Research on minimum wage is still divided
Minimum wages are one of the most contentious topics in economic policy. Many American states and cities are experimenting with big minimum wage increases, so that there is now a lot of variation across the country.
To many in the news media and in the world of think tanks and activists, being pro- or anti-minimum wage is akin to a religious belief. But even in the world of economics research, there’s plenty of disagreement.
A slew of recent minimum-wage studies illustrate the point. The first is a paper by a team of academics charged with studying the effects of the Seattle minimum-wage hike that started phasing in three years ago. Ekaterina Jardim, Mark C. Long, Robert Plotnick, Emma van Inwegen, Jacob Vigdor and Hilary Wething looked specifically at the non-chain restaurant industry. They found that although the first increase in 2015 caused little change in employment levels, by 2016 the higher wage was starting to reduce the number of hours worked by low-earning restaurant employees.
Another recent study, by Ed Leamer, Jerry Nickelsburg, Till von Wachter and Frederic Zimmerman, was presented at last weekend’s American Economic Association (AEA) conference in Philadelphia. These authors found even larger negative effects from minimum-wage increases. Estimating a model of labour demand relying on data for various California counties, they looked at how that state’s recent series of minimum wage hikes—from $6.75 in 2006 to $10.50 in 2017—affected employment. Their numbers are big—for the limited-service restaurant industry, for example, they estimate that minimum wages held down employment by 22%.
Those studies both seem like bad news for advocates of higher minimum wages. But three important new studies found a much more benign effect. In one paper, economists Doruk Cengiz, Arindrajit Dube, Attila Lindner and Ben Zipperer look at all state minimum wage changes between 1979 and 2016—a huge data set. When a state raised its minimum wage, Cengiz et al. looked at the change in the total number of jobs in the state paying a little more than the new floor. They found that these jobs tended to increase by about the same amount as the number of jobs below the new minimum wage that were eliminated. The implication is that minimum wage hikes raise wages but don’t kill many jobs. Cengiz et al. estimate that a 50% increase in the minimum wage leads to only a 3% decrease in employment—a relatively modest impact.
The second of these studies, by economists Sylvia Allegretto, Anna Godoy and Michael Reich, looked at big recent minimum wage increases in seven cities. In each case, they compared the area to a synthetic control—a weighted average of other places constructed to have similar employment trends and demographics. They found no statistically significant effect on employment in low-wage industries like food services.
The final paper, by economists Amanda Agan and Michael Makowsky, took a different approach—instead of looking at employment levels, they looked at imprisonment. This is an important addition to the literature because it measures not just dollars and cents, but real human behaviour—if minimum wages help keep people out of jail, it’s a strong hint that the boost in material comfort and dignity they provide outweigh the despair they create through layoffs. Agan and Makowsky found that a minimum wage hike of 8% reduces the recidivism rate of both male and female prisoners by 2% during the following year. That’s a significant and positive effect, and it implies that minimum wages benefit the poor more than they hurt.
It’s important to remember that these studies are all very limited. Some of them, like the Leamer et al. and the Allegretto et al. studies, are preliminary and subject to change once the final analysis is concluded. The studies that use synthetic controls—Jardim et al. and Allegretto et al.—could be wrong if they’ve chosen the wrong controls. The method used by Cengiz et al. requires its own set of assumptions, which could be wrong.
Anyone following the research debate will be tempted to throw up their hands. What can we learn from a bunch of contradictory studies, each with its own potential weaknesses? Some cynics even see the contradictory results as reason to doubt the usefulness of empirical economics itself.
But this is the wrong response. The right reaction to the contradictory studies is caution. Policy makers and advisers should read the whole literature, including studies that yield conclusions they don’t like. They should try to get a picture of which research methods are considered the most reliable, and why. And then they should move forward cautiously with policy, taking steps to try to help the poor, but not making the steps so big and bold that they can’t be reversed if things go wrong.
In the case of minimum wages, a majority of the evidence seems to indicate that raising the wage floor by modest amounts isn’t very dangerous. That means that experiments like the ones now underway in places like Seattle should continue. But the studies showing larger harm from minimum wages boosts should be a reason not to make the increases too large or abrupt, and not to implement big hikes at the federal level. To borrow a phrase from Deng Xiaoping, the right approach is to “cross the river by feeling the stones.” Bloomberg View
Noah Smith is a Bloomberg View columnist.
Comments are welcome at firstname.lastname@example.org