Evaluation using random assignment experiments: Demonstrating the effectiveness of earnings supplements

Abstract

The UK is preparing to start one of the largest random assignment evaluations of a new social policy that has ever been undertaken in Europe or North America. This juncture is a useful time to examine the merits of random assignment evaluation using new results from one-of the most widely cited experimental evaluations the Self-Sufficiency Project in Canada. Random assignment experiments are the most reliable approach to measure the impacts of changes in social policy. However, they are often expensive and cannot answer all relevant research questions. The Canadian Self-Sufficiency Project demonstrates these qualities. It showed that the provision of earnings supplements to lone parents who leave welfare for full-time work can increase employment and earnings and decrease welfare receipt. The high quality of research provides credible evidence that the large, initial programme expenditure can mostly be recovered through reduced welfare payments and higher taxes.