Friday, February 11, 2011

Aggregating Preferences

. Friday, February 11, 2011

Every time a new poll of voter attitudes towards spending and taxes comes out, liberal commentators gleefully point out just how irrational the American public is. They seem to always want lower taxes and higher spending, but also a lower deficit/debt. Here, for example, is Krugman saying "The conclusion is inescapable: Republicans have a mandate to repeal the laws of arithmetic." And here is the Washington Post making their readers dumber:

Budget cutting is a top