There has probably never been a National Bureau of Economic Research working paper that produced as much glee in the media as last week's report showing that Seattle's minimum wage law may have led to a net loss in wages for low wage workers. According to the analysis, there was a reduction in average hours worked among those in the low wage labor market that more than offset the gain in wages. The result was a net loss in wages for exactly the group of people the law was intended to benefit.
This finding was quickly picked up in every major news outlet. While some, notably the New York Times, reported the finding with appropriate cautions, others (e.g. here, here, here, here, and here) were nearly gleeful at the idea that workers in Seattle were losing their jobs. Most of the reporting ignored the fact that the same week a team of researchers from Berkeley produced an analysis using a very similar methodology that found no statistically significant impact on employment.
There are important differences in the studies. The Berkeley study follows much prior research and only looks at the restaurant industry, a major employer of low wage workers. The University of Washington NBER paper looked at all workers getting paid less than $19 an hour. It also had two additional quarters of data. However, the Washington study also excluded the roughly 40 percent of the workforce that worked at multi-site employers (think Starbucks and McDonald's).
In other words, it it not obvious that the Washington study is the "better" analysis. The Berkeley team has produced much of the cutting edge research on the minimum wage over the last fifteen years. I doubt that many of the reporters touting the Washington study would be able to explain why it is a better analysis of the impact of Seattle's minimum wage hikes.
To be clear, I don't see myself as having a dog in this race. Many of the minimum wage advocates are personal friends, but I have always been cautious on this policy. We all know that there is some point at which a higher minimum wage is counterproductive, leading to job losses that more than offset the benefit from the higher wage.
I don't know if $15 an hour in Seattle in 2020 is above this point. I felt that $15 nationally by 2020 is. This is the reason I refused to sign the petition circulated last summer by Bernie Sanders supporters that endorsed his call for a $15 an hour minimum wage in 2020. (I did sign a letter calling for a $15 an hour wage in 2024, which would be roughly 15 percent less, adjusting for inflation and productivity growth.)
Anyhow, I have tried to look at the Washington University study with open eyes. It is a serious work, but there are clear problems. The fact that multi-site employers are excluded could well be an issue. It is possible that larger employers responded less or in the opposite direction than single site employers. I wouldn't assume that this is the case, but we certainly can't rule it out.
Perhaps more importantly, the study has no data on people who are contract workers or self-employed. This could matter in the age of Uber. If more people turned to these forms of employment in Seattle than in the synthetic control set up in the analysis, this could offset the job loss they found.
The biggest concern is the possibility that the boom in Seattle has pulled a large number of people over their cutoff for low wage employment ($19 an hour). In its analysis of total employment it finds no statistically significant negative effect on total employment. This is even true in the restaurant industry, with its large concentration of low-paid workers.
This raises the possibility that employers have been forced to pay workers more for reasons having little to do with the minimum wage. If the boom is pulling large numbers of workers over this $19 an hour threshold, the study's methodology would attribute the reduction in low paying jobs to the minimum wage.
There is actually an easy way for the Washington team to test for this possibility. They can run their model looking at employment above $19 an hour. They have all the data necessary for this sort of analysis.
If it did turn out that they found an increase in high wage employment associated with the minimum wage this would mean either that the minimum wage hike actually led to the growth in high wage employment (fairly implausible), or that Seattle's boom is pulling people out of the low wage labor market, thereby negating their finding that the minimum wage hike had led to the reduction in low-wage employment.
Anyhow, we are far from having the final word on Seattle's minimum wage hike or the issue more generally. The Washington study furthers our understanding of the issue and it helped to expose the enormous bias in the media.
Note: Additional media citations were added in response to requests.