Abstract
The study examines the roles of experience and education in explaining the increase in wage inequality among Philippine male workers between 1988 and 1995. It also provides a methodological approach to the analysis of wage inequality by combining non-parametric methods with semiparametric additive models, using the variance accounting framework. Non-parametric density estimators allow flexibility in dealing with distributional inference while additive models yield marginal effects estimates under minimal assumptions on the functional specification of the wage–schooling and wage–experience relationships. The results show that much of the inequality increase from 1988 to 1995 was caused by greater variabilities in returns to schooling and experience among 1995 workers. The rise of the p90/p10 percentile ratio was caused by greater return variabilities on schooling and experience in the 90th percentile.
Notes
1 Lemieux (Citation2002) reports that empirical findings support the convexity of schooling in relation to its effects on earnings.
2 In instances wherein secondary computations require scalar values, purely non-parametric regression functions need to be replaced by their semiparametric counterparts that combine parametric and non-parametric features.
3 For interesting model contrasts within the non-parametric paradigm, see Zheng (Citation2000) who used local constant non-parametric regression and Ginther (Citation2000) who employed local linear regression and quantile methods.
4 A useful collection of tools can be implemented using STATA. Stephen Jenkins (Citation1999) wrote a program for decomposing inequality measures.
5 Consistent tests for parametric quantile regression functions have been developed and applied by Bierens and Ginther (Citation2001) in wage specification analysis. In explaining rising wage inequality, Gonzales and Miles (Citation2001) validated quantile wage functions for Uruguay.