The Minimum Wage

2/16/2013 04:56:00 PM
Everyone is talking about the minimum wage, so I figure I should too (see here).

I'm not prepared to give the minimum wage hike quite the bear hug that Krugman did, where he says that "it's actually good policy." Krugman and a host of others have noted that the conventional Econ 101 analysis that minimum wage laws cause unemployment does not stand up to empirical analysis, as study after study finds no effect of the minimum wage on employment. And it isn't just employment--studies have looked at other hypotheses, such as whether minimum wage hikes cause inflation, and find very little effect there too.

But a finding of no effect is not the same as saying that minimum wage laws work. We would get negative results if, for example, the minimum wage remains below the market clearing wage, if the population at risk of a minimum wage hike is exceptionally small, or if obtaining exemptions or loopholes around such hikes are relatively easy. In fact, all of these things are true of almost all the minimum wage studies that have been conducted on US data.

I'm having trouble locating the right dataset at the moment, so I will just use this to illustrate the point. Just looking at the new proposed change from the current 7.25 to 9 per hour, we see that this would affect only about 6 percent of the sample. And that is considerably larger than the population "at risk" in all previous minimum wage surveys, because all previous hikes (except Washington, which is at 9.19) were smaller than 9 dollars an hour, and therefore involved a much smaller population of people. But! Note that at the current 7.25 minimum wage, nearly 5 percent of full-time workers are earning less than the federal minimum. That is, because so many people are exempt from the minimum wage one way or another, we know that the maximum plausible effect size is way, way lower than 6 percent. My point is that it takes fairly large sample sizes to be able to statistically detect employment effects from minimum wage changes as small as what we've observed in the sample. That's not to say that the existing research hasn't been successful in ruling out large effect sizes, but I'd say it remains to be seen whether small employment effects of minimum wage hikes do occur. And given the bell-shaped distribution of income, the more we hike the wage, the larger those effects are likely to become.

The second problem is that minimum wage hikes are relatively easy to ignore. Historically, minimum wage laws have been riddled with exceptions for servers, teenagers, and businesses who demonstrate an inability to afford the higher wages in general. But, here's a clarifying way to think about it: it is illegal to pay your neighbor less than 7.25 per hour to mow your lawn, but perfectly legal to pay him less than 7.25 to mow your lawn. Part of the definition of an employee, as opposed to independent contractor, is that you (the employer/customer) control the number of hours they work. But, from your perspective, all you care about is how much it costs to mow the lawn, not how much it costs per hour. Hence, giving up your employer status is costless, but often allows you to legally ignore the minimum wage.

The ease of these loopholes in the minimum wage law explains why there are so many more workers earning less than 7.25 an hour than actually making that much. The empirical consequence is that it is much harder to discern what effects the minimum wage actually has on welfare.

I will also add that a lot of the minimum wage studies have looked at relatively small polities, like states, to examine the effects of minimum wage changes. Unlike the US as a whole, it is relatively easy to move across state lines, meaning that, for example, we wouldn't expect a state change in minimum wage to affect unemployment rates even if it does decrease employment--laid-off workers will simply work in a neighboring state. Not all study designs face this type of problem, but I remain doubtful about a number of them.

That said, there are a variety of reasons that minimum wage laws could be welfare-improving. For example, if employers are able to set monopsonistic wages. Moreover, the fact that there is no conclusive evidence of unemployment as a result of the law is promising. But, I'm not totally persuaded that this is good policy.

As a final remark, let me offer an answer to Greg Mankiw. I don't really know why Obama chose 9 dollars. Maybe it is because it is close to the highest minimum wage in the country, Washington's 9.19 per hour. Or, if I may, maybe because a childless adult working full time would have to earn at least 9 dollars an hour in order to stay above 133 percent of federal poverty level, the line below which he becomes eligible for the medicaid expansion included in the Affordable Care Act.

Or to put it another way, 9 dollars an hour amounts to the federal poverty level plus the cost of individual health insurance.