Why the supply-side matters—always

Paul Krugman has a new post on unemployment compensation:

A reader writes in, having just heard somebody or other claim that the current unemployment rate would be much lower if unemployment benefits were even less generous, and asks whether this can be true. And the answer is no. People who say things like this are fundamentally confused about what the economic research actually means.

I think he’s right about unemployment, but I’m afraid that it is Krugman who is “fundamentally confused” about the effects of UI:

What this means, in turn, is that UI may raise the “full employment” level of unemployment “” the level at which labor shortages start to appear, wages start an upward spiral, and inflation becomes a potential problem. Since the Fed raises interest rates when it sees signs of inflation, there is a sense in which UI raises unemployment on average over the business cycle.

But all of this is totally irrelevant to our current situation, where inflation is running below target, the target is too low anyway, and the reason we have mass unemployment is that there just isn’t enough demand, and hence there just aren’t enough jobs, no matter how desperately people search for them.

There are actually two or three mistakes here.  It’s easier to see them if we use the AS/AD workhorse.  Krugman is saying that UI can reduce the LRAS curve, but if the AD and SRAS curves intersect to the left of the LRAS curve, then it won’t be binding.

One problem is that if you lower the LRAS curve by 1%, then the entire business cycle is shifted downwards by 1%.  Real GDP is 1% lower in both booms and recessions.  That’s the natural rate hypothesis.  We see this pattern in European countries like France and Italy, where unemployment cycles around a relatively high natural rate, say 8% to 10%.

But Krugman also forgets that the SRAS curve is impacted by UI.  If workers are pickier about accepting crummy new jobs, then nominal wages are less flexible, and the SRAS curve shifts to the right more gradually than usual.  The self-correcting mechanism may work, but it will take longer.  During the interim the unemployment rate is higher than it would otherwise be.

I think Krugman also confuses stylized models with reality, which is much messier.  So while I agree with Krugman that a demand shortfall is the major factor restraining RGDP at the moment, in particular industries like housing construction there are already shortages of skilled workers.  In fairness, the aggregate wage data shows no sign of aggregate labor shortages, so the effect of these spot shortages on unemployment is small, but not zero.

Here’s another mistake:

Here’s what is true: there’s respectable research “” e.g., here “” suggesting that unemployment benefits make workers more choosy in the search process. It’s not that workers decide to live a life of ease on a fraction of their previous wage; it’s that they become more willing to take the risk of being unemployed for an extra week while looking for a better job.

Actually it’s both. Unless you have lived a very sheltered life, then you’ve occasional come across poorly educated workers who hate their crummy jobs.  (I haven’t led a sheltered life.)  These workers will sometimes jump at the chance for a 99 week paid vacation.  I would too.  Yes, this is hard for a college professor to understand, as the UI benefits are low, it hurts your reputation, it reduces your prospects for the future, etc.  But I’ve known a number of people who don’t care about any of that. Life is complex.  Some have a bit of saving.  Others pick up extra change working part time in the underground economy.   Don’t get me wrong, this group is only a tiny percentage of the labor force in America, but it’s not zero.

FWIW, at the current level of unemployment benefits (73 weeks max, less in many states) I’d guess that extended UI is only raising the unemployment rate by 0.3% to 0.6%, but I admit that’s just a guess—I could be wrong in either direction.  So I agree with Krugman’s claim in the first paragraph, but would warn readers that the problem is much more complex than most assume.

The past 50 years in Western Europe should teach us a bit of humility.  When the natural rates of unemployment in Europe rose sharply in the 1970s and 1980s, there was initially widespread denial, especially among Keynesian economists.  Over time people have gradually accepted the notion that the eurozone natural rate is higher than in America, and also higher than in some Nordic countries with generous UI benefits.  We don’t yet understand all the factors that explain changes in the natural rate of unemployment, which means that claims by people like Casey Mulligan, while likely overstated, should not be dismissed out of hand.



