Evaluating the performance of performance-based budgeting: how do states intend to use performance measures in the budget process, and do they do it?

Abstract

Most states and many local governmetns use some form of performance-based budgeting (PBB). PBB is a method of public budgeting that provides policymakers with information about the performance of an agency or program when making appropriations decisions. The intent is to introduce some rationality and acountability into the budget process. Despite its popularity, most research finds that PBB has little impact on appropriations at the agency level. Additionally, existing theories of PBB are often diluted with caveats and qualifications. What is described here as the naïve theory predicts that agencies that meet or exceed most of their performance goals witll receive budget increases in the next budget cycle and those that fail to reach most of their goals will receive budget cuts. The naïve theory is tacitly endorsed by many scholars, even when they simultaneously argue that it would be rational for policymakers to increase funding for underperformance or that other considerations besides performance should still matter with PBB. Because PBB theory is unclear, it becomes almost impossible to determine if it is having the intended effects on governments To address these issues related to PBB, I used surveys of state legislators in Illinois, Kansas, Oregon, and Texas to derive the intended effect of PBB on appropriations. I also surveyed the public in the same four states along with a nationwide sample. The goal of these surveys was to clarify the intended effect of performance on appropriations and then use the findings to develop alternative theories that would more accurately describe PBB. In general, legislators supported funding decreases for poor performance and steady funding for good performance. Respondents to the public surveys tended to support higher funding in almost all cases, but moreso when agencies met more of their goals. This indicates different priorities for legislators and the public, but there were also significant partisan differences between the two classes of respondents that may be more important. Using actual appropriations and performance data from the same states surveyed, I then tested the alternative theories I derived from the results. In many instances, I found the same nonexistant relationship between performance and funding that the previous literature had. In cases where there was a statistically significant relationship, they were different from what would be predicted by the naïve theory or any of the alternatives. The effect of performance was usually negative, meaning that funding increases were highest for agencies or programs that met fewer of their goals. Further analysis revealed a u-shaped assocaition, where budgets increased the most for underperforming agencies and programs, followed by those with good performance records, and average performance meriting the smallest increases, or even decreases. These findings suggest that for the most part, legislators do not respond to perfromance information the way they said they would in the surveys, or how their constituents would.