When Good Measurement Goes Wrong
本研究复制了Hoffman对最低工资就业效应的估计，发现其双重差分法的平行趋势假设不成立；改用合成控制法后，估计出低技能劳动者最低工资就业弹性为-0.5，与原始研究结果相近。
Hoffman’s (2015) replication of Sabia, Burkhauser, and Hansen (SBH 2012) suggests that “unlucky” measurement error in low-skilled employment in the Current Population Survey Outgoing Rotation Groups (CPS-ORG) led SBH to produce upwardly biased estimates of the labor demand effects of the 2005 to 2006 New York State minimum wage increase. This study replicates Hoffman’s preferred policy estimates from the full CPS and finds evidence that the parallel trends assumption underlying his difference-in-difference approach is violated. When a synthetic control state with pretreatment employment trends similar to those in New York is constructed, this study estimates a relatively large negative employment elasticity with respect to the minimum wage for low-skilled individuals (–0.5), similar to the estimate SBH obtained using the CPS-ORG (–0.6).