. Instead, increased substantially less in Belgium, it appears that relative Denmark and Sweden than demand for in the U.S. or other Anglo-Saxon comitries. A 5 I Model now develop of Differential Technology Responses a simple theory which links the adoption of skill-biased technologies to labor market institutions, may have been framework I and provides an explanation less skilled-biased in Europe than on training investments why in the U.S. Tliis developed in Acemoglu (1999a), and combines of the literature for it technical change theory builds on a with some of the insights in the presence of labor market frictions (e.g., Acemoglu, 1997, Acemoglu and Pischke, 1998, 1999). The basic idea to invest in greater skill new is that in a labor market with technologies is wage compression may workers. I affected wage bargaining, the incenti\-es of finns by the degree of wage compression. In particular, increase firms' incentives to raise the productivity of low- then suggest that the greater institutional wage compression in Europe may have encouraged firms to adopt certain technologies with low-skiU workers that U.S. firms did not adopt. Throughout, to I wiU keep the presentation at a heuristic Acemoglu (1999a) for level. The reader is referred a fuUer discussion of a similar model (but without an analysis of the impact of differences in wage-setting institutions 16 on technology clioices). The Environment 5.1 There a continuum is of workers that are infinitely hved 1 value of income discounted at the rate have hu < human capital /i^, wliile the A r. of these workers are skilled, fraction remaining workers are unskilled with human in this economy are created imemployed workers, there matters will skilled that the function for V there are be M({7, V) new matches created. and unskilled workers will ha^•e M exhibits constant returns to simply by the tightness of the labor market, 6 worker When vacancies and a vacancy rate for an Notice that what is q(6) where g unemployed worker, is = the same matching probability rates. scale, so matcliing rates are jj. The determined a decreasing function. The probability of matching in turn, is 6q[6), which is assumed to be increasing at the 7^ when they post a vacancy at time in d. exogenous Once they match t. with a worker, they decide what type of a job to create. In Acemoglu (1999a), I assume s. Firms incur a setup cost more I probability of matching with a assume that matches between firms and workers come to an end also U the total nun:iber of unemployed workers (skilled plus unskilled), and as a is both result, Here capital Diamond via costly search as in the models by Mortensen (1982), and Pissarides (1990). (1982), the and hs- Jobs I and maximize the net present realistic I analyzed (and involved) case where firms decide the type of job before matching. focus on the case where this decision the analysis. Specifically, at time f, is made after matching in order to simplify a firm matched with a worker of hmnan capital h can either produce Ath units of the final good, without incurring any additional costs, or additional investment of cost kt One it can undertake an and produce {I where a > (10) + a) Ath, (11) 0. possible interpretation of the choice between the two technologies 17 is that firms can either use last period's technology at no cost, or adopt the frontier teclinology by incurring an additional cost every period. Notice that the multipHcative nature of the production function in (11) makes technology complementary to new skills. Also in terms of the framework discussed in the previous section, the elasticity of substitution between skilled and unskilled workers unaffected by relative supphes. Tliis is and wage inequaUty is infinity, to simplify the discussion and focus on is differential technology responses of different economies. Analysis 5.2 The discounted net present value of a firm with technology adoption decision Xf e {0, 1} and matched with a worker of Jt (xt, h) = Ath + Xt {aAth - skill kt) ~ h G {hu,hs} can be written as Wt (xj, h) 1 + 1 + sVt+i + (1 - s) r max Jt+i {xt+i,h) ^t+i (12) where Wt (xf, h) of the worker, is the wage as a fimction of the technology decision and the and Vt+i is the value of a vacancy at time t + I. skill level The wage may depend on the teclinology decision because of bargaining over the quasi-rents created by search frictions. The term sVt+i is added because with probability and the firm becomes an imfiUed vacancy. value conditional on no separation, and I Finally, the last s, there term have imposed that there separations. This will be the case in steady state, and I will is will is a separation, the continuation be no voluntary focus on the steady state allocation for simphcity.