Skip to main content

Decomposing wage discrimination in Germany and Austria with counterfactual densities

Empirica volume 42, pages 49–76 (2015)Cite this article

Abstract

Using income and other individual data from EU-SILC for Germany and Austria, we analyze wage discrimination for three break-ups: gender, sector of employment, and country of origin. Using the method of Machado and Mata (J Appl Econom 20(4):445–465, 2005) the discrimination over the whole range of the wage distribution is estimated. Significance of results is checked via confidence interval estimates along the lines of Melly (Estimation of counterfactual distributions using quantile regression. Working Paper, SIAW, University of St. Gallen, 2006). The economies of Germany and Austria appear structurally very similar and are highly interconnected. One would, therefore, expect to find similar levels and structures of wage discrimination. Our findings deviate from this conjecture significantly.

This is a preview of subscription content, access via your institution.

Access options

Buy single article

Instant access to the full article PDF.

USD 39.95

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9

Notes

  1. see http://epp.eurostat.ec.europa.eu/tgm/table.do?tab=table&plugin=0&language=de&pcode=tsdsc340.

  2. A major rational for this sample restriction is the removal of self-selection bias. But, as our results on self-selection in the “Appendix” will show, filtering along the gender dimension is most likely insufficient for that purpose.

  3. In the econometric literature dealing with decomposition this discriminatory part is called structural effect, whereas the part associated with different characteristics is known as composition effect. We will keep using the terms “discrimination” and “explained differences” instead.

  4. A more thorough discussion of the shortcomings of the OB-decomposition is found e.g. in Fortin et al. (2011).

  5. Similar ideas are found in Gosling et al. (2000), Albrecht et al. (2003) and Melly (2005a). Testing with these approaches only yielded marginally different results relative to those of MM and are not reported here.

  6. Results using RIF-regressions with reweighting were not found qualitatively different from the MM-results. These results are available from the authors upon request.

  7. We found that the number of bootstrap samples S required to get stable results should be a multiple of the total number of observations. For the application we have chosen S = 40000 which is roughly four times the n 1 + n 2 number of observations in the case of Germany and eight times in the case of Austria. With this number of bootstraps the differences between the MM approach and Melly (2005b) are negligible for practical purposes.

  8. Formulated as a programming problem, quantile regression coefficients β(τ) for quantile τ are estimated as solution to min β(τ) (1/n)∑ i ρ τ [w − x i β(τ)] with ρ τ (u) = τ u for u ≥ 0 and \(\rho_\tau(u)=(\tau\!-\!1)u\) for u < 0. We use the R-package quantreg by Roger Koenker for that purpose (see Koenker 2012).

  9. Step 2 (b), by the probability integral transformation principle, simulates random sampling from the (estimated) conditional distributions of w ki conditional on X k , for k = 1,2. Or, put differently: The w ki consistently estimate the corresponding quantiles of the conditional distribution, see Koenker and Bassett (1978). Repeating these quantile estimates for S random draws of characteristics from the original distributions then amounts to integrating out these characteristics from the corresponding conditional distributions.

  10. For more details see Machado and Mata (2005). A formal proof of consistency and asymptotic normality of the derived difference measures is contained in Albrecht et al. (2009).

  11. Numerous non-basic counterfactual distributions can be imagined and found in the literature (see Cahuc and Zylberberg 2004 pp. 280–282 for a short discussion). For example, one based on fictitious non-discriminatory market remuneration coefficients β m for both groups. Such non-basic counterfactuals are not considered here.

  12. Melly provides a corresponding R-source code on http://www.econ.brown.edu/fac/Blaise_Melly/code_R_rqdeco3.html.

  13. Relevant approaches are found e.g. in Buchinsky (1998), Albrecht et al. (2004), Neuman and Oaxaca (2004a) or Ivanov (2008).

  14. This latter criterion may potentially introduce another type of sample selection bias, as it ignores different likelihoods of longer unemployment spells for each subgroup considered. See Sect. 3.5.

  15. Unfortunately, the understanding of these education levels has been different in Germany and Austria. This explains the implausible, massive differences in the proportions of these three levels between the two countries (see Tables 2 and 4 in the “Appendix”). This prohibits comparing the estimated standard quantile regression coefficients for these variables between countries. To our knowledge, statistical offices are aware of the corresponding shortcomings and currently work on improved definitions and comparable coding.