19 Responses to “Why the supply-side matters—always”

  1. Gravatar of jknarr jknarr
    11. June 2013 at 08:50

    Brilliant. Boom. I see fireworks in June. Is school out or something?

    The NYT should be giving you equal space. They need your CV.

  2. Gravatar of Floccina Floccina
    11. June 2013 at 08:52

    Actually it’s both. Unless you have lived a very sheltered life, then you’ve occasional come across poorly educated workers who hate their crummy jobs. (I haven’t led a sheltered life.) These workers will sometimes jump at the chance for a 99 week paid vacation.

    I have known people who have worked for cash as long as they can collect and right away switched to taxed work right when there UI runs out, sometimes with the same employer. I guess that is a different issue because they are in fact employed they just look unemployed to the Government. Also wives of good earners sometimes purposefully stay on UI because they work for in family consumption. Both of these though are taking a job that a taxed worker could be doing but there are working.

  3. Gravatar of J Mann J Mann
    11. June 2013 at 09:06

    I wish Krugman would be more precise.

    I assume if pushed, he or DeLong would clarify that he was only responding to the claim in his first paragraph that without current UI benefits, unemployment would be “much lower.” But the rest of his column reads like he’s claiming that there is no effect at all.

    It would be nice if someplace, he said that the best estimate is that but for current UI benefits, unemployment would be 0.X% lower, and that of those Y hundred thousand people, this many would probably be out of the measured workforce altogether, and that many would probably be working at measurable jobs.

  4. Gravatar of Niklas Blanchard Niklas Blanchard
    11. June 2013 at 09:26

    I am related to one of those people who received 99 weeks of unemployment, and in the 100th week promptly got a job.

  5. Gravatar of TallDave TallDave
    11. June 2013 at 09:39

    “These workers will sometimes jump at the chance for a 99 week paid vacation.”

    I don’t know, our society is much, much higher on the Maslow scale than we were 50 years ago, so marginal incentives are lower. People who are working non-career jobs (which is most of them) have only the incentive of the difference between UI and work — which can actually be negative for some of them because of high MTRs caused by various welfare cutoffs, meaning we are asking them to work to get less money — and in many more cases the effective wage is a few dollars an hour which won’t affect their lifestyle much (it’s certainly nothing like the difference between eating and not eating that workers in 3W countries face).

  6. Gravatar of ssumner ssumner
    11. June 2013 at 10:48

    I agree with all the comments. Especially jknarr’s suggestion that the NYT give me some space. 🙂

  7. Gravatar of Doug M Doug M
    11. June 2013 at 10:52

    Krugman has never been one to let the facts get in the way of a party.

  8. Gravatar of Karen Karen
    11. June 2013 at 11:04

    “WIW, at the current level of unemployment benefits (73 weeks max, less in many states) I’d guess that extended UI is only raising the unemployment rate by 0.3% to 0.6%,”

    That’s actually a pretty close guess to an estimate a colleague and I found on the effect of extending UI back in 2008 (approx 0.22 percentage point increase in the UI rate) http://www.heritage.org/research/reports/2008/11/extended-unemployment-insurance-no-economic-stimulus?ac=1.

    I only wish I had understood the monetary policy effects better in 2008. You’ve convinced me, Scott, that monetary policy has been tight and that structural adjustments could be helped and much smoothed if we’d get the supply of money to meet the demand. Thanks for your blog and helping me think about the macro economy much more clearly.

  9. Gravatar of Brian K Brian K
    11. June 2013 at 11:32

    It amazes me how easily Krugman refutes right wing ideas with ECO101 arguments, but will jump through hoops to show ECO101 doesn’t really apply to left wing planks like this and minimum wage.

  10. Gravatar of ssumner ssumner
    11. June 2013 at 12:15

    Thanks Karen, It’s good to know that people are getting similar results with better methods that my seat of the pants estimate.