^" Next suppose that At and kt grow exogenously at a constant rate, g, At = {1 + gYA kt = {l + gfk. i.e., (13) '"Away from the steady state, workers may want to end the relationship with a firm that does not adopt the new technology voluntarily, and I'm ignoring this case in writing equation (12). In any case, such voluntary separations would never happen along the equilibrium path, since the technologj- adoption decisions are taken every period, so the firm would never choose not to adopt the new technology when it knows that this will result in a quit. 18 ' In addition, assume that wages are equal to a fraction wt{xt,h) = The important assumption incorporated l3[l + (3 of the worker's ex post product: Xta)Ath. in this expression new takes place after the firm sinks the cost of (14) that wage determination is teclmology. This is tice given the impossibility of wTiting binding wage contracts, and are negotiated throughout the duration of the employment technology-related decisions are made ' reasonable in practhat, wliile wages many of the relationship, before or at the beginning of this relationship. Using (14) and the fact that both productivity and costs are growing at the rate g as specified by (13), new technology we can rewrite the steady-state value (x = 1) and that of a job that never adopts the new technology J{x,h) where 8 = ~^^^^ is of a job that always adopts the = 6[[l-l3){l (;r = + xa)Ah-xk] 0): (15) the effective discount factor, taking into accomit time preference, separation probabihty and productivity growth. Notice that the firm only receives a fraction cost of investment. full is a standard result, As a 1 — /3 of the output, though it incurs the This result, firms will tend to underinvest in teclmology. going back at least to Grout (1984), and also emphasized in the context of search equihbrium by Acemoglu (1996). It is straightforward to see that, in steady state, firms will choose to adopt the teclmology with a worker of skill level h if and only if J (x = 1, h) > .7 (x = 0, /i), or {l-l3)aAh>k. The skills, and therefore more if (16) fact that the left-hand side of this expression is increasing in technologies are complementary to new h reiterates that likely to new be adopted with skilled workers. Next, consider the case: {\-0)otAK>k. (17) '^More formally, I'm assuming that wages are determined by outside-option bargaining as in Rubenand Shaked and Sutton (1984) (see Acemoglu, 1996, for a justification in the search equi- stein (1982) librium context). 19 This condition ensures that firms are happy to adopt a new technology even with unClearly (17) immediately implies that (1 skilled workers. — (3) aAhs > k. So the new technology will be adopted with skilled workers as well. Then, wages will be given by w^ = P so the skill premirun (or the {1 + Ahu a) _ w^t _ ~ w^ implemented both with towards t in each period hu stable, because new all technologies are and unskilled workers, so technical change rate. In particular, given condition (18), is "neutral" = -it (1 + q{Ot) )max{Jt -(/))max{.7t(x 7j is the cost of unskilled. The the value of a vacancy as:^^ = {x — l,/i„); 1, hs) Jt(x ; = Jt{x = 0, hs)} + 0,/i„)} (19) - (1 g(^t)) max{T4+i;0} J opening a vacancy. With probability q{9t), the vacancy matches with a worker, and with probability is simply hs X he is steady technical change, and wages are increasing at a skilled can be written + Here (18) . straightforward to characterize the equilibrium tightness of the labor market and the unemployment Vt Ath„ a) skills. It is also at time is But wage inequality remains constant rate. = p {1 + measure wage inequality) ^ In this equilibrimn, there and wt 0, this worker is skilled, and with probability firm then decides whether to adopt the captured by the "max" operators. With probability 1 — new technology q{9t), there 1 — d, or not as no match, and is then the firm decides whether to keep the vacancy, obtaining the value Vt+i or shut down. Assume and impose that the cost of vacancy free entry, -y^ also grows at the rate g, that which implies that Vt = K+i = 0. is, 7j = Then, using (15), - = (1 -1- g) 7, we obtain that in steady state: V = ~j + q {9''"') [6 ((1 -f3){l + a)A- k) (</>/i, + (1 0) K)] 0, ^^ Without loss of any generality, I'm using the timing convention that a vacancy that gets period can start production in this period. 20 (20) filled this . 6^"' is the steady-state tightness of the labor where denoting the fact that and the whole tliis market, with the superscript "Pre" refers to the case before the tecliriology shock. This tightness equilibriiun allocation are defined uniquely. Next, the steady state unemployment rate for both skilled and mrskilled workers ^ The important ment will rate, Change Now > i.e., and that is this in unemployment From Ar+t shock, we = rate a decreasing function of is same unemploy- ^: unemployment is tighter. Technology Regime imagine that the economy kx-^'^ '''' 7Tkwy that both skilled and unskilled workers have the be lower when the labor market 5.3 k'j^ point = is:^'' this point (1+5)'^t and is liit by an adverse shock A grow onwards, both k and fcj+j = - aArh, > [l + gYk'j.. Assume at time T, increasing at at the same kj- to rate as before, also that following this technology have: (1 P) > k'j- [l - aArK. (3) Condition (22) implies that firms, fi-om this point onwards, new technology with wages at any time t is now will continue to adopt the with unskilled workers. That skilled workers, but not technology adoption condition (16) (22) satisfied for skilled is, the workers only. As a new result, >T are now Wf = so after this technology shock, l3Athu and u'f = wage inequahty wf , '^• = ^ (3 {I <' a) Ath^. increases to . = + . + hs "'fc:- unemployment rate equation follows by equating to the flow out of unemployment, Oq (6) u, to the flow into unemployment, (1 — s) u. The fact that both types of workers have the same unemployment rate is a consequence of the simplifying assumption that all search is undirected, and no matches are turned down. It is straightforward, but cumbersome to generalize the model by