Arthur Laffer, whose work I’ve often respected and who, I think, has been underappreciated by the economics profession, has a piece in today’s Wall Street Journal on unemployment insurance (UI). It’s titled, “Unemployment Benefits Aren’t Stimulus.” I wish I could say I like it. I don’t. What’s strong is part of his theoretical argument, one that many economists have made over the years. What’s weak is another part of his theoretical argument and his evidence.

Strong argument:

The most obvious argument against extending or raising unemployment benefits is that it will make being unemployed either more attractive or less unattractive, and thereby lead to higher unemployment.

and later:

Employers don’t usually hire people to assuage their consciences. They hire people to make after-tax profits. And if workers require more pay because of higher unemployment benefits, employers will hire fewer employees. Whether increased unemployment benefits incentivize workers to work less or disincentivize employers from hiring more workers, the effect will be the same–higher unemployment.

That’s great. But check out this part of this theoretical argument:

To see this, imagine an economy that produces 100 apples. If 10 of those apples are given to the unemployed, then people who otherwise would have had those 10 apples now won’t. The stimulus of 10 apples for the unemployed is exactly offset by the destimulus of 10 apples for those people from whom the 10 apples were taken.

This should be in economics textbooks as an example of begging the question. Notice that he assumes a fixed supply: 100 apples. So, of course, if he starts with that assumption, he can’t allow for the conclusion that unemployment benefits could lead to, say, 101 apples. In other words, he assumes, from the start of the analysis, that unemployment insurance is not a stimulus.

I hasten to add that I don’t think unemployment insurance is stimulus. But you don’t prove something by assuming it from the getgo.

His weak evidence is actually a clever graph that I have never seen before. On the left-hand vertical axis is total real benefits paid per UI recipient (average payment per week times average duration of benefits.) On the right-hand axis is the unemployment rate. Of course, they track closely. He never claims explicitly that the UI benefits are what caused the unemployment, but that seems to be what he wants the reader to conclude. The obvious alternate hypothesis goes the other way: in times of high unemployment, the federal extension of UI benefits is triggered and so average benefits per recipient rise. I think he’s got something with his graph. But he needs to dig further and try to get the causation right.

I’m against the extension of UI and, indeed, am more radical on this than Art Laffer. He writes:

No one opposes unemployment benefits as a transition aid for people to get back on their feet and find a new job.

I do. If unemployment insurance is such a great idea, let the free market provide it. You can argue that a private insurer would have trouble enforcing against moral hazard. And you would be right. But so does the government, which has no particular expertise at that. You could also argue that unemployment insurance would be underprovided in the free market because it carries a positive aggregate demand externality. But if you’re going to have a government macroeconomic policy, the way to pump up aggregate demand, without the disincentive effects of UI, is, as Scott Sumner points out again and again, to increase the money supply.

Update: I’ve learned from conversations with regular readers of Econlog that many of you don’t read any of the comments. For that reason, I recommend that you see my comment below in response to previous commenters.