Trialing for the Public Good: Building Effective Programs from the Ground Up

The President’s fiscal year 2014 budget emphasized the creation of “a culture of performance improvement.” In his 2013 Academy of Public Administration review of GPRA goals, Donald Moynihan recommended that modernization should involve, among other things, building within government a “learning culture.” In this paper we argue that rigorous experimentation is an important feature of a learning culture in government, and that lack of attention to experimentation as a means of doing government better is a significant shortcoming of the performance movement. Those promoting “evidence-based policy-making” acknowledge the importance of experimentation but generally source evidence outside of normal government operations. The posture of both the performance management and evidence-based policy enthusiasts contrasts with the growing acknowledgement of experimentation as an important tool of management in the private sector, where experimentation is coming to be seen as a normal function of good management within firms in many industries. We offer experimental management as a “third way” of improving the delivery of public sector services. We discuss reasons for the difference in experimentation between private and public management and the potential gains from encouraging more in government. We conclude with ideas for strategy. 1. The Problem Public managers have long struggled to implement effective policies and programs for accomplishing specific social welfare goals – reducing poverty, increasing employment and earnings, supporting learning and academic success. Over the past fifty years two primary mechanisms have emerged to support the sound implementation of effective programs. Rigorous research, particularly randomized control trials (RCTs), generally supported by the federal government, has sought to answer the big policy and program design questions – does a particular policy or program, when well implemented, achieve its desired impact? These studies,