adding some heterogeneity, or by allowing firms to turn down workers as in Acemoglu 1999a) '"Notice that I am taking the technology shock to be an increase in the cost of new investment. This is consistent with the fact that all the countries in question experienced a TFP slowdown over this time period. Nevertheless, many other economists view the 1980s as a period of more rapid technological change (e.g., Greenwood and Yorukoglu, 1997, Hornstein and Krusell, 1996). See Gordon (1998), Jorgensen and Stiroh (2000) and the discussion in Acemoglu (2002). '''This ( 21 unemployment In addition, both types of workers increases. To see now changes the free-entry condition V=-^ + q for (^'-'S-Po^t^) [^ (^ ((1 _ ^) wliich incorporates the fact that miskilled workers. it The to + (1 now ^) ^_ ;,) ;,^ + _ (1 ^) (j change is in technology in the _ ^) y^/^^^] = (23) q, new technology with firms will not adopt the steady-state labor market tightness refers to the equilibrium after the this note that denoted by 9 ~^°^^^ since model economy supposed to approximate the U.S. Comparing this expression to (20) change, opening a vacancy So fallen). ^^^'P°''* < is, increase. 9, tlie 9^^'^, is now shows that, as a consequence of the teclinology less profitable (i.e. the term in square Vjrackets has tightness of the labor market, has to decline to satisfy (23) This impHes, from equation (21), that the imemployment —that rate will ^'^ So at a very simple model level, this illustrates how a technological shock might increase wage inequality by changing the technology adoption decisions of firms. After the change in technology, firms find skilled workers, premium and as a profitable to invest in technologies only with result, the relative productivity of skilled in the Technology Regime Now, consider a different environment, meant tal new workers and the skill increase. Change Hbrium 5.4 it European labor market. In this is a "European" Equi- to proxy at a very crude level the continen- environment, there compression, for example because there for unskilled workers. in the is institutionally minimum wage imposed wage at a level that is binding This contrast captures in a styUstic way the differences between the American and continental European labor markets. For example, Blau and Kalm (1996) provide evidence in favor of the hypothesis that labor market regulations compress wage '^This is differences at the bottom consistent with the fact that of the distribution in Europe. They show that unemployment both among low and high education workers the U.S. increased during the 1980s (see Acemoglu, 1999a). 22 in although wage differentials between the 90th and the 50th wage percentiles are similar and 10th percentiles across countries, the differential between the 50th much is larger in the U.S. than Europe. Union wage setting, which extends wage floors across firms, or mimmum relatively high probably at the root of wages, as well as more generous social insurance progr'ams, are wage compression. tliis Denote the minimum wage in minimum wage has the So I assume ^t = to grow as well ,5(1 is, the + minimum wage not adopt the new —otherwise, u^;. it skilled worker. Since would quickly become Ahu > q) w> PAhu and i3Ahs binding for a firm that is it is And unskilled worker. same configuration minimmn wage In the presence of the Irires it is at all = > t = 0: (24) an miskiUed w^orker and does never binding for a firm employmg a minimum, the worker is the unregulated economy, camiot apply, since to:^'' + Xta)Ath,w^} (25) . mimmum wage is below simply paid the minimum. Since nothing has changed for i.e., (1 But the technology decision will (14) than the minimum. Hence, consider the "eciuilibrium bargained wage" in the absence of the minimum wage at time w. wage equation less rmix{Pi\ skilled workers, the condition for the the irrelevant. t law, the some workers would be paid Wt{xt,h) the growing, not binding for a firm that adopts the a straightforward generalization of this wage equation if is both productivity and minimum wages grow at the same constant rate, (24) ensures the So Since the economy most convenient to assume that it is technology. In contrast, new technology with an (24) implies that by ^ + gYw (1 Also to illustrate the main point, That time effect at — new technology /3) (1 4- a) Ahg for unskilled be binding > to be adopted is the same as in k. workers has changed. In particular, for a firm now employing an unskilled worker with the old technology, so J{x = 0, K) = 6 {Ahu -w). (26) '^This wage rule follows from is straightforward application of the Rubinstein (1982) bargaining model with outside options (see Shaked and Sutton, 1984). 