Thursday 28 February 2013

Why we cannot have nice things

Economists disagree on some policies; you can usually find an economist willing to support some policy proposition that his colleagues won't like. But, surely, were there some set of policies agreed upon by economists from across the spectrum, it would probably be a good one, right?

Over the last decade, Bryan Caplan produced a series of papers and a book showing that economists' views on matters of positive economics diverged widely from the views of the public, even after correcting for income, ideology, education, and a host of other covariates. Broadly speaking, relative to economists, the public were shown to be pessimistic, xenophobic, anti-business, and to love make-work projects.

Paola Sapienza and Luigi Zingales have added* to this literature with a new Chicago Booth working paper: Economic Experts vs Average Americans. They compare public and economist support for a series of policy propositions and conclude that the set of policies that would command substantial agreement among economists would have little chance of electoral success.

I've a few quibbles with their method: they collapse "agree" and "strongly agree" into single categories; there can be useful information in strength of agreement. When we're thinking about giving up a bit on a less-important policy to get concessions on something that's more important, preference-intensity matters.

I'll also quibble a bit with the Booth Panel as being that representative a sample. First off, the survey shows that Booth Panel economists trust government more than does the general public. I'm not sure that that's a finding that would be found in a broader survey of economists, but it could be. But I do know that the Booth panel's views on the minimum wage differ vastly from every prior survey of economists I've seen on it: the Booth Panel had 47% support for the $9 US minimum wage (albeit with many supporters saying in the comments that hiking the EITC would be even better); Whaples 2006 (see here or here) had 87% of economists agreeing that minimum wages hurt the job prospects of the young and unskilled, and 47% wanting to abolish the minimum wage entirely. Whaples 1996 had 87% of economists agreeing that minimum wages hurt the job prospects of the young and unskilled while Whaples 2006 had 47% wanting to abolish the minimum wage entirely. [Thanks to David Friedman, in comments, catching that I'd meant to put "1996 and 2006, respectively," rather than 2006. I'd linked both sources but screwed up the attribution in text. It matters because consensus could have deteriorated over the decade.]

But, the differences remain striking. Even among a sample of high-trust-in-markets Democrat members of the public, average differences in agreement between economists and the public across questions was thirty percentage points.

Well, maybe this is all because people can always find some economist who'd agree with their position. But in that case, we'd expect to find smaller differences between economists and the public where economists agree with each other more strongly; they find the opposite. They put the difference up to specific training in economics and expert knowledge among economists - about what Caplan had concluded a decade ago. But they add a very nice twist: one wave of the public survey included information in the relevant questions, "Nearly all economic experts agree that ...". Telling the public that economists agree on something usually very slightly reduced disagreement, but sometimes made things worse.

They probe further into disagreements and find some evidence that the public have different ceteris paribus assumptions. Economists pretty much all agree that carbon taxes are much better than CAFE regs. But economists also tend to assume that carbon taxes will be handled in some revenue-neutral way with offsetting transfers. And the public doesn't trust government to actually make things revenue-neutral.** They then put most of the differences down to differential auxiliary assumptions around ceteris paribus. But Caplan's prior work was on purely positive questions like whether foreign aid spending is a major reason that the economy is performing poorly. I'm not sure what auxiliary assumptions might have driven the vast differences he was finding.

I don't have particular reason to doubt their results around trust in government. But even if you didn't trust the government to make sure a carbon tax were revenue-neutral, CAFE standards are a really wasteful way of achieving any particular goal: you force the public to pay $1 for enhanced fuel economy that they maybe value at $0.80.*** Wouldn't you then only prefer CAFE over carbon taxes if you valued the marginal dollar of government spending somewhere south of $0.80? That's plausible for me, but is it plausible in a general sample? Well, maybe if they have systematic misperceptions about the composition of the federal budget and where the marginal dollar goes....

HT: Andres Marroquin

* And, in the next draft of their paper, I hope that they make some note of Caplan's pioneering work here. I've found similar results in a New Zealand sample.

** I'm not sure the public's wrong here either. Yeah yeah, we put up some nice revenue-neutral policy with tax cuts. But once a new domain's been opened to taxation, it's a lot easier to ratchet back the income tax cuts while keeping the new tax. So is the problem then voters' lack of trust, or governments' inability to make binding commitments?

*** Numbers here purely for illustration; I've not looked around to see if anybody's tried to estimate this. I know that people value fuel economy and that the value of fuel economy varies with petrol prices, but I've not ready numbers on the costs of increasing fuel economy via CAFE regs.


  1. Following your link to the Whaples 2006 pdf, I find the figure of 47% opposing the minimum wage but not the 87% thinking it hurts the young and unskilled. And your other link is to an abstract of a considerably earlier Whaples survey.

    Where is the 87% figure from?

  2. You are right. The 87 is from the older 96 piece first linked, the 47 is from the 06. Will fix when at the desktop.

  3. Nice post, Eric. You may want to take a look at my recent article summarizing the results of two recent studies. One looks at over 100 published papers on the minimum wage. The other is a micro-micro study with about 600K data points.