Government increasingly relies on complex arrangements of providers to deliver public services. Because many of these providers are private sector organizations (both nonprofit and for-profit) that deliver public services through government contracts, issues related to contract management and performance have gained prominence within the public administration literature. While research has demonstrated contract incentives drive performance (Heinrich 2002; Heinrich and Choi 2007), there is a dearth of literature on how incentives might subvert larger public policy goals. In services with difficult to measure outcomes, proxies such as processes and cost efficiency are frequently used to measure performance. Proxies may not correlate with the overarching goals of a public program.

Private contractors have been called the “new street-level bureaucrats” (Smith and Lipsky 1993; Van Slyke and Roch 2004). Thus, the inherent discretion afforded to contractors further complicates the production of publically funded services. Because of the assumed altruism motivating nonprofit organizations, scholars have conceptualized the relationship between government and nonprofit as one of principal-steward rather than principal-agent (Van Slyke 2007). Though this idea has gained traction recently, studies tend to be qualitative or small N, and more empirical evidence is warranted.

In this study, we examine contract performance in mixed markets (multi sector) for juvenile justice services funded by the state of Florida. In Florida, private organizations (nonprofit and for-profit) implement the majority of juvenile justice residential treatment programs on behalf of the Department of Juvenile Justice (DJJ). Organizations may provide multiple, independent programs. Discretion is afforded to individual programs about how to rehabilitate youth. Examples of program discretion include selection of specific treatment models for juveniles.

The goal of the DJJ is to reduce youth recidivism. The state tracks youth recidivism by program for the year after juveniles are released. However, the DJJ rewards and sanctions contractors based on “quality” of programs, though quality is determined using contract compliance rather than recidivism reduction. Low recidivism is not significantly correlated with the DJJ’s program quality scores. This mismatch between the incentives for contract maintenance/renewal and the overarching policy goal of reduced recidivism motivates our research questions: Do contractors prioritize contract compliance or the broader public value of reduced recidivism? Are there differences between nonprofit and for-profit contractors?

Our study draws from panel data based on DJJ’s five years of administrative records for juvenile rehabilitation programs delivered by nonprofit, for-profit and public organizations.Using five years of contract performance data, we demonstrate that contract performance proxies are not closely correlated with broader public policy goals and that this correlation weakens over time as emphasis shifts to performance proxies. Performance on DJJ’s quality dimensions (e.g., process) and cost efficiency are not correlated with the state policy goal to reduce juvenile recidivism. Results are consistent for both nonprofit and for-profit contractors, although the effect less pronounced for nonprofit organizations. This lends support to principal-steward theory, though our results also counter an assumption of perfect goal alignment. We end with a discussion of implications for governance and the accountability of private agents producing services on behalf of the state.