Managerial Incentives in Public Service Delivery: Evidence from School-based Nutrition Programs in Rural China

Abstract

Growing evidence indicates that weak or misaligned incentives facing providers pose a significant barrier to service delivery in many developing countries. To address weak supply-side incentives, performance pay and related approaches explicitly linking provider pay and performance have become increasingly common in public service delivery. Despite the growing prominence of these approaches, however, many conceptual issues surrounding the use of performance pay in this context remain unexplored. A fundamental one is the role of performance pay for managers in the organizations commonly tasked with delivering public services. Although a large literature examines performance pay for managers in private firms, much less is known about the use of performance pay for their counterparts in public service organizations. Improving public service delivery may nonetheless depend heavily on aligning the incentives of managers with social objectives. Drawing on a large-scale field experiment involving 300 primary schools in rural China, this dissertation explores how performance incentives for school administrators affect their implementation of new, school-based nutrition programs targeting anemia. School-based nutrition programs are an important function of schools, particularly in settings with less developed public health infrastructures. Weak incentives for schools to effectively implement these programs are compounded as these programs compete with more traditional functions for finite school resources. I report the results of this field experiment which was designed to test three main issues concerning the use of performance incentives for school administrators in this context. First, I study the effect of offering administrators performance pay contracts tied to reductions in school-level anemia prevalence. As part of the experiment, a subset of schools were randomly allocated to receive one of two levels of performance incentives for reductions in student anemia or to a no-incentive comparison group. I find that large incentives led to meaningful reductions while smaller incentives (10% of the size) were ineffective in reducing anemia. Further, I find that an important channel through which large incentives impacted student nutrition was by motivating administrators to engage households and influence feeding at home. I discuss the implications of this finding for the design of performance incentives tied to jointly produced outcomes. Second, I study the impact of providing administrators with more resources to implement a nutrition program and how this interacts with performance incentives. To test this, schools were orthogonally assigned to two levels of block grants within each level of performance incentives. I find that, absent explicit anemia-based incentives, increasing the size of block grants under the control of administrators led to sizable reductions in anemia prevalence but were nearly twice as costly as performance incentives. This impact was not purely the result of additional inputs; larger block grants also caused a more efficient use of inputs and an increase in effort devoted to reducing anemia. I also find that additional resources and incentives are substitutes in this context. I provide evidence that this substitution is due, at least in part, to incentives re-framing the task of implementing the nutrition programs from one that was part of the professional role of administrators to one that was not. Finally, I approach the health promotion and education roles of schools as a multi-tasking problem and use remaining experimental groups to examine how performance incentives for school administrators to reduce anemia and improve test scores each affect anemia prevalence and academic performance. Although the theory of multitasking is well-developed, there are few empirical studies testing this theory directly. I emphasize three main findings. First, incentives in the two dimensions (given in the context of an anemia reduction program) both led to significant reductions in anemia prevalence. Second, anemia-based and test-based incentives serve as substitutes in the direction of anemia reduction: providing administrators with both types of incentives did not lead to significantly larger reductions in anemia. Third, I find that anemia incentives caused an allocation of resources away from education 'inputs' but this did not lead to significantly lower student performance on standardized exams after one year. These results reflect that test-based incentives are well-aligned with improving nutrition, but anemia-based incentives are not well aligned with effort to improve academic performance. Strengthening incentives to improve academic performance while also emphasizing the relationship between good nutrition and academic performance may therefore be sufficient to motivate administrators to effectively implement school-based anemia reduction programs while causing less reallocation of resources away from education.