This week, researchers at the Federal Reserve Bank of Minneapolis released two analyses of the economic impact of several recent increases in the separate citywide minimum wages in Minneapolis and Saint Paul. The authors of the two separate, but linked, analyses interpret their findings as suggesting that the series of minimum wage increases reduced employment in the Twin Cities, but this conclusion is not supported even by the results presented in their own reports.
The data and methodology used in the study are simply not sufficient to distinguish between the effects of the minimum wage increases and other changes in employment happening around the same time. Even more problematic are the studies’ unsuccessful attempts to estimate effects through 2020, when the study’s methodology is unable to separate the effects of the minimum wage from the devastating impact of the pandemic on the two cities’ service-sector workforces. In fact, the studies’ findings implausibly suggest that the entire decline in restaurant employment during the pandemic is due to the minimum wage increases, as opposed to lockdowns, curfews, and pandemic-related declines in consumer spending on eating out.
The first methodological problem is the studies’ choice of a comparison group for assessing the impact in Minneapolis and Saint Paul. To determine the effects of the minimum wage increases, the studies compare actual outcomes in the two cities to other Minnesota cities that had similar wage and employment growth trends prior to the minimum wage increases. These other Minnesota cities form a “control” group or counterfactual of what wages and employment in Minneapolis and Saint Paul would have looked like in the absence of their minimum wage increases.
While this approach is in principle a reasonable starting point, there are several signs that the methodology failed to generate sensible estimates in this case even prior to the pandemic period. Notably, the studies’ methodology failed two of their own “placebo” tests, finding “effects” of the minimum wage when there clearly should have been none—for example, by predicting a change in employment years before the minimum wage was put into place.
The methodology used in the studies suggests that the 2018-2019 minimum wage increase caused large employment changes in low-wage sectors in 2014 and 2015, three to four years before the minimum wage increases actually took effect (see Table 7 of the two studies). A valid methodology would have predicted zero effects this far in advance of the actual increases.
Their methodology also estimated that the 2018-2019 minimum wage change in Minneapolis increased employment by 17% in the finance and insurance industry, even though this higher-wage sector is not plausibly affected by the minimum wage.
Further, the studies incorrectly attribute large swings in employment to the Saint Paul minimum wage up to two years before employers were legally required to raise their wages. While the authors frame these results as “anticipation” effects, we find this to be implausible because turnover in low-wage jobs is typically high, allowing employers to make adjustments quickly and easily. The detection of “effects” years in advance of the actual minimum wage change, for example, is often used as a falsification or placebo test in other minimum wage research. Moreover, the pattern of these estimates is erratic: accommodation and food services saw small wage increases and large employment reductions; retail had small wage increases and no employment change; and “other services” had no wage increases and large employment increases.
The broad pattern of findings suggests that the study’s methodology is prone to estimate large swings in employment regardless of any change in the minimum wage. Rather than reflecting actual consequences of the minimum wage, all of these findings are signs that the studies’ methodology did not successfully distinguish the actual effects of the minimum wage from other unrelated changes in employment happening around the same time.
Instead of concentrating exclusively on the pre-pandemic 2018 and 2019 period, the studies also attempt to extend their analysis into 2020. This part of the analysis relies on the heroic assumption that, in the absence of any minimum wage increases, Minneapolis or Saint Paul and their respective comparison cities in Minnesota would have followed the same trajectories during the period marked by the pandemic and racial justice protests as they had experienced before those events.
The magnitude and timing of the employment effects in the paper suggest this assumption is false. The authors estimate, for example, that the Minneapolis minimum wage reduced restaurant employment by more than 40%, substantially more than the actual 30% decline in restaurant employment in Hennepin County (which contains Minneapolis) between 2017 and 2020. In other words, the study suggests that in the absence of the minimum wage increases, restaurant employment in Minneapolis would have increased by about 10% during the pandemic and period of racial justice protests. These results are even more implausible given that three-quarters of the full-service restaurant employment losses the studies attribute to the Minneapolis minimum wage occurred exactly during the pandemic and racial justice protests of 2020, a year when the Minneapolis minimum wage increased by only $0.75 at small employers and $1.00 at large employers.
In general, the restaurant employment responses estimated by the studies are implausibly large and well outside the range of existing research on the minimum wage. For example, the magnitudes of the effects estimated through 2020 are consistent with every single minimum-wage worker in full-service restaurants losing their job.
Other research, however, finds much smaller effects of the minimum wage on restaurants. The median estimate in a comprehensive review of research by Arindrajit Dube implies employment losses two orders of magnitude smaller than what was estimated by the studies for full-service restaurants. Using the Congressional Budget Office’s methodology, which has been criticized for being too pessimistic, the wage increases estimated by the Minneapolis study imply full-service restaurant employment should have declined in Minneapolis by only 2%. In contrast, the study estimated declines of more than 12% in 2019 and more than 40% in 2020.
The erratic, outsized, and frequently counterintuitive results in the two studies make clear that neither of these studies are a reliable guide to the effects of recent increases in the minimum wage in the Twin Cities.