23 , In contrast, worker tlie is still when the firm adopts the Comparing the expressions introduce the (1 + q) Ahu - technology, the (27) new technology with an p) shows that the firm unskilled worker intuition — /?) not binding. will find it profitable to P)aAh^-k> PAK-w. than for the unregulated economy (1 is if (28) is economy to adopt new technologies with new technology when (27) meant that the minimum wage was binding (24), wliich the condition for this restrictive k) minimum wage immediately ensures that the right hand-side of (28) ers, also The new and (26) {1Assumption unskilled given by J[x^l,K) = b{[lbecause, new technology with an value of the firm that adopts the aAhu — k > (recall the work- for unskilled negative. Consequently, unskilled workers is less unregulated economy adopts the 0). as foUows: the institutions already force the firm to pay an unskilled is worker a higher wage than what the firm and the worker would have bargained impMes that the firm can invest more and increase production, without Wage compression productivity being translated into higher wages. is to. This this increased therefore making the firm the residual claimant of the increase in the productivity of the worker. This reasoning is similar to the intuition for why firms find it profitable to invest in the general training of their employees in the presence of labor market imperfections in Acemoglu and Pischke (1999). Also as in Acemoglu and Pischke (1999), labor market rents for firms are crucial for this result. for firms, if institutions With /? = 1, that is, when there are no rents pushed the wages of unskilled workers up, firms would lay off all unskilled workers.'^ Now imagine a situation in which firms invest in new technology with both skilled and unskilled workers, even in the in the absence of the "European" equilibrium workers. minimum will also invest in wage. Then, a fortiori, firms new technology with both types of Next, as discussed in subsection 5.3, imagine a change in technology' regime '^Of course, in this case we also need 7 = for firms to 24 break even in the first place. at time T, increasing the cost of investment from kr to the minimmn wage, stop investing in the fiixms /cj, snch that in the absence of new technology with Because the minimum wage increases the incentives to invest unskilled workers, continue to do so. it In particular, (1 the quite possible that the is new technology - economy with wage compression less skill biased in in this Europe than in the U.S., but in fact, there it why teclinical change in the U.S. over the past 20 years. may have Interestingly, there less skill-biased technical change in Europe than is none. is As a result, wliile the skill premium increases in remains unchanged in Europe. ^^ In light of these results, indicated similar relative little (29) be adopted with unskilled workers in Europe, but not in the U.S. simple model, not only the U.S., will if Therefore, this model offers a possible explanation for been new tecknology with oAtK - PAtK + w> k'T>{l-P) oAtK, /?) will in unskilled workers. wage compression interesting that the brief empirical investigation above it is demand shifts in other Anglo-Saxon comitries. Since there is Anglo-Saxon countries, the analysis here suggests that in these firms there should have no fmrther incentives to adopt workers than U.S. finns. Relative demand new technologies with unskilled shifts in favor of skiUed slower only in economies with significant institutional workers should be wage compression, such as the continental European economies. Wliat about vmemployment in the technology regime? Since firms are "European" equilibrium now adopting the change in the new teclmology with both skilled after the and unskilled workers, the free-entry condition becomes V=-^ + q (0^-^°^*) [6 (((1 -P){l + Recall that in the absence of the a)A- k) [4>K minimum wage, + (!-</>) K))] firms did not find = it 0. (30) profitable to adopt the new technology with unskilled workers, so the term in square brackets is lower ^*The reader might note that the assumption of rent-sharing in the U.S. economy is important for Without rent-sharing, whenever the European economy with wage compression finds it profitable to adopt the new technology, so will U.S. firms. This is not essential for the overall argument, however. It is possible to generalize (and complicate) the model such that wage compression forces European firms to invest in new technologies with unskilled workers, even when the competitive economy would not introduce these new technologies. these results. 25 in (30) (23) than the comparable term in expression has to be liigher than 9 unemployment more compression in ' °'^* imphed by Europe than in the U.S. forcing firms to adopt is (30), new Therefore, (23). ' °^' and the technology shock The impUed by will increase intuition for this result is that wage technologies with unskilled workers, while without wage compression firms would have preferred not to adopt these technologies. This implies that wage compression reduces profits from employing unskilled workers. may be matched with Since each vacancy an unskilled worker, wage either a skilled or compression discourages job creation in general, increasing both skilled and unskilled This pattern of high unemployment both among the skilled and the unemployment. Europe unskilled in As a result, the is '^ consistent with the broad facts. theory presented here is consistent with the differential inequality trends across countries, the differential behavior of the relative documented in the first part of tliis paper, Europe increased and with the 6 fact that relative to that in the U.S. over the 1980s. micro evidence supporting the underlying mechanism of demand But for skills, as unemployment as yet, there is in no story. tliis Concluding Remarks Despite the social importance of inequality trends, and their prominence in academic debates, the economics profession currently lacks a consensus on why, over the past several decades, inequality increased in the U.S. and the UK, but not in continental Europe. In this paper, I reviewed the two traditional