Portal de Periódicos da CAPES

Human Capital and Regional Wage Gaps

2011; Routledge; Volume: 46; Issue: 10 Linguagem: Inglês

10.1080/00343404.2011.579092

1360-0591

Enrique López‐Bazo, Elisabet Motellón,

Regional Development and Policy

Abstract López-Bazo E. and Motellón E. Human capital and regional wage gaps, Regional Studies. This paper uses micro-level data to analyse the effect of human capital on regional wage differentials. The results for the set of Spanish regions confirm that they differ in the endowment of human capital, but also that the return that individuals obtain from it varies sharply across regions. Regional heterogeneity in returns is especially intense in the case of education, particularly when considering its effect on the employability of individuals. These differences in endowment and, especially, in returns to human capital account for a significant proportion of regional wage gaps. López-Bazo E. and Motellón E. 人力资本及区域工资差距,区域研究。本文利用微观数据分析了人力资本对于区域工资差异的影响。西班牙区域研究结果证明了人力资源禀赋方面存在的差异,同时个体回报率也存在巨大的区域差异。回报率的区域异质性在教育领域表现尤为明显,特别在考虑其对个体就业率的影响时。秉赋上的差异尤其是人力资本回报率差异构成了区域工资的显著差异。 教育 经验 区域差异 人力资本回 报率 工资差异分解 López-Bazo E. et Motellón E. Le capital humain et les écarts des salaires régionaux, Regional Studies. Cet article emploie des microdonnées afin d'analyser l'impact du capital humain sur les écarts des salaires régionaux. Les résultats pour l'ensemble des régions espagnoles confirment qu'elles diffèrent par la dotation du capital humain, mais aussi que le rendement qu'en obtiennent les individus varie sensiblement suivant la région. L'hétérogénéité régionale des rendements s'avère particulièrement vive pour ce qui est de l'éducation, notamment quand on tient compte de ses effets sur l'employabilité des individus. Ces différences de la dotation et, surtout, du rendement du capital humain expliquent une proportion non-négligeable des écarts des salaires régionaux. Éducation Expérience Écarts régionaux Rendements du capital humain Décomposition des écarts des salaires López-Bazo E. und Motellón E. Humankapital und regionale Einkommensunterschiede, Regional Studies. In diesem Beitrag wird mit Hilfe von Daten auf Mikroebene die Auswirkung von Humankapital auf regionale Einkommensunterschiede untersucht. Die Ergebnisse für die untersuchten spanischen Regionen bestätigen, dass sich die Regionen hinsichtlich der Ausstattung mit Humankapital, aber auch hinsichtlich der von den Personen dafür erzielten Erträge untereinander drastisch unterscheiden. Im Bildungsbereich fällt die regionale Heterogenität der Erträge besonders stark aus, vor allem wenn man ihre Auswirkung auf die Beschäftigungschancen der Personen berücksichtigt. Auf diese Unterschiede bei der Ausstattung und insbesondere bei den Erträgen des Humankapitals lässt sich ein erheblicher Anteil der regionalen Einkommensunterschiede zurückführen. Bildung Erfahrung Regionale Disparitäten Erträge aus Humankapital Dekomposition von Einkommensunterschieden López-Bazo E. y Motellón E. Capital humano y diferencias salariales por regiones, Regional Studies. En este artículo utilizamos datos a nivel micro para analizar el efecto del capital humano en las diferencias salariales por regiones. Los resultados para el grupo de las regiones españolas confirman que existen diferencias en cuanto a la dotación de capital humano, pero también varían en gran medida los beneficios que las personas obtienen a partir de este capital en las diferentes regiones. La heterogeneidad regional de los beneficios es especialmente aguda en el caso de la educación, en concreto cuando se considera qué efecto tiene para la capacidad de inserción laboral de las personas. Estas diferencias en las dotaciones y, especialmente, en los beneficios del capital humano representan un porcentaje significativo de las diferencias salariales por regiones. Educación Experiencia Desigualdades regionales Rendimiento del capital humano Descomposición de las diferencias salariales Keywords: EducationExperienceRegional disparitiesReturns to human capitalWage gap decompositionJEL classifications: C24J31R11R23 Acknowledgments The authors acknowledge financial support from the Spanish Ministry of Science and Technology, National Program of R&D, ECO2008-05314/ECON, and from the European Community's Seventh Framework Programme (FP7/2007-2013) under Grant Agreement Number 216813, IAREG Project. The views expressed here are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission. Notes An alternative is to estimate the effect of human capital on firms' productivity. However, the lack of firm-level data from a representative survey for each Spanish region prevented the authors from considering this approach. In any case, under well-known assumptions, the marginal productivity explanation of wage determination establishes the link between wages and productivity. The assessment of the return to human capital based on the estimation of a wage equation is standard in the labour market literature. The ECHP has frequently been used in wage studies for the Spanish labour market and for other European Union Member States (Montuenga et al., Citation2003; Rodríguez-Pose and Vilalta-Bufí, 2005; García-Pérez and Jimeno, 2007). Although the Earnings Structure Survey (a data set also produced in European Union countries under the auspices of EUROSTAT) contains the most complete information on wages and workers, jobs' and firms' characteristics, it does not provide information on the non-employed. This prevents one from controlling for sample selection and computing the indirect effects of human capital on wages through its effect on the probability of employment, which is one of the objectives of this paper. The regional representativeness of the sample for the entire panel of the ECHP is only guaranteed at the NUTS-I level, which corresponds