by hilzoy
Since the Democrats are talking about raising the minimum wage, I thought it might be a good time to round up some good discussions of the question: does raising the minimum wage cause employment to fall? As far as I can tell, everyone agrees that, other things equal (and assuming that demand is not perfectly inelastic), when you raise the cost of something, demand for that something tends to decline, and thus that raising the minimum wage should have some negative effect on employment. As far as I can tell, no one disputes that; and so when people argue about the effects of the minimum wage, it's generally not helpful to point out that raising the price of something obviously causes demand for that something to drop. As an a priori argument about what happens when other things are equal, it's true, and no one I've been able to find seriously disputes it.
The argument for the claim that the minimum wage does not cause employment to fall grants this basic point, but goes on to say: in the actual world, other things are not equal, and so we need to ask: does raising the minimum wage by the kinds of amounts people normally propose have other effects on employment that tend to counterbalance or even outweigh this negative effect? If it does, then arguing that since (other things equal) raising prices tends to lower demand, raising the minimum wage must reduce employment would be exactly like arguing that since, other things equal, cutting someone open with a knife tends to make that person less healthy, surgery can never improve your health. As I can attest, having an appendectomy does have negative consequences for your health. I was laid up for ten days after mine. But someone who kept pointing to the fact that opening someone's abdomen up with a knife necessarily harms that person would be missing the point of appendectomies, namely: that those negative effects are vastly outweighed by the benefits of not having your appendix rupture and kill you. Same here.
Whether or not raising the minimum wage does have effects that outweigh its costs is, of course, an empirical question. Luckily, Kash Monsour has assembled some of it in a series of posts. He explains his starting point at the outset; I think it's exactly the right one:
"When I started my graduate studies in economics, I was perfectly accepting of the classical economic analysis that illustrates why a minimum wage should cause low-income people to lose jobs. But by the time that I had finished grad school, I had learned that there are economic theories that lead to different conclusions, and I felt that I had seen enough evidence to call into question the classical prediction of the effects of raising the minimum wage. Since then, the additional evidence that I’ve seen has tended to generally confirm that minimum wage laws only have a very small negative impact (or very possibly no impact at all) on employment.This is a subject on which I have tried to let the empirical evidence guide my opinion. And personally, I'm persuaded that the benefits of a higher minimum wage for low-income individuals (and the distribution of income more generally) outweigh any possible negative employment effects."
He then presents some of that empirical evidence, which I'll summarize below the fold (with helpful graphs, which I copied from him.) If you prefer arguments from authority to annoying graphs, you could just read this statement (pdf), signed by 650 economists, including five Nobel laureates.
Apparently, one of the things that led people to start questioning the assumption that raising the minimum wage would depress employment was that some researchers decided to change their methods. Most researchers had originally just looked at the question: are changes in the minimum wage in one place correlated with changes in employment in that place? This is not an ideal method, since you don't have anything like a control group, and therefore you can't really ask whether any changes you observe were actually caused by increases in the minimum wage, or whether they would have happened in any case for unrelated reasons.
In the early 90s, however, some researchers started looking at states and localities that were near one another, where some states/cities/whatever had raised the minimum wage and others had not. This is a better research design: if the states/cities/whatever are near one another, then presumably many of their background economic conditions are similar. That means that those states/cities/whatever that do not hike the minimum wage can serve as rough controls for those that do: if one state hikes the minimum wage and its neighbor does not, and we observe some effect on employment in the first but not the second, we can have more confidence that that effect is caused by the hike in the minimum wage than we could have if we only looked at the first state. On the other hand, if those states' employment rates look pretty similar even though one raised its minimum wage and the other didn't, that's evidence that raising the minimum wage doesn't do much.
With that in mind, look at this graph. It shows private sector employment and average weekly earnings for nine northeastern states, over a five year period. Those states are ranked by how much they changed their minimum wage during those five years: Maine, on the left, raised its minimum wage by the greatest amount ($1.20), while New Hampshire, New Jersey, and Pennsylvania, on the right, left their minimum wage at %5.15 for the entire period.
If the minimum wage had an effect on employment, you'd expect to see the red bars (change in employment) gradually increasing in size from the left side of the graph to the right. Maybe there would be a few states here and there that didn't fit the pattern, but broadly speaking, the bars should be getting higher from left to right. But, as Kash says:
"If you see a systematic relationship in this chart between raising the minimum wage and employment or wages, then you have better eyes than I do. (Note: their formal statistical correlations are almost exactly 0.0 in both cases.)"
Here's a second case, which Kash describes as follows:
"In the following table I’ve selected all of the instances where one of the Northeastern states increased its minimum wage, but where a similar neighboring state did not. The table shows what happened to overall employment in the private sector and to the average weekly wage in the private sector each case."
And here's the table:
Not much of a correlation there either.
Here's a third example, also via Kash, from a study of what happened to restaurant wages in San Francisco and Alameda County (across the bay) when San Francisco raised its minimum wage to $8.50 in 2004. The table sort of speaks for itself:
Minimum wage increases in SF but not in the East Bay; employment goes up in both SF and the East Bay; the increase in employment is higher in SF than in the East Bay.
This is some of the evidence people point to when they argue that raising the minimum wage would not have adverse effects on employment. (At least, if we're talking about the kinds of raises normally envisioned: I imagine that raising the minimum wage to a million dollars a minute would cause a dramatic drop in (reported) employment, but since no one is proposing anything like that, I propose to leave that detail out of consideration.) There's also a good FAQ on the minimum wage here, and a more technical literature survey here. Since I try to be evidence-driven on this question, I'm open to counterarguments and other studies; I just thought it would be good to get these out there.
Two questions about that Northeastern study:
(1) What's the percentage -- in those states -- of laborers at the minimum wage? Looking at the aggregate change in employment owing to a change in minimum wage law might not be the most sensible comparison. After all, do we really consider it as confirmation of minimum wage changes not affecting employment that computer programmers are not getting laid off? In Delaware, for example (someone stop me if I am reading the chart from Phil's link wrong), there are 219,000 workers at hourly wages, of which 4,000 are at the minimum wage. That's only 2% of the workforce to begin with.
(2) I have to admit that I'm skeptical that looking at the next year's employment numbers should be regarded as compelling evidence. Few of these employment situations are of the classic fixed plant/flexible employment scenario where managers lay off or take on new labor immediately based on market conditions. This is in its way ironic, because labor being variable in the short-run, you would think that business would be very responsive to changes in labor costs. But I think you have to look a couple of years down the line (which is itself dicey, because then other factors cloud the picture). Shorter me: businesses may be captive in the short run, but local changes in minimum wage could make a difference the next time they decide to buy a plant, the next time they make major capital decisions, and those, being long-run effects, would only begin to show up three, four years in the future.
Posted by: Ara | November 21, 2006 at 05:44 PM
If one does not have a right to live without roommates;
And if one does have a right to live;
Then does one have a right to roommates?
Posted by: Rebecca Borgstrom | November 22, 2006 at 01:16 PM