Thu, 20 July 2017
My guest for this is Ekaterina Jardim of the University of Washington. Ekaterina is one of the authors of the new minimum wage study that has been making headlines recently, "Minimum Wage Increases, Wages, and Low-Wage Employment: Evidence from Seattle." One reason this study is so interesting is that it was funded by the City of Seattle, which is something that governments aren’t obligated or expected to do when they enact major policy changes like these minimum wage hikes.
There was a broad theoretical and empirical consensus in the 1980s that higher minimum wages have disemployment effects on the low skilled, and then Card and Krueger (1994) started a new empirical literature that found no evidence of disemployment effects.
A major problem with Card and Krueger (1994) and with many of the other studies conducted over the past quarter century was their use of proxy measures for low-skilled workers. Instead of looking at workers who actually earned less than the new minimum wage, these studies looked at groups that they knew to contain many minimum-wage workers: generally teenagers or restaurant workers. This new study does not face this limitation because Washington State requires firms to report both the hours worked and the wages of all workers.
One criticism I’m seeing a lot in response to the media coverage of this study is the fact that they had to drop multi-location firms from the sample. The reason for this is that the data only shows what firms people work for, not their location. So if a firm has locations both inside and outside Seattle, you don't know whether a given worker in that firm belongs in the treatment or the control. Still, despite this limitation, the study's sample included over 60 percent of workers in Seattle. Furthermore, the study authors surveyed employers and found that the multi-site firms that were excluded from the sample actually reported more reductions in work hours than did the firms that remained in the sample. So if anything, this omission understates rather than overstates the effect of the minimum wage increase.
One big concern people have is just how much this study's results deviate from the established literature. The authors address this by repeating their analysis using employment in the restaurant industry as a proxy for low-skilled labour. They find that using this proxy for low-skilled labour reduces the measured impact of the minimum wage to near zero, consistent with past studies that have looked only at the restaurant industry.
It seems that this apparently robust finding, replicated in study after study over the past few decades, was actually a quirk of studying the restaurant industry, which tends to substitute high-skilled labour for low-skilled labour rather than cutting total labour hours as a short-run response to minimum wage hikes.
Kevin Grier explains the synthetic control method, which the minimum wage study uses to construct a control group.