Marginal Utility

by Ken Houghton

There are appreciations aplenty of* Rutgers alumnus Milton Friedman in the blogsphere today (Mark Thoma notes the publication of Friedman's possibly-final WSJ editorial from the day after he died); endemic to a person dying and the general de mortuis nil nisi bonum of civilized society.

I'll post more tomorrow, perhaps—Virginia Postrel's 2003 hagiography in the NYT is well worth discussion, especially as it appears the only person with whom she spoke was Ben Bernanke—but I want to refer to ways in which so-called rational thinking can lead one astray, and Friedman and Stigler's Roofs or Ceilings (September 1946) (warning: PDF) (h/t DeLong, way back) should be a classic example of this.

The authors wish to compare two periods of housing crisis in the San Francisco area: immediately after the 1906 Earthquake and the then-current return of soldiers from fighting overseas in World War II (and, perhaps though not mentioned, the disinterment of the Japanese-American communities). The two periods are clearly not directly comparable, so they searched for evidence of a common proxy.

Let's ignore that they are ballparking numbers back-of-the-envelope style and focus instead upon the thing that amazes them, which is not that it takes 36 days (18 April - 24 May 1906) for the newspaper to be printed again, but rather that that issue has few people advertising to rent housing:

The classified advertisements listed 64 offers (some for more than one dwelling) of flats and houses for rent, and 19 of houses for sale, against 5 advertisements of flats or houses wanted

The authors contrast this with 1946:

[I]n 1946 there were 30 advertisements per day by persons wanting to rent houses or apartments...During this same period in 1946, there were about 60 advertisements per day of houses for sale, as against 19 in 1906.

There are several fundamental leaps of illogic in the above comparison:

1. The authors compare one day of advertisement with an average over several months
   
2. The authors treat the first day a paper is published in over five weeks as if it were a "normal" day. (One doubts James Hamilton teaches this approach to time series analysis to his La Jolla students.) This leads to:
   
   1. Assuming that people would have voluntarily paid for adverts without knowing when they would be published
      
   2. Assuming that people would view the first paper published, five weeks after the event, as one that would have people scanning classifieds for price-takers or price-makers
      
   3. Assuming that other options found and explored in the intervening five weeks would not affect the desire to advertise
      
   4. Assuming, most importantly perhaps, that people who have been homeless and without their possessions for five weeks would have the available resources to buy an advertisement and would view that as a proper allocation of their more-than-usually-scarce resources.
   
   
3. The authors seem to believe that a negative productivity shock (houses destroyed, producing homeless families with reduced resources and infrastructure) is directly comparable to an increase in demand (returning soldiers and families with plenty of argent de poche preferring to disrupt neighborhoods choosing to locate themselves in San Francisco).
   
4. The authors leverage this specious comparison of two different types of market failure — though their argument in the 1906 case appears to be that it did not fail—under extraordinary circumstances to argue that the market should be left to itself, as it is self-correcting.

And what would the effect of the solution be?

After a year or so, average rents might be up by as much as 30 percent. But even this would mean a rise of only about 5 percent in the cost of living, since rents account for less than one-fifth of the total cost of living. A rise of this magnitude...is hardly likely to start a general inflation.

Ah, the good old days, when a 5% COLA was nothing serious, and rents and rent-equivalents were only 20% of the COL itself.

No solution to the [San Francisco area] housing problem can benefit everyone; some must be hurt....Existing methods of rationing housing [i.e., those who built the neighborhood are permitted to remain living in it-kh] are forcing a small minority—primarily released veterans and migrating war workers, along with their families, friends, and relatives—to bear the chief sacrifice. [emphasis mine]

So (1) the majority are currently benefiting from the policy and (2) those who are not benefiting are those whose relocation to the area is one of convenience, not necessity—that is, they could choose to live elsewhere with minimal interruption, as the "crisis" is a local phenomenon, while the current residents have built and established their neighborhoods, their relationships, shopping patterns, and their own "family, friends, and relatives" who would be separated on an involuntary, piecemeal basis.

Why does this solution appear to be addressing a problem caused by the very people who have the most viable alternatives?

As a final note to the reader—we should like to emphasize as strongly as we can that our objectives are the same as yours: the most equitable possible distribution of the available supply of housing and the speediest resumption of new construction. The rise in rents that would follow the removal of rent control is not a virtue in itself [emphasis mine]. We have no desire to pay higher rents, to see others forced to pay them, or to see landlords reap windfall profits.

Good to know that would just be collateral damage, along with uprooted neighborhoods and the removal of any incentive for tenants to maintain the property they rent.

Yet we urge the removal of rent ceilings because, in our view, any other solution of the housing problem involves still worse evils.

*This one may be especially worth reading.