The unequal distribution of the public-private sector wage gap in Greece: evidence from quantile regression

Abstract

Quantile regression analysis is used to estimate the public-private sector wage differential in Greece. The results suggest that wage differences between sectors are mainly attributed to the employee's endowment. The decomposition of the wage differential shows that the endowment component (characteristics differential) increases as we move up to the upper quantiles and the unobserved components decrease at higher quantiles. (The views expressed in this paper are those of the author and not those of the Bank of Greece.)

Notes

¹ For detailed surveys on the wage differentials among the private and public sectors see, among others, Ehrenberg and Schwarz (1986) and Bender (1998, 2003).

² The studies of Kanellopoulos (1997) and Kioulafas et al. (1991) examine, applying ordinary least square analysis, the wage differentials between the public and private sectors in Greece.

³ Christofides and Pashardes (2002) have applied the Oaxace and Ransom (1994) decomposition technique to examine the wage differentials between the public and the private sector in Cyprus in the context of Ordinary Least Square equations.

⁴ In particular, the independent variable of occupation consists of eight dummy variables for nine different occupations. Full-time employment is a dummy variable, which takes the value one if the employee is employed full-time and zero otherwise. Finally, the variable for skills is a dummy variable, which takes the value one if the employee believes that he/she has more skills or qualifications to do a more demanding job than the one he/she has now and zero otherwise.

⁵ Separate wage equations for public and private sector for both genders and for each one of them separately have been estimated. All the control variables included in the wage equations for the total sample by sector of employment are statistically significant at different levels of significance with the appropriate sign. However, the size of the estimated coefficients differs between the two sectors.

⁶ In addition the wage equations were performed including a selection variable (inverse Mills ratio). The selection variable was negative in all the estimated regressions and was statistically significant at 5% level of significance in the private sector wage equation. In the estimated wage equations by gender the selection variable was significant in the male equations for both sectors and not significant in the female equations for both sectors. However, the non-significance of the estimated coefficient of the selection variable indicates that this variable is not necessary to be included in the wage equations. Therefore, in the present analysis the wage equations are estimated without the inclusion as explanatory variable the inverse Mills ratio.