Month: January 2014

There has been a recent kerfluffle over the Sabia and Burkhauser paper (ungated here) suggesting that minimum wage increases do not very much help the American poor.  Sabia and Burkhauser report facts such as this:

Only 11.3% of workers who will gain from an increase in the federal minimum wage to $9.50 per hour live in poor households…Of those who will gain, 63.2% are second or third earners living in households with incomes three times the poverty line, well above 50,233, the income of the median household in 2007.

That’s what I call not very well targeted toward helping the poor.  To the best of my knowledge, these numbers have not been refuted or even questioned.

There has been a significant campaign lately to elevate this Arindrajit Dube piece (pdf) into a rebuttal of Sabia and Burkhauser.  I’ve now read through it, and while it is pretty dense, I don’t see that it supplies any such effective rebuttal (it is however a valuable paper, and survey paper, in its own right).

Here is an excerpt from the Dube paper:

An additional contribution of the paper is to apply the recentered influence function (RIF) regression approach of Firpo, Fortin and Lemieux (2009) to estimate unconditional quantile partial effects (UQPEs) of minimum wages on the equivalized family income distribution.

Dube also writes:

The elasticity of the poverty rate with respect to the minimum wage ranges between -0.12 and -0.37 across specifications with alternative forms of time-varying controls and lagged effects; most of these estimates are statistically significant at conventional levels.

Dube in fact counts up twelve papers on the side of “minimum wage hikes can make a reasonably-sized dent in poverty.”

Now, I don’t intend this as any kind of snide, anti-theory, or anti-technique comment, but when there is a clash between simple, validated observations and complicated regressions, no matter how state of the art the latter may be, I don’t always side with the regressions.

One interpretation of the Dube results is:

a) although a minimum wage hike applies only to some members of a community, its morale or network effects spread its benefits much more widely, or,

b) through some kind of chain link effect, a minimum wage hike pushes up the entire distribution of wages for lower-income workers

Alternatively, I would try

c) the public choice critique of econometrics is correct, these minimum wage hikes are all endogenous to complex factors, and no one has a properly specified model.  We are seeing correlations rather than causation, despite all attempts to adjust for confounding variables.

So far I am voting for c).  And there is a very simple story to tell here, namely that states which are good at fighting poverty, through whatever means, also tend to have higher minimum wages for political economy reasons.  It seems unlikely that controls are going to pick up that effect fully.

Or try another model, more tongue in cheek but instructive nonetheless.  If government is quite benevolent and omniscient, and has always done exactly the right thing in the past, we will see in the data that the minimum hikes of the past are at least somewhat effective in fighting poverty.  At the same time, the remaining options on possible minimum wage hikes will not help at all.

Dube’s paper, econometrically speaking, is a clear advance over Sabia and Burkhauser.  But Dube pays little heed to integrating econometric results with common sense facts and observations about the economy.  As Bryan Caplan has stated, the knowledge and judicious invocation of simple facts about the economy is one of the most underrated skills in professional economics.

I also get a bit nervous when the number of studies on one side of a question is counted and weighed up against common facts.  Some of these pieces are simply measuring the same correlation in (somewhat) differing ways, and the number of them says more about the publication process than anything else.  These pieces also are not all in what I would call great journals.  Maybe that is an unfair metric of judgment — I am writing this on a blog, after all.   Nonetheless I looked at the list of cited sources and pulled out the two clumps with what appeared to be the highest academic pedigree, in terms of both economist and outlet.

The first clump is a group of papers by David Neumark, with co-authors.  I find that Neumark does not himself think that minimum wage hikes do much if anything to help poverty, and he has a good claim at being the world’s number one expert on the economics of minimum wages.  In fairness to Dube, he does have some good (although I would not say decisive) criticisms of one of Neumark’s papers pushing this line.

The second source is a paper by Autor, Manning, and Smith, an NBER working paper.  They write “…the implied effect of the minimum wage on the actual wage distribution is smaller than the effect of the minimum wage on the measured wage distribution.”

Of course that hardly settles it.

You might call this one a draw, but then we return to the question of where the burden of proof lies.  I’m still stuck on, to repeat the above quotation, this:

Only 11.3% of workers who will gain from an increase in the federal minimum wage to $9.50 per hour live in poor households…Of those who will gain, 63.2% are second or third earners living in households with incomes three times the poverty line, well above 50,233, the income of the median household in 2007.