Association for Public Policy Analysis & Management

Can we reduce inequalities using performance-based incentives? Although there has been some evidence that performance-based incentives could improve performance of job training agencies, healthcare services, and research institutes, empirical evidence of its’ influence on inequality is very limited. Recently, a growing number of universities have adopted performance-based incentive system to promote their ranking and reputation, and, between 2004 and 2012, the studied university has also increased financial incentives for researchers’ publications in SCI journals. This study examines effects of financial incentives on inequalities in research performance using a panel data of nine-year of 450 scientists in a Korean university. As financial incentives for SCI publications rose over time, inequalities among scientists reduced: the GINI index for average number of SCI publications per scientist declined from 0.75 in 2004 to 0.66 in 2009, and the GINI index for quality of publications which was measured by impact factors of journals also declined from 0.37 in 2004 to 0.28 in 2009. In addition, the share of publications explained by high 25^th-50^th percent ranked scientists has expanded while the share of publications by top 10 percent scientists has reduced, which meant that financial incentives had different effects on scientist with different performance-level. Fixed-effect OLS analyses support it: the regression coefficient of financial incentives on SCI publications was the highest with high 25^th-50^th percent scientists controlling for unobserved scientists’ characteristics. The model with top 25^th percent scientists did not show a statistically significant relationship between financial incentives and research performance. Thus, this study presents evidence that, if designed carefully, the performance-based incentives could improve equality of performance among individuals in nonprofit or public organizations, and different types of incentives should be considered for heterogeneous groups of individuals with dissimilar levels of productivity.