The Economist doesn’t understand marginal analysis

Strange for a magazine with such a name that they don’t know about Alfred Marshall and his marginalist revolution. See, when someone uses marginal analysis to show, unequivocally, that increases (*ahem* marginal increases *ahem*) in unemployment insurance benefits depresses employment, its not ok call them names for not doing inframarginal analysis. Its not ok to call them names for not doing inframarginal analysis because inframarginal analysis in our current policy environment is wildly inappropriate.

The first $1838 of UI (and food stamps and TANF) keeps families from starving and gives them shelter for a month or so ((This amount keeps a family of four above the poverty threshold. UI benefits in California are up to $450 a week per wage earner. If the family is low income, then the family probably qualifies for other welfare programs and then the issue is redistribution not efficiency.)) while the bread winners look for new jobs and it prevents them from having to sell off their assets, which, as The Economist mentions, would exasperate the demand shock. Every dollar above that encourages them to search longer and thus increases unemployment.

Casey Mulligan is talking about marginal increases in UI not eliminating them!

UI-apologetics come from a confusion between redistributional policy and efficient policy. You don’t need to make tenuous arguments for efficiency (“without UI people would sell their assets off and cause further decreases in AD!!!”), just argue directly for redistribution. Here, I’ll help you: the recession has made me realize that the poverty guidelines are too low. Too many people are too near the jaws of poverty for my comfort. The guidelines should be increased and more people should qualify for welfare programs.

8 thoughts on “The Economist doesn’t understand marginal analysis”

  1. I don’t see the marginal analysis there. He is saying that unemployment benefits prolong unemployment, not that marginal unemployment benefits prolong unemployment.:

    “I don’t quite understand this obsession with UI-apologetics, because UI (unemployment insurance) is just one of many policies that collectively (and some by themselves) create terrible incentives:”

    I can’t see where Casey Mulligan supports a single dollar of unemployment benefits.

  2. Look at the list of policies he’s talking about… “hike” and “extension” are common verbs seen in the list. He’s talking about increases in these programs. He hasn’t proposed their elimination.

  3. I see it now. The paragraph in the body is less clear, he doesn’t seem to support UI at all.

    The implication that extending UI benefits will have measurable effect on the depth of the recession is still ridiculous.

    Does anyone really think that ending UI will mean any more jobs will be created? Can labor supply meet non-existent labor demand?

    I don’t really get why Casey Mulligan prefers to focus on minor factors like UI and mortgage modifications (mortgage modifications which are not actually happening, while downplaying things like, say, a banking crisis:

    “The non-financial sectors of our economy will not suffer much from even a prolonged banking crisis, because the general economic importance of banks has been highly exaggerated.”

    Has Casey Mulligan been right about anything?

  4. There’s another, more credible, argument for efficiency, within a search-and-matching framework. Without (with less) UI workers will more willing to accept a relatively worse match and the average, economy-wide quality of matches will decrease. With (more) UI, you get (a) more unemployment; and (b) better matches.

    With regular functional forms and standard assumptions, there’s going to be a range over which the social cost implied by (a) is less than the social benefit that comes from (b).

    This is, of course, all about frictional (and not “cyclical”) unemployment…

  5. Angry, in the search/matching literature is there a distinction made between frictional and cyclical unemployment? Just curious how those two things are modeled.

  6. I really don’t know.

    I was trying to say that I think of UI as relevant to trend/mean unemployment (e.g. the buttload of a literature on EU vs. US trend rates) and maybe less for movements at cyclical fluctuations, but you’re right, in a fully specified model you probably can’t break it down that way.

Comments are closed.