Compensation Comparisons Between the Public and Private Sectors in the United States: A propensity matching approach

This study attempts to find evidence on the extent of the rent paid to public sector workers in the United States by comparing human capital returns between the public and private sector, controlling for selectivity bias. Using information from the Health and Retirement Study (HRS) and the Current Population Survey (CPS) for the period 1992–2000, I apply propensity score matching methods in order to obtain workers from both sectors comparable in terms of their propensity score or conditional probability of working in the public sector. The HRS showed a wage premium paid to public sector workers ranging from 3.5 percent to 11.1 percent when controlling for union status, and from −1.6 percent to 6 percent without controlling for union status. This difference is attributed to the portion of the rent public sector workers obtain through unions. The wage premium increases with the level of education, but declines for workers with a college degree or more, indicating a relatively higher compression of the distribution of wages in the public sector. Wage differentials widen when adjusted for nonwage pecuniary benefits, with public sector workers enjoying relatively better jobs. In the presence of selectivity bias, the wage premium seems to be underestimated. The HRS data also showed a high level of homogeneity indicating that there are no significant differences in means between the two sectors in most of the variables considered in the study. This result facilitated the estimation of the propensity score and provided more and better matches, regardless of how the treated and the control groups are defined. The CPS showed a high level of heterogeneity, with the individuals showing such deep and irreconcilable differences that it was impossible to find appropriate matches. Consequently no clear or reliable conclusions were obtained. This suggests that the CPS may not be suitable for the wage comparison across sectors, at least not using matching methods. Since the HRS does not disaggregate by level of government, and given the poor results obtained from the CPS, it was unable to be determined whether the wage premium decreases with the level of government.

Labor Market Segmentation (D437400)

12662