- Public Sector Wages and Bureaucratic Quality:Evidence from Latin America
Public sector employment is commonly perceived as being less demanding than private sector employment. In many countries, the public sector offers a noncompetitive work environment and a level of job security that cannot be matched in the private sector. Quinn, for instance, finds that private sector workers tend to be subject to more workplace disamenities (such as pace of work, supervision, and workplace haz-ards) than their public sector counterparts.1 Bellante and Long show that the fringe benefits enjoyed by public sector workers are substantially larger than those offered in the private sector.2 Poterba and Rueben show that in 1993 benefits averaged 43.8 percent of wages for U.S. public sector workers and 40.3 percent for U.S. private sector workers.3 At the same time, it is commonly believed that these advantages of public sector employment are offset by lower average wages. This situation is summed up in a joke in which employees of state-owned enterprises in the former Soviet Union describe their employment arrangement as follows: "We pretend to work, and they pretend to pay us." This does not actually seem to [End Page 97] be the case, however. Mounting evidence indicates that in many countries, wages paid in the public sector are not lower and are often significantly higher than wages paid in the private sector.4 In essence, there seems to be a public sector wage premium.
The purpose of this paper is twofold. First, the paper presents evidence on public-private wage differentials for a sample of seventeen Latin American countries covering 88 percent of the total population of the region. Second, the paper discusses the relationship between the public-private wage differential and the quality of the public sector and presents evidence for a sample of Latin American countries.
The first part of the paper aims at documenting public-private wage differentials for a sample of Latin American countries over the 1980-98 period. While there exists a large literature testing for the presence of a public sector wage premium in the United States (Ehrenberg and Schwarz survey twenty-three studies that support the presence of a public sector wage premium), very little work has been done at the cross-country level.5 One exception is Blanchflower, who estimates public-private wage differentials for fifteen member countries of the Organization for Economic Cooperation and Development (OECD). He finds that eleven of the countries have a positive and statistically significant public sector wage premium, while only one (Norway) has a statistically significant public sector wage penalty.6 Gregory and Borland, who survey more than thirty-four studies, find that the public sector wage premium is high for women, but it is often not statistically significant for men.7 Panizza and Qiang find a significant public sector premium for both men and women in several Latin American countries.8
One interesting issue is the correlation between the wage premium and skills. Katz and Krueger run separate regressions for U.S. workers belonging to different educational groups and show that college-educated males face a public sector penalty, while males with lower education enjoy a public sector premium. (Similar findings are obtained for female workers).9 [End Page 98] Quantile regression analysis applied to U.K. data reveals that the public sector premium is inversely related to an employee's position in the distribution of earnings.10 Similar results are found for Australia and Sweden.11 The more concentrated earning distribution in the public sector may have an important selection effect. If the sorting of employees between the public and private sectors is due to unobserved productivity-related characteristics, ordinary least squares (OLS) estimations of the public-private wage differential could yield biased results.12 In van der Gaag and Vijverberg's study of public-private pay differentials in the Ivory Coast, OLS estimates yield a large public sector premium, but this premium disappears when the authors control for the selection bias.13
Even with this last caveat, the existing empirical literature on public-private wage differentials can be summarized by the following three stylized facts: a rent is...