explanations for these patterns, that ative supply of skills increased faster in Europe and European labor market prevented inequality from increasing, and I institutions developed a simple framework to quanti- whether these explanations are tatively assess rel- satisfactory. I concluded that, although these traditional explanations accotmt for a large fraction of the differential cross-coimtry '^Notice that if the model is unskilled workers, only unskilled the model, under modified so that firms can open separate vacancies for skilled and unemployment will increase. Nevertheless, it is some assumptions, can generate increases in the workers, contrasting with standard reasoning based on the predicts that it is unskilled unemployment that should 26 unemployment Krugman increase. important to see that rates for both types of hypothesis, which unambiguously inequality trends, they do not provide an entirely satisfactory approach. Instead, pears that relative demand Motivated by this fact, developed a simple theory where labor market institutions I Europe also encourage more investment creasing the productivity of less-skilled workers. This demand for skills over the past 20 years in showed that I this explanation technology adoption ap- for skills increased differentially across countries. creating vv^age compression in in the it based on the may have Europe than in teclinologies in- led to a smaller increase in the U.S. effect of labor market institutions on consistent with the differential inequality trends, the differential is behavior of the relative demand for skills documented in the first part of the paper, and the differential behavior of unemployment rates in Europe and the U.S. These macro facts of course do not establish that lines, this theory is and there can be many other potential theories explaining why change differentially across countries. empirical avenues to pursue. I along the correct relative demands There are therefore both important theoretical and hope that this paper will stimulate others to investigate alternative approaches with differential degrees of skill-biased technology across countries (resulting from differential adoption decisions). On fully the empirical front, work using microdata is necessary to investigate more care- whether differences in relative supplies, with similar shifts in relative demands, can account for the differences in these cross-country inequality trends. Future work could also look at whether there is any direct evidence that technical change has biased in Europe than in the U.S. adoption and/or at rates of industries in tries in both Europe possibility is less skill to look at the rates of teclinology growth and capital accirmulation in unskill-intensive relative to similar industries in the U.S. (using skill-intensive indus- set of countries as a control group). direct investigation of ers, TFP One been Alternatively, one could undertake a whether certain advanced technologies, such as personal comput- computer-assisted or nmnerically-controlled machines, used seldom with unskilled workers in the U.S., were introduced with imskilled workers in Europe. Finally, also be fruitful to it may use data on industry-level skill-upgrading across different countries and expand the approach used imi Berman, Bomid and Macliin (1998) estimates of relative demand shifts across countries. 27 to obtain alternative On why tlie theory front, relative demand it would be useful for skills including theories where the Finally, the current among OECD may have changed economy may function differentially in different countries, off its relative paper has focused on the causes of demand differential document rapid increase and a number of studies in the and low-income demand for skills in find increasing inequality in Freeman and Oostendorp, 2000, or Dmyea nations. many the study Herman and Machin developing countries An causes of the increase in inequality in developing countries over this 28 is middle-income countries, and Szekely, 2000). an important research area. curve. inequahty trends economies. Another important and potentially exciting area of inequality trends in middle-income (2001) to investigate alternative explanations for (e.g., investigation of the same period is also . Data Appendix 7 The samples come from the Luxembourg Income Studies Database is (LIS). The LIS data a collection of micro datasets obtained from annual income sm'veys in various coun- While these surveys are similar tries. form to the Cirrrent Population Survey in the United States and extensive effort has been made to for make information on income and household characteristics comparable across countries, important problems of com- parabihty remain. In order to maintain consistency across countries, several restrictive have been used when defining variables and samples. Following Gottschalk and criteria Joyce (1997), who weighted by weeks worked, Moreover, the data individuals is LIS data, a simple count of the number of workers, not also used is is restricted to not available for measme used as a all of sui.)ply from each educational group. male heads of households, since the earnings of other survey years Comitries for which data in all countries. on gross wages were not available were discarded, as were data from countries with only one survey in the LIS data base. In addition, do not use the German surveys uses data from