Performance-related Pay in the Public Sector : A Review of Theory and Evidence
The objective of this paper is to provide a review of the theoretical and, in particular, empirical literature on performance-related pay in the public sector spanning the fields of public administration, psychology, economics, education, and health with the aim of distilling useful lessons for policy-makers in developing countries. This study to our knowledge is the first that aims to disaggregate the available evidence by: (i) the quality of the empirical study; (ii) the different public sector contexts, in particular the different types of public sector jobs; and (iii) geographical context (developing country or OECD settings). The paper's main findings, based on a comprehensive review of 110 studies of public sector and relevant private sector jobs are as follows. First, we find that overall a majority (65 of 110) of studies find a positive effect of performance-related pay, with higher quality empirical studies (68 of the 110) generally more positive in their findings (46 of the 68). These show that explicit performance standards linked to some form of bonus pay can improve, at times dramatically, desired service outcomes. Second, however, these more rigorous studies are overwhelmingly for jobs where the outputs or outcomes are more readily observable, such as teaching, health care, and revenue collection (66 of the 68). There is insufficient evidence, positive or negative, of the effect of performance-related pay in organizational contexts that that are similar to that of the core civil service, characterized by task complexity and the difficulty of measuring outcomes, to reach a generalized conclusion concerning such reforms. Third, while some of these studies have shown that performance-related pay can work even in the most dysfunctional bureaucracies in developing countries, there are too few cases to draw firm conclusions. Fourth, several observational studies identify problems with unintended consequences and gaming of the incentive scheme, although it is unclear whether the gaming results in an overall decline in productivity compared to the counterfactual. Finally, few studies follow up performance-related pay effects over a long period of time, leaving the possibility that the positive findings may be due to Hawthorne Effects, and that gaming behavior may increase over time as employees become more familiar with the scheme and learn to manipulate it.