Minimum Wages: The State of the Research

In January of 2016, the city of Seattle raised its minimum wage from 11 dollars an hour to 13 dollars an hour for many businesses. This was part of a phase-in for a 15 dollar minimum wage that activists had been fighting for for years. You might have heard this story, but what you may not realize is that this change produced one of the most dramatic reversals of an established scientific literature in modern history.

Let's start at the beginning. For a long time, economists have had a theory that minimum wage laws create unemployment by incentivizing employers to economize on low-skilled labour. But in 1992, two economists, David Card and Alan Krueger, came up with a clever new way to test this theory. They looked at New Jersey's minimum wage increase in 1992 from $4.25 to $5.05 per hour, and compared employment in fast food restaurants in New Jersey to employment in similar fast food restaurants in nearby Pennsylvania, which didn't raise its minimum wage.

They chose to look at fast food because the fast food industry has a high concentration of low-wage workers, and because they expected fast food restaurants to have a high response rate to their telephone survey.

What they found surprised both them and the economics profession as a whole. Card and Krueger found no evidence of disemployment effects in the New Jersey fast food restaurants they studied. Their study was released in 1994, and over the following two decades, economists produced hundreds of studies with similar results: minimum wage increases seemed to have no impact on employment. These studies tended to focus on the restaurant industry, because of its high concentration of low-wage workers, and on teen workers, many of whom work at or close to the minimum wage.

This research shifted economists' views to be more friendly towards minimum wage increases, and it gave fuel to labour activists who wanted even higher minimum wages. After the financial crisis of 2008, this crystalized into the "Fight for $15" movement, which demanded that all businesses should pay their workers a minimum of $15 per hour.

The city of Seattle was one of the first to adopt this as a policy, and began their phase-in with an increase to $11 for large businesses in April 2015 and $13 for many of those businesses in January of 2016. The city government was confident that their policy would work out for the best, so they commissioned a study by a non-partisan group at the University of Washington to measure the effects of the change.

Importantly, unlike Card and Krueger and many other researchers since them, the researchers at the University of Washington didn't have to use restaurant workers or teenagers as a proxy for low-wage workers. That's because they had access to data from the State of Washington's Employment Security Department, which catalogued every worker's wages and hours worked.

This study showed a stunning change in response to the minimum wage increase. When Seattle raised its minimum wage to $13, the hours worked by low wage workers fell by so much that they actually earned less in total because of the change than they would have without it. The average worker actually earned $125 less per month at a $13 minimum wage than they did at an $11 minimum wage.

So why the sudden change, when all the previous research suggested the minimum wage should have zero impact on employment? It turns out that the focus on restaurant workers and teenagers as proxies for low-wage labour was biasing the results of previous studies. This new study even checked and found that, had they used restaurant workers as a proxy in their sample, they would have found a zero effect just like all the other studies done in this way. It turns out that the entire empirical literature on the minimum wage, from 1994 to 2017, was wrong. Even though there were hundreds of independent studies, they all suffered from the same fatal flaw: the use of bad proxies for low-wage labour.

Even if you measure something a hundred times and get the same answer, it could still be wrong if all the measurements were made with a broken ruler. Similarly, even if hundreds of serious scientists independently study a topic and get the same result, they could all be wrong if their data was bad. The economists studying the minimum wage used proxies for low-wage workers because they were limited by the data available to them. But we now know that those proxies weren't so good after all.

Thank you for watching this video. If you want to know more about the latest research in economics, check out my podcast at economicsdetective.com, or by searching for Economics Detective Radio in your favourite podcast app.