  16. Using “age”, “age2” and “experience” instead of “age”, “experience” and “experience2” lead to a worse fit and was formally rejected by corresponding tests.

  17. This is a fairly standard result and easy to interpret: Negative values for the bottom percentiles arise naturally, if the lowest incomes are associated with manual labor, which deteriorates in quality with age. Positive values for higher incomes simply reflect widespread seniority pay.

  18. Synonymously we will speak of overall wage differences or raw discrimination.

  19. Fournier and Koske (2012) report a difference of 25 % at the median (Fig. 7) where we find 20 %. One reason for this difference might be that we classify all persons born abroad as immigrants, while Fournier and Koske count only those born outside the EU. Furthermore, their underlying regression specification is not quite clear. The basic data set instead is the very same as used here.

  20. These results are unweighted and, thus, are not strictly comparable to the OLS results from above. But, as the comparison of total wage gaps reported in Tables 7 and 11 shows, the differences are quantitatively negligible.

References

  • Albrecht J, Björklund A, Vroman S (2003) Is there a glass ceiling in Sweden? J Labor Econ 21(1):145–177

    Article  Google Scholar 

  • Albrecht J, van Vuuren A, Vroman S (2009) Counterfactual distributions with sample selection adjustments: econometric theory and an application to the Netherlands. Labour Econ 16(4):383–396

    Article  Google Scholar 

  • Albrecht JW, van Vuuren Aico, Vroman S (2004) Decomposing the gender wage gap in the netherlands with sample selection adjustments. IZA Discussion Paper No. 1400, Institute for the Study of Labor, Bonn, November 2004

  • Altzinger W, Berka C, Humer S, Moser M (2012) Die langfristige Entwicklung der Einkommenskonzentration in Österreich 1957–2009. Teil 2: Methodik und Ergebnisse. Wirtschaft und Gesellschaft 38(1):77–102

    Google Scholar 

  • Arulampalam W, Booth AL, Bryan ML (2007) Is there a glass ceiling over Europe? Exploring the gender pay gap across the wages distribution. Ind Labor Relat Rev 60(2):163–186

    Google Scholar 

  • Blinder AS (1973) Wage discrimination: reduced form and structural estimates. J Hum Resour 8(4):436–455

    Article  Google Scholar 

  • Böheim R, Hofer H, Zulehner C (2005) Wage differences between men and women in Austria: evidence from 1983 and 1997. IZA Discussion Papaer No. 1554, Institute for the Study of Labor, Bonn, April 2005

  • Böheim R, Hofer H, Zulehner C (2007) Wage differences between Austrian men and women: semper idem?. Empirica 34(3):213–229

    Article  Google Scholar 

  • Böheim R, Himpele K, Mahringer H, Zulehner C (2013) The distribution of the gender wage gap in Austria: evidence from matched employer-employee data and tax records. J Labour Mark Res, 46(1):19–34. ISSN 1614-3485. doi:10.1007/s12651-012-0113-y

    Google Scholar 

  • Buchinsky M (1998) The dynamics of changes in the female wage distribution in the USA: a quantile regression approach. J Appl Econom 13(1):1–30

    Article  Google Scholar 

  • Cahuc P, Zylberberg A (2004) Labor economics. The MIT Press, Cambridge

    Google Scholar 

  • Depalo D, Giordano R, Papapetrou E (2013) Public-private wage differentials in euro area countries: evidence from quantile decomposition analysis. Temi di discussione (Economic working papers) 907, Bank of Italy, Economic Research and International Relations Area, April 2013

  • DiNardo J, Fortin N, Lemieux T (1996) Labor market institutions and the distribution of wages, 1973–1992: a semi parametric approach. Econometrica 64(5):1001–1044

    Article  Google Scholar 

  • Dustmann C, Ludsteck J, Schönberg U (2009) Revisiting the german wage structure. Q J Econ 124:843–881

    Article  Google Scholar 

  • Firpo S, Fortin N, Lemieux T (2009) Unconditional quantile regressions. Econometrica 77(3):953–973

    Article  Google Scholar 

  • Fitzenberger B, Kunze A (2005) Vocational training and gender: wages and occupational mobility among young workers. Oxf Rev Econ Policy 21(3):392–415

    Article  Google Scholar 

  • Fitzenberger B, Wunderlich G (2002) Gender wage differences in West Germany: a cohort analysis. Ger Econ Rev 3(4):379–414