in Spain to an artificial grouping of regions based on geographical criteria alone. In any case, individuals working fewer than fifteen hours a week were removed from the sample, given that in this case the ECHP does not provide information on some variables that are important for the analysis (for example, tenure). It should be mentioned that the results are robust to the exclusion from the sample of those individuals working fewer than thirty hours a week. This information was kindly provided by the Catalan Institute for Statistics (IDESCAT), which estimates the parity power standards for the seventeen Spanish regions from the aggregate Spanish figures used by the Statistical Office of the European Union, EUROSTAT, to produce data net of the cost-of-living differences across the Member States. Note that, given the common currency for the spatial units under analysis, parity power standards only account for differences in the cost of living. It can be argued that jobs' and firms' characteristics also differ across regions. And as far as wages vary within these characteristics, the composition effect should include them as well. However, here the focus is on individuals' characteristics, given the authors' interest in the effect of human capital. In any case, a great deal of the wage variability associated with different jobs and firms is likely to be captured by differences in workers' human capital if there is a process of sorting across jobs and firms depending on the endowment of human capital. Results for the seventeen regions are not reported here to save space, though they are available upon request. Note that, as is usual in this type of analysis, a simple specification of the Mincerian wage equation is used to obtain a better insight into the global effects of the human capital variables on wages (Pereira and Silva, Citation2004). For the full list of characteristics included in X and in Z, see Table 2. For further details, see the fourth section. It must be noticed that this type of selection process does not distinguish between non-participants and unemployed individuals. It is possible to design a two-step sequential selection procedure for the decision of participation and then for employment for those participating (Arrazola and De Hevia, Citation2008, considered this type of selection process in their study of the returns to schooling in Spain). However, it was decided here not to use such a process due to the low number of observations for some of the categories in some regions and, above all, given the authors' interest in the derivation of a detailed decomposition of the regional wage gaps. In any case, results reported by Arrazola and De Hevia suggest that the estimate of the returns does not vary markedly when the selection process distinguishes between non-participants and the unemployed. For the derivation of the expressions and the discussion of these marginal effects, see Greene Citation(2003) and Cameron and Trivedi Citation(2005). Hoffmann and Kassouf Citation(2005) and Arrazola and De Hevia Citation(2008) used these expressions to compute different types of returns to education. As usual in the specification of wage equations, a quadratic form is used for experience: βEXP·EXP ir + βEXP 2·EXP2 ir As a result, the return to experience (conditional and unconditional) is: βEXP + 2·βEXP 2·EXP ir In the case of tenure, its return will be measured by the estimation of the coefficients of each of its categories. Despite the arguments suggesting the overestimation of the returns to schooling based on the OLS estimator, the conclusion from the results in the literature based on IV estimates is that the causal effect of education is as big or bigger than the OLS estimated return. García et al. Citation(2001) also used a similar instrument for their analysis of the gender wage gap in Spain. Results from the Sargan test were clearly against the exogeneity of the inverse Mills ratio. This is an important difference when comparing the results for the entire country with those reported by Arrazola and De Hevia Citation(2008), as they did not take into account that the inverse Mills ratio is a function of human capital, and thus considered it as an exogenous regressor. The authors thank an anonymous referee for raising that point. The authors thank an anonymous referee for raising this concern. A two-step procedure was implemented in STATA 11. In the first step, the inverse Mills ratio was estimated by using the command HECKMAN. In a second step, the IVREG2 routine (Baum et al., 2010) was used to estimate the wage equation by TSLS, considering schooling and the inverse Mills ratio as endogenous regressors. The full set of estimated coefficients for the wage and the selection equations, for each of the seventeen regions and for Spain, are available from the authors upon request. These results are available from the authors upon request. Notice that in what follows it is assumed that the no-discrimination wage structure is that in region A. Notice that and are the weight in the standard linear decomposition. It is impossible to summarize the results for the decomposition of the wage gap for all pairs of regions (17*16*0.5 = 136) in this type of publication. An alternative to that in this study is to consider the gap with regards to a benchmark region (for instance, that with the highest average wage), although this is subject to the criticism of the selection of the benchmark and slightly complicates the comparison of results across regions. In any case, the qualitative conclusion on the important contribution of regional differences in the return to human capital (in particular to schooling) is also obtained when a benchmark region is used to compute the gaps. These results are available from the authors upon request.