Public Employment Agencies and Unemployment Spells: Reconciling the Experimental and Nonexperimental Evidence
研究发现，非实验研究可能高估公共就业机构与失业时长的正向关系，因为求职者往往在尝试其他方法失败后才使用该机构，英国1987-88年调查数据支持这一假设。
Econometric evidence strongly suggests that unemployed job-seekers who use the services of a Public Employment Agency (PEA) have longer unemployment spells than those choosing alternative search methods. Yet, in some well-designed U.S. experiments, increased use of PEA services has been associated with faster transitions into jobs. The author argues that the nonexperimental studies may be biased toward finding a positive relationship between unemployment spell duration and PEA use because they ignore the possibility that PEAs are chosen by many job-seekers only after other search methods have been tried unsuccessfully and a period of unemployment has elapsed. An analysis of U.K. survey data with information on the timing of PEA use in 1987–88 supports that hypothesis.