Australia (1985 91, 94 and 97), Denmark for 1981 and 89), (1987, 92, 95 for the reasons and 1984. As a Belgium (1985, and explained in the text, we result, at the 88, 92 and 97), Finland (1987, 91 96), and end this study Canada (1987, 95), Germany (1989 and 94), Israel (1979, 86 and 92), the Netherlands (1983, 87, 91 and 94), Norway (1986, 91 95). Data The or more and 95), for the college Sweden (1981, 87, 92 United States premimn is, in all and 95), the United Kingdom (1979, 86, 91 and those years comes from the generally, the coefficient March CPSs. on workers with a college degree relative to high school graduates in a regression of log real annual gross wages. Specifically, four education categories were constructed, conceptually corresponding to less-than 12 years of education (liigh school dropouts), 12 years (high school graduates), 13 to 15 years U.S. The recoding (some college), into these groups the education variable is measured is is or more years (college graduates) for the straightforward in countries and years where in years of Finland, at least for some years), but the education variable and 16 schooUng (these are: Israel, somewhat more problematic already grouped (all 29 the rest, except Canada and for countries where Sweden and the United Kingdom) Since no education information is . and the United Kingdom, returns LIS data base for Sweden to broad occupations are used rather than education. Three occupation groups are constructed professional available in the samples, roughly corresponding to for those and managerial workers, blue coUar workers and a residual category which The includes lower-level white-collar workers. regression also includes dummies — remaining education/occupation cattegories and a quartic in age for the since the exact num- ber of years of schoohng would have to be inputed for most surveys, a quartic in age, and not experience, is used as a control in the wage regression. Only full-time, full-year workers aged 18 to 64 are used in these computations. as those whose hours of work per week were no defined as those who worked at least 48 those observations with the lowest 1 less Full-time workers are defined than 35, and The samples weeks per year. workers are full->-ear also lea\-e out percent earnings. Finally, earnings for top coded observations are calculated as the value of the top code times 1.5. The relative supply of skills is calciilated from samples that include salary workers between the ages of 18 and 64. non-college equivalents = = college graduates 0.5 x workers with school dropouts. For the Netherlands, it some -I- Katz and Krueger 0.5 x workers with college was impossible -t- some UK high school graduates 30 + and liigh to distinguish between workers where only occupational categories were available, used in the relative supply computations. (1998), college, with some college and workers with a college degree, so for this country, as and the wage and defined as the ratio of college- Following, Autor, equivalents to non-college equivalents. these are: college equivalents It is all onl}- tliree for Sweden groups were References 8 Acemoglu, Daron, "A Microfoiindation for Social Increasing Retmns in Human Capital," Quarterly Journal of Economics, 111 (1996), 779-804. Acerhoglu, Daron, "Training and Irmovation in an Imperfect Labor Market," Review of Economic Studies 64 (1997), 445-464. Acemoglu, Daron, nical "Why Do New Technologies Complement Tech- Skills? Directed Change and Wage Inequality" Quarterly Journal of Economics, 113 (1998), 1055- 1090. Acemoglu, Daron "Changes Theory and Some Evidence" in Unemployment and Wage Inequality: An Alternative American Economic Review, 89 (1999a), 1259-1278. NBER Working Paper No. 7018, Acemoglu, Daron, "Technical Change, Inequality and the Labor Market", NBER Acemoglu, Daron, "Patterns of Skill Premia", (1999b). Working Paper No. 7800, (2002), forthcoming Journal of Economic Literature, March 2002. Acemoglu, Daron and Jom-StefTen Pischke, "Wliy Do Firms Train? Theory and Evidence," Quarterly Journal of Economics 113 (1998), 79-119. Acemoglu, Daron and Jorn-Steffen Pischke "The Structure of Wages and Investment in General Training" Journal of Political Angrist, Joshua, Strip" Economy, "The Economic Returns 107, (1999), 539-572. to Schooling in the West Bank and Gaza American Economic Review, 85 (1995), 1065-1087. Beaudry, Paul and David Green "The Changing Structure of Wages in the Germany: What Explains the Differences?" Berman, Eli, NBER Working US and Paper No. 7697 (2000). John Bound and Stephen Machin "Implications of Skill-Biased Tech- nological Change: International Evidence," Quarterly Journal of Economics, CXIII (1998), 1245-1280. Berman, Eli, and Stephen Machin "SBTC Happens! Evidence on the Factor Bias of Technological Change in Developing and Developed Countries", Blau, Francine D. and Lawrence M. Kahn 31 mimeo (2000). "International differences in male wage inequality: Institutions versus market forces," Journal Political Economy 104, (1996) 791-837. Card, David, Francis Kramartz and Structure of Wages and Employment: A Thomas Lemieux "Changes ; Comparison of the United in the Relative States, Canada and Prance" Canadian Journal of Economics (1996). Card, David and to College for Diamond, Thomas Lemieux, "Can Falling Supply Explain the Rising Younger Men?," Quarterly Journal of Economics Peter, "Wage Determination and Return (2001). Efficiency