    Article  Google Scholar 

  • Fortin NM, Lemieux T, Firpo S (2011) Decomposition methods in economics. In: Ashenfelter O, Card D (eds) Handbook of labor economics, vol. 4A. North-Holland Elsevier, Amsterdam, pp 1–102

    Google Scholar 

  • Fournier J-M, Koske I (2012) Less income inequality and more growth—are they compatible?: part 7. The drivers of labour earnings inequality—an analysis based on conditional and unconditional quantile regressions. Economics Department Working Papers, No. 930, OECD

  • García J, Hernández PJ, López-Nicolás A (2001) How wide is the gap? An investigation of gender wage differences using quantile regression. Empir Econ 26(1):149–167

    Article  Google Scholar 

  • Gosling A, Machin S, Meghir C (2000) The changing distribution of male wages in the UK. Rev Econ Stud 7(4):635–666

    Article  Google Scholar 

  • Heinze A (2010) Beyond the mean gender wage gap: decomposition of differences in wage distributions using quantile regression. ZEW Discussion Papers, No. 10-043, Zentrum für Europäische Wirtschaftsforschung

  • Ivanov AV (2008) Immigrant discrimination in Germany? Quantile regression decomposition of the wage gap. CDSE Discussion Paper No. 41, Center for Doctoral Studies in Economics, University of Mannheim

  • Koenker R (2012) Quantreg: quantile regression. http://CRAN.R-project.org/package=quantreg. R package version 4.81

  • Koenker R, Bassett Jr G (1978) Regression quantiles. Econometrica 46(1):33–50

    Article  Google Scholar 

  • Machado JAF, Mata J (2005) Counterfactual decomposition of changes in wage distributions using quantile regression. J Appl Econom 20(4):445–465

    Article  Google Scholar 

  • Melly B (2005a) Decomposition of differences in distribution using quantile regression. Labour Econ 12(4):577–590

    Article  Google Scholar 

  • Melly B (2005b) Public-private sector wage differentials in Germany: evidence from quantile regression. Empir Econ 30:505–520

    Article  Google Scholar 

  • Melly B (2006) Estimation of counterfactual distributions using quantile regression. Working Paper, SIAW, University of St. Gallen, 2006

  • Mueller RE (1998) Public-private sector wage differentials in Canada: evidence from quantile regressions. Econ Lett 60(2):229–235

    Article  Google Scholar 

  • Neuman S, Oaxaca RL (2004a) Wage decompositions with selectivity corrected wage equations: a methodological note. J Econ Inequal 2:3–10

    Article  Google Scholar 

  • Neuman S, Oaxaca RL (2004b) Wage differentials in the 1990s in Israel: endowments, discrimination, and selectivity. IZA Discussion Paper No. 1362, Institute for the Study of Labor, Bonn

  • Oaxaca R (1973) Male-female wage differentials in urban labor markets. Int Econ Rev 14(3):693–709

    Article  Google Scholar 

  • Peters H (2008) Development of wage inequality for natives and immigrants in Germany—evidence from quantile regression and decomposition. SOEPpapers on multidisciplinary panel data research 113, Deutsches Institut für Wirtschaftsforschung, Berlin, June 2008

  • Picchio M, Mussida C (2010) Gender wage gap: a semi-parametric approach with sample selection correction. CentER discussion paper series 2010-16, Tilburg University

  • Pointner W, Stiglbauer A (2010) Changes in the Austrian structure of wages, 1996–2002: evidence from linked employer-employee data. Empirica 37(2):105–125

    Article  Google Scholar 

  • Poterba JM, Rueben KS (1995) The distribution of public sector wage premia: new evidence using quantile regression methods. Working paper nr. 4734, NBER

  • Weichselbaumer D, Winter-Ebmer RR (2005) A meta-analysis of the international gender wage gap. J Econ Surv 19(3):479–511

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Vienna University of Economics, Vienna, Austria

    Thomas Grandner & Dieter Gstach

Authors
  1. Thomas Grandner
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Dieter Gstach
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Thomas Grandner.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

PDF (113 KB)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Grandner, T., Gstach, D. Decomposing wage discrimination in Germany and Austria with counterfactual densities. Empirica 42, 49–76 (2015). https://doi.org/10.1007/s10663-014-9244-4

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10663-014-9244-4

Keywords

  • Wage discrimination
  • Decomposition
  • Quantile regression