  11. Gravatar of Brenb Brenb
    11. June 2013 at 13:15

    The number of people accepting extended federal benefits (EUC) was 1.8 million in April 2013, down from over 5.6 million in January 2010, yet re-employment rates haven’t improved very much. Im sure extended benefits have had an effect, but it probably has not been severe. The decline in insured unemployment is probably one of the reasons the labor force participation was falling in 2011-12.

    The same can be said for the rising number of disability rolls. Just by eyeballing it, you can see the percent of the population on SSDI increased faster than its pre-recession trend following 2008, but the increase relative to trend accounts for only about 0.2-0.3 percentage points of the total 2.7 percentage point decline in the labor force participation rate (most of the SSDI increase is among individuals age 45-64). These effects aren’t nothing, but surely too small to be the most important thing on people’s minds!

  12. Gravatar of Edward Lambert Edward Lambert
    11. June 2013 at 14:47

    The natural rate of unemployment has risen in the US. It does not seem to be common knowledge yet.


    I tend to think that unemployment insurance did help cause this rise “in this recession”. But what was the alternative really?

  13. Gravatar of Doug M Doug M
    11. June 2013 at 15:51

    Edward Lambert,

    The labor force participation rate is lower.

    It was declining before the recession. Cratered during the recession and and has not recovered.

    Not that that is an explanation. It may be a symptom of the same cause. It also means that employment may be more anemic than the numbers suggest.

  14. Gravatar of Michael Michael
    11. June 2013 at 16:39

    I agree there probably is an effect but that it is probably minor.

    According to JOLTS data, this recession and recovery have been characterized by an extremely low quit rate. Quits have recovered somewhat but are still lower than they have been at any time between 2003 and 2008.


    If we had a healthy labor market – but for those people who are enjoying “99 week vacations” – we expect to see a lot more people quitting their jobs (since the most common reason for quitting a job is to take a better job).

  15. Gravatar of drycreekboy drycreekboy
    11. June 2013 at 17:45

    It’s purely an anecdote, but consider the following. I have a trustworthy friend who’s an administrative law attorney working unemployment cases for the relevant, benefits-dispensing agency of a Southeastern state. About a month ago he related to me a case wherein the UI receiver blurted out in testimony that the State had taken so long to get his benefits to him that he’d had to look for a job. A bit more choosy indeed.

  16. Gravatar of kebko kebko
    11. June 2013 at 18:15

    “… if unemployment benefits were even less generous”

    I love how the “liberal conscience” is always exasperated at the stinginess of social services. Here we are, coming off a time with 99 weeks of UI, still at 73 weeks, and Krugman can’t imagine how we could be less generous. Of course, these things are supposed to be funded out of wages, and I doubt many workers would agree to increase the levels, all things considered. But, it’s basically being treated as a redistribution, now, and of course all reasonable people are always in favor of more of that.

  17. Gravatar of Jerry Brown Jerry Brown
    11. June 2013 at 18:48

    Your last paragraph undermines most of your argument. Sure we don’t know everything that changes the “natural rate ” of unemployment or why it is different in different countries or why it is different in the same country at different times. But why let that stop us from using this “hypothesis”. Especially if it might show Krugman made a mistake.
    The common sense non-economics argument about poorly educated people with crummy jobs makes more sense than basing an argument on this ever changing and unidentifiable variable of natural rate. However, I believe unemployment is up a lot even for educated people.

  18. Gravatar of ssumner ssumner
    12. June 2013 at 07:45

    Brenb, That’s right.

    Edward, The alternative was not running a tight money policy. But I see your point.

    Michael, That’s right.

    Jerry, The unemployment rate for college grads is pretty low.

  19. Gravatar of Geoff Geoff
    13. June 2013 at 12:49

    “Edward, The alternative was not running a tight money policy.”

    Money was not tight, if the standard is what the free market would have otherwise produced.

    Money has been extremely loose if the proper standard is used, not some silly arbitrary aggregate spending metric.

Leave a Reply