in Search EquiUbrium." Re- view of Economics Studies 49:2 (1982): 217-27. Duryea, Suzanne and Miguel Szekely "Labor Markets in Latin America: the Supply-side" Emerging Markets Review, Vol 1, A Look at 199-228. Freeman, Richard "Demand For Education" Chapter 6 in Orley Ashenfelter and Richard Layard (editors) Handbook of Labor Economics, North Holland, Vol I (1986), 357-386. Freeman, Richard and Lawrence P. Katz, "Introduction and Sirmmary," man and L. Katz (eds.). Differences in R. Free- and Changes in Wage Stmctures. Chicago, IL: The University of Chicago Press, pp. 1-22. Freeman, Richard and Remco Oostendorp "Wages Around Occupations and Countries" NBER Working Gottschalk, Peter and Timothy ings and the World: Paper No. 8058, (2000). M. Smeeding, "Cross-National Comparisons Income Inequality," Journal of Economic Literature 35 Gordon, Robert Pay Across of Earn- (1997), 633-687. "Current Productivity Puzzles from a Long-term Perspective" Northwestern University mimeo, (1998). Gottschalk, Peter and Joyce, Mary, "Cross-National Differences in the Rise in Earnings Inequahty: Market and Institutional Factors" Review of Economics and Statistics 80, (1998): 489-502 Gould, Eric D., Omer Moav, and Bruce A. Weinberg, "Precautionary Demand for Education, Inequality, and Technological Progress," Mimeo, Feb 2000. Greenwood, Jeremy and Mehmet Yorukoglu, "1974" Carnegie-Rochester Conference Series on Public Policy, XLVI (1997), 49-95. 32 Groshen, Erica, "Sources of Wage Dispersion: How Mucli Do Employers Matter," Quarterly Journal of Economics, 106 (1991), 869-84. Grout, Paul, "Investment and Wages in the Absence of Binding Contracts: A Nash Bargaining Approach," Econometnca, 52 (1984), 449-460. Hornstein, Andres and Per Krusell (1996), "Can Technology Improvements Cause NBER Productivity Slowdowns?," Jorgensen, Dale, W. and Kevin Macroeconomic Annual, J. Stiroh, "Raising the vol. 11, pp. 209-259. Speed Limit: U.S. Economic Growt.h in the Information Age" forthcoming in the Brookings Papers on Economic Activity (2000). Juhn, Chinhui, Kevin M. Murphy and Brooks the Rise in Returns to Skill" Jounial of Political Pierce (1993); "Wage Inequality and Economy Katz, Lawrence, and David Autor, "Changes in the Vol 101, pp 410-442. Wage Structure and Earnings Inequahty" in O. Ashenfelter and D. Card, eds.. The Handbook of Labor Economics Volume III, Amsterdam: Elsevier, 2000. Katz, Lawrence and Kevin Murphy, "Changes in Relative Wages: Supply and De- mand Factors" Quarterly Journal of Economics, Katz, Lawrence, Gary Changes (1992), 35-78. W. Loveman and David G. Blanchflower "A Comparison of OECD Countries" in Richard Freeman and in the Structure of Lawrence Katz op. CVII Wages in Four cit. Katz, Lawrence and Siunmers, Lawrence tions" Brookings Papers "Industry Rents: Evidence and Implica- on Economic Activity (Macroeconomics), (1989) 51, pp 209-90. Krueger, Alan B., and Jorn-Steffen Pischke "Observations and conjectures on the U.S. employment miracle," NBER Working Paper No. 6146, (1997). Krugman, Paul, "Part and Prospective Causes of High Unemployment" in Federal Reserve Bank of Kansas City Reducing Unemployment: Current Issues and Policy Options (1994). Mortensen, Dale, "Property Rights and Efficiency in Mating, Racing and Related Games." American Economic Review 72 (1982): 968-79. Miirphy, Kevin M. and W. Graig Riddell and Paul M. Romer, "Wages, Technology in the United States and Canada," in E. Helpman, 33 ed., Skills and General Purpose Technologies Cambridge, MA, MIT Press, (1998). and Brian Bell "The Collapse in Demand Unemployment Across the OECD" Oxford Review of Economic Policy Vol Nickell Stephen pp For the Unskilled and 11, (1996) 40-62. OECD Jobs Study Volumes 1 and 2 (1994). Pissarides, Christopher, Equilibriiim Unemployment Theory. Cambridge, MA: Basil Blackwell, 1990. Rubinstein, Ariel "Perfect Equilibrium In a Bargaining Model" Econometnca, 50 (1982), 97-109. Shaked, Avner and Jolni Sutton, "Involuntary Unemployment as a Perfect Equilib- rium in a Bargaining Model" Econometnca, 52 (1984), 1341-1364. Welch, Finis, "Education in Production," 312-327. 34 Journal of Political Economy 78 (1970), — — ;v. in ON ON '-^ o CO — .^ CO CO ^~ ^ — — rr- p ^ o ^ ^ ^ CD o H. ^ " aj CO n' S IN^ "O c C S^ c^ r-~ '=! h- CO T— CD ONl h- o ^ 00 CO - CO u u •" -a c^ OJ 5 (D 1_ Lu ^ C^J CD CD CD CO CO r^ CNI r^ o T— ^ T3 -^ CD CO ^— ON ^ m " O ON o <U CJ *- O "O (U On u— c r3 o ON On a. aj v7 -S^r^ c O 2; 1§ o m 05 CD .^ CD CNJ r^ r- 00 oq i_ 1- „ • ac CO ^^2 5 V >>,!" a -S "O .<D CNJ til O (_> On 'y^ m ON o o _- ^ ^ Co as Ln CNJ CD in CD o -* I/, On U QJ -o On n> Jai L^ oo ON o ^ M Ol rn :§ "^ i> r^ c^ ^- B cj « p Q. ^ CJ) CD CD o CD ^ o CX) ''~ CnI CD 00 CO CO ^ CD 'a- CO ''" ''" r\i CNJ ^ o CO 00 ''~ ^ m in '^~ c c^ C3 UJ i ^ i '^~ -o a> (^ — ^' Lm (L> S « -o CNJ CNJ CD 00 r^ •^^ !+-' u-, ty^ -a _j oc (t) on cvt & ^— CO r^ in in CJ) oq <i- CD 00 CO T CD ^ CD o ^ •^ C ^- O CN CNJ CD CO ^ c — o c ^U ,P <;j; ^ o cd o CNJ CJ) (U ^ o o CO ^ i c 3 "v^ C/5 On DO t^ CO r- CD t-- T ^ CO T— CD (J) CD 00 ^~ E •" CD CO ro CD T ^ n o r^ CNJ CNJ ^ CD "^~' ~ c/:) LU c O ^ o 00 CO CO t CD 00 r- ^ en On oq ^~ CO CD CD O) 1-' '^~ '^~ T-' o ^ rt — O " o OO CD oo CNI 00 tq hCD °9 r- o CO CO in CJ) CNJ o 3 c t/5 3 < CO (L> c3 CQ CJ J3 e (D Q -a c c J2 cl! c a 6 <u O c lU — c 1 L> Crt a> o 2 Z ^ oo 5 00 D D u- X) 5i j„ 5 """ i. ^ *" a3 ON — H<- _ " o ,p r-^ — o '•*-• >- fe -id OO OO ON rt Cl "S I" ^ .fD X I/) il> 3 OO K c « o . .-P On cu a ' S —m oo -o -C On c ~ « ra C3 ON <j=; d} c "O S ^ On ;« <S OO Oo ON E S t^ < D I— C S "^ in On 3 — o O iw 00 .2 <u tn ^- a ^ « '> O « <ii CD 1;; O ON ON — 1-. i2 I a. . o o .^ CO C o g- < 2 — oo O P On On On C/5 ca J^ X J= -a -4-» H CO ^ o CO ^H cS Xi -o O c o H O (U c/5 i= ,(L> i: 1 x: -J 9 ^ „ -a as ON ^5 ta — C/) C |2 oo U H 2.H o) oa Oo g 2i — cj ^ a° h^ H) — — — ^ ~ CO CO ^ CvJ rt ^ CD ^ rCD ^ ^ CM CD "^ CD eg Oi QO O) •^ CO r- CO Csl ^ CD r^ V5 ,_ Q. <- .^ Q n z: '_ S"; C/) X t- ^/ Q n n OJ T-J =/) rri cx> Csj OJ -^ d §- - K X -— ^ ^' -Q io ) C/) :S -a * '"" u c/l CO (U — CA -^. CD <t ri? ' C\l CL CO 00 LD -^ CD CM CD in V- CD CD ^ ^ CO CO X h- T00 CSI r^ in r^ 'd- CO CD 3 u^ V} ID o o au > ^ a .^ ri^ S ^ a> in r^ T— CSi O) ^ ^ in CM CM ^ 00 in -tr CSI T CD CO CD in n: OJ so jj u. "0 -a on c "5 £ C/i -a a. II CO "^ c^ ^§ CM in CN CO CNJ CO CO CD rin CO OJ ^ CD in 3 0. 0- c n _c n OJ i> OJ TD a :3 !U ID JZ T3 CD Qi r3 OJ IT) (U ^ i3 o + > E U to 2 —J 1 c^ ^ > ~ n ^ — JZ vO CO C3 ^ ^* -^ 00 o ? CO -^c^ r-- 'd- CO CX) 500 CM til <i> in ^ ^ in -n *-i i-^ "U ^J 'c3 f > =r .= i^n <!> + -p c^ CO en aa ^m Ol ^ ^ T-H 1^ ^ ^- S C\J '^r CO CM 00 in ?3 io £1 H (1> T— T > CO CO CJ5 CO h~ CM r-- CO CD CD CO CSI en h- ^ (J) T— CO ^ m CO E ^ t^ ;5 2 OJ c ^ Q. H 3 C3 P ^ <!> L-. F L. o t*- Q-) a <" <u c= -3 (U U c3 o CO a. ^ O C o — --> CO bi) — 'I- un i:; S o S = CTJ _u _!. O 03 CO = + o p o n. o- CO <1> L^ •o c rcl in CO C3 C3 B (U Q. m LO 00 CD CO CO CO CO CO ^ in 00 CO CSI CM CM '^ CO CD CO ^ in ^ Q CO '^t; L> .— i2"*- S <u >" " 5^ U ra _ ° c n ^ I- « ? O <U a> tC o c a. s (U 0J3 r3 -> CD 00 C\l zr 15 CSI CO 5 LD CD CM in CO CO CM h^J- CO CD CM r^ 00 CO CM a> in CSI _c C " « Si (a y &> -a « ca ^ c^ ^§ CO in CSI CD r^ CO (O c» CO CO in CO CSI " CM CD CO J CJJ o 500 r- CSI CO CD in in r^ Csl CNl m 1 3 < E 3 [3) c a E c <u CU <u CQ u Q c -a c c iJu (U E <L> 1 !». c/5 •T3 OJ 2 z -a E T3 >> U 5 00 -0 DO -C l- c 5 D •'"> ca OJ CO — 00 f- 2 to CO 0 CO ^ I i oi_ OJ OJ > (U c *- 00 c =0 CO OJ > og a> 2 5 00 II lU -a C3 to o 3 C to z5 O " o o 3 Bc n 00 tL> c/1 . c2 c CJ zz * E -a a. 3 3 c2 c 3 E 3 CJ c to 0. .2 y a 3 g u C/3 _-^ 00 to 03 ^ OJ 00 2 « 00 NCD ^ u > a o « C/3 'I- a> F tO 00 00 c o 00 OJ •3 10 "3 Relative supply Skill premium of skills European supply Q)US-Post shift coP^^ and E Relative co^-p^^' demand for skills SkilPbiased tech. change or increased trade H/L Figure 1: Differential inequality trends HVL' due to 35 differential relative supply changes. Relative supply Skill CO premium of skills US-Post Institutional wage-setting E (oP^= and line for Europe coE-P°st Relative demand or skills Skill-aiased tech. change or increased trade Relative unemployment in Figure 2: Europe Differential inequality trends because of differences in labor institutions. 36 market ill Australia # - observed — ^— 1 1979 United Kingdom Figure .3a: 19B6 - Actual observed skill I Canada - observed premium vs. predicted skill togw —4 predicted logw, eos-2 vs predicted predfcted logw eos=2 skill premium pfedided togw. eos=2 1 1 1991 vs. predicted skill 1995 premium premia and predicted and a 37 skill = 2. premia from equation (7) for cr = 1.4 — predicied O — prwJided togw eo5-2 logw e<»=? predicted tog-v eos== T- -T 1983 Belgium - observed B—— predrcted logw vs. predicted skill premium — predicted logw, Netherlands - 1987 199J 1991 observed vs. predicied skill premium A^— predicted logw eos=2 eos=2 eos= _fc ? 1991 Finland - observed Figure 3b: Actual vs. predicted skill skill 1995 1992 Sweden obsen/ed premium - premia and predicted and cr 38 skill = 2. vs. predicted skill premia from equation 1995 premium (7) for a = 1.4 - predided logw. eos=2 1992 Denmark- observed vs. predicted A Israel - observed vs predicted Figure 3c: Actual skill skill —B 1995 - piedided piedicted itugw 1997 Germany premium observed predided logw, eos=2 skill logw. eos=Z bos=-14 vs. predicted ft premium Norway premia and predicted and a 39 skill = 2. - observed skill premium predided logw. eos=2 vs. predicted skill premia from equation premium (7) for a = lA " Q A —a ptedicled logw. eos=2 pcedicied logw. txa=2 predidwi logw eos==1 4 1985 Australia » - observed 19V9 1986 - Figure 4a: Actual observed skill Canada premium —^—predicted logw United Kingdom 1»4 1989 vs. predicted skill logw. 1»1 vs. predicted skill - observed vs- predicted skill 1997 premium eos=2 1995 premium premia and predicted and a 40 skill = 2. premia from equation (9) for cr = 1.4 —A 1955 Belgium - observed —— - observed Figure 4b: Actual predided logw eos-2 —ft 1988 vs. predicted —4 Finland predicled logw eos=2 1991 skill Netherlands - observed Sweden premium premia and predicted and (7 41 skill = 2. - observed 1994 vs. predicted skill a predicted logw eos=2 vs. predicted skill skill premium . I premium pfBd»cted logw eos=2 vs. predicted skill premium preinia from equation (9) for (7 — 1.4 A — — predicted predicled iog/i. eos=2 —ft pretfKietf logw* eQS=2 logw eos==l 4 199! 1997 1995 1997 year Denmark - observed vs. predicted skill - pfedided Israel - observed vs predicted Figure 4c: Actual skill skill Gemiany premium k>gw, - observed vs. predicted eos=2 skill premium preilicied togw, premium Norway premia and predicted and (7 42 skill = 2. - observed vs, predicted skill premia from equation eo5=2 premium (9) for a = 1.4 iMo Date Due MIT LIBRARIES 3 9080 02246 2342