Bryan Caplan  

An Amazing Consensus among Economists: Not

PRINT
Krugman on Gains from Trade... The Feds' Suit of S&P...
My friend and colleague Dan Klein has written a guest reply to my recent post on economists' consensus.  Here's Dan:

I was surprised to read Bryan's words: "Compared to non-economists, economists enjoy an amazing consensus." Bryan points to a 1996 survey administered to both a set of economists and to a set of ordinary Americans. Yes, the economists agree with each other more than do the ordinary Americans, but not all that reliably. On the thirty-seven questions, the standard deviation of economists' responses is greater than that of ordinary Americans on one-third of the questions (see Table 2 of the JLE paper by Bryan).

Yes, on some questions the economists' standard deviation is much lower, such as on whether technology is beneficial (Q21) and whether the economy is in a state of depression, recession, stagnation, slow growth, or rapid growth (Q37). It isn't surprising that economists agree more than ordinary Americans on questions such as these.

But when we talk about whether economists "enjoy" a consensus, we are naturally most concerned with the most important matters. We are concerned with important issues of public policy. And we are concerned with all important consequences, not just particular effects of particular mechanisms. If you asked a group of people who are trained to think about the particular effects of particular mechanisms, it should not surprise us that they agree more than those who are not so trained. The real issue is whether economists agree on the big questions and overall consequences. Do economists agree on the desirability of major policy reforms?

Fortunately, an abundance of data is available from surveys of economists (Whaples 2006 and supplement; Klein and Stern 2006; Whaples 2009; May, McGarvey, and Whaples 2012; Klein, Davis, and Hedengren 2013). Let's look at the last of those just listed, the one that Bill Davis, Bob Figgins, and I designed and fielded in 2010 to economics professors, with 299 respondents. It asked about seventeen reforms, each reform ratcheting upward the restriction on individual liberty or the activism of a tax-funded government activity. The five-point scale was "support strongly," "support, not strongly," "neutral," "oppose, not strongly," "oppose strongly," so lower-valued responses are more interventionist and higher more laissez-faire. Here are histograms on the seventeen questions:

[View image]

Of the seventeen questions, the one on which economists most clearly show consensus is international trade. That histogram is what strong consensus looks like. But notice how few look like that. Consensus can be attributed to perhaps four: international trade, prostitution, immigration and abortion.

One might expect that a science about the welfare consequences of a policy would generate single-peakedness, even pronounced single-peakedness, as with the international trade question. After all, perfectly random responses would tend toward a uniform distribution. But notice that, in fact, many of the questions show a lack of single-peakedness.

And, yes, the four histograms with consensus lean on the laissez-faire side, which is another matter, and one that Bryan also perhaps sometimes overplays. Notice how on some issues the preponderance--if not quite strong enough to be dubbed a consensus--leans in the other direction.

Mind you, the questions are about a ratcheting upward in the intervention or activism. Besides speaking to the matter of consensus, the results on all seventeen questions speak to attitudes toward governmentalization. For example, only 49.3 percent of economics professors oppose increasing the minimum wage (for the data see "Histograms" here).

The survey also asked about voting, so we can see by how much the mean answer of economists who vote Democratic differs from that of those who vote Republican. In the following figure, blue = Democrats, and red = Republicans:

[View image]

Sometimes the blue Democrat and red Republican bars are about the same height, but frequently they are quite different. The results relate to the recent paper by Roger Gordon and Gordon Dahl,  "Views among Economists: Professional Consensus or Point-Counterpoint?," which spins consensus among the 41 economists selected, apparently by Anil Kashyap, to be on the Chicago Booth IGM Panel as consensus among economists in general, and as a lack of ideological cleavage among economists. But the questions put to the IGM Panel are often very uncontroversial. A Washington Post piece ironically titled "Economists Aren't Politically Driven, Economists Report" explains that the panel's ideological range runs from Emmanuel Saez to Robert Hall.

Finally, about Bryan's speaking of economists enjoying a consensus, here are words from Bill Davis, Dave Hedengren, and me:

The economics profession exhibits greater ideological diversity than other fields (Klein and Stern 2005, 283-286). Is ideological diversity a good thing? The answer, for each of us, will depend on what sort of state of consensus the state of diversity is being compared to. Wherever one stands, one can imagine something worse than a lack of consensus.


I'll reply to Dan later this week.


Comments and Sharing





COMMENTS (6 to date)
Daniel Kuehn writes:

This is the same mistake you made with the EJW article. You're jumbling normative and positive questions. Of course economists disagree on normative questions. Some may have trouble believing this, but in fact economists are human beings. We have a lovely diversity in values and priorities. So we're going to disagree on normative questions.

That doesn't challenge Bryan's point at all - there is considerable consensus among economists. There should be. We wouldn't be much of a science if there wasn't.

Seth writes:

DK - The general public jumbles those types of questions as well, generalizing authority to economists on normative questions based on the scientific cover of positive questions.

For those economists who are relied upon to give answers to normative questions, I don't see a great deal of effort to make this clear.

Perhaps I'm just not seeing it, but if this is true, why not? If not true, I'd appreciate examples where economists offer the disclaimer.

Justin writes:

"there is considerable consensus among economists. There should be. We wouldn't be much of a science if there wasn't."

This nails on the blackboard to actual scientists. I don't know what "considerable" consensus is supposed to mean, but as a computer scientist by profession following the blogosphere for the past few years that statement is laughable. I've found economists to be insightful studiers of and commentators on society, but economics indeed "isn't much of a science". Hayek was right. (Or was he? Is there consensus wrt scientism in econ? I suspect not)

Alexandre Padilla writes:

When an economist is asked a normative question, that economist can indeed have moral/normative values inconsistent (or consistent) with what economics shows (I am thinking about drugs, prostitution, etc) but then this economist should abstain to answer the question or clearly indicate that the answer is guided not by economic considerations but by other non-economic considerations. But I would argue that if you are an economist 24 hours a day as opposed to an 8-5 economist (to borrow from a distinction I heard on Peter Boettke's econtalk interview about his new book. Note that I don't pretend that Peter Boettke would agree with me - this is my opinion), economics should inform your answers even on normative questions. What's the intended (explicitly articulated) goal of the policy (normative) and does the policy achieve the goal (positive)? For example, there are many reasons why increasing minimum wages could be justified but the articulated goal is to help the poor. Does increasing minimum wages help the poor? [note I didn't give an answer]

Surveys are imperfect but I would hypothesize that Dan Klein's questions were addressed to economists and expected answers from people who answered the questions from an economic viewpoint not from a non-economic viewpoint. Maybe what Klein's research shows is that economists don't always rely on economics to express their opinions about normative issues and therefore we should be careful when it comes to listen to experts because experts actually might not use their expertise when it comes to express opinions on normative issues.

Daniel Kuehn writes:

Alexandre -
re: "When an economist is asked a normative question, that economist can indeed have moral/normative values inconsistent (or consistent) with what economics shows (I am thinking about drugs, prostitution, etc) but then this economist should abstain to answer the question or clearly indicate that the answer is guided not by economic considerations but by other non-economic considerations."

How can normative values be consistent or inconsistent with what economics shows? How does economics "show" whether someone should be pro- or anti-prostitution? That doesn't make any sense to me.

Aaron writes:

"What's the intended (explicitly articulated) goal of the policy"
This is actually a positive statement, not normative.

Comments for this entry have been closed
Return to top