This week, University of Washington (UW) researchers released what is sure to be a key paper in the literature on the effects of the minimum wage on low-wage workers. The study, which looked at Seattle’s recent multistep minimum wage increase, is valuable because much of the recent interest in minimum wage increases has come from local governments rather than the state or federal government.
Ekaterina Jardim and her colleagues found that, although the first stage of the hourly increase from $9.47 to $11 had modest effects on workers’ total income, the second stage of the increase from $11 to $13 reduced low-wage workers’ income an average of $125 a month. Incomes fell because the reductions in employment were more substantial than the increases in hourly wages mandated by the City of Seattle.
The UW study is not the final word in minimum wage research. The study has important strengths associated with data quality, but it has weaknesses related to study design. These strengths and weaknesses yield an important trade-off: the unusually high-quality data that make the study so valuable limit what kinds of analysis and conclusions are possible.
Researchers face these trade-offs all the time, so any policy decision should be based on the picture that emerges from a broad range of research rather than a single study. Nevertheless, the UW study highlights the real limits of the minimum wage as a policy solution for boosting incomes.
The UW study’s strength: Good data on low-wage workers
The major contribution of the new UW study to the minimum wage literature is its use of unusually high-quality labor market data collected in Washington State. Most prior research on the minimum wage uses either payroll data on employment and total earnings or highly imperfect survey data on hourly wages and hours worked.
Because Washington is one of the few states that requires payroll data reporting hourly wages and hours worked, the UW team can focus on the slice of workers with wages close to the minimum wage. They also can estimate the effect of the minimum wage on hours worked rather than just the number of workers employed.
Although this distinction seems trivial, it matters a lot. Because much of the adjustment to a new minimum wage may come through reduced hours rather than fewer jobs, these data advantages are an important part of the new study’s appeal.
One problem with the higher-quality Washington State data is that they make the new UW study difficult to compare with other research. Because no other studies can focus so narrowly on low-wage workers directly affected by the minimum wage, and because other studies generally look at the impact of the minimum wage on employment rather than hours worked, they are measuring a different phenomenon from the new UW study.
If these earlier studies had the same high-quality data as the UW team, their results could be comparable with those of the UW study, but they may also be different.
The UW study’s weakness: Problematic comparison groups
The method employed by the UW study to estimate the effect of the minimum wage raises questions. To get at the effect of the minimum wage, Jardim and her colleagues need to compare what happened to workers in Seattle with the outcomes for a group of similar workers outside Seattle that did not experience the minimum wage increase. They used two comparison groups:
- Workers in the portions of King County that exclude Seattle
- Workers in the rest of Washington State
Both comparison groups were statistically adjusted using standard methods to make them more comparable with workers in Seattle.
The problem arises in that the city of Seattle and its outlying suburbs are likely to experience different employment trends and respond to economic fluctuations in different ways. These differences are magnified when counties that are even further away from Seattle are included in the comparison group.
Some of these differences can be accounted for through statistical adjustments, but many of them cannot. It’s possible that Jardim and her colleagues are picking up on the differences between large cities and suburbs rather than the actual effect of the minimum wage.
Because cities, suburbs, and rural areas grow and react to change in different ways, prior research on the minimum wage has tried to make apples-to-apples comparisons: cities are matched to similar cities as a comparison group, and counties are matched to other counties.
The best prior research finds that the minimum wage has little effect on employment. And although the UW study does its best to form comparison groups, the apples-to-apples case is not as compelling as prior studies.
Why would Jardim and her colleagues choose these comparison groups? It goes back to the powerful labor market data collected in Washington State. To use these better data, the UW study had to limit its comparison group to Washington State. Prior studies used lower-quality data that allowed them to select what was arguably a better comparison group.
What does the UW study tell us?
The new study of Seattle’s minimum wage is an important contribution to an informed policy debate. It is not conclusive, but highly suggestive that some localities may be reaching the limits of beneficial minimum wage increases.
All analysts agree that there is some level at which the minimum wage is so high that hiring low-wage workers becomes unprofitable for employers and more harm is done to workers than good. The new UW study suggests that some of the most proactive local governments in the United States may be pushing against these limits of effective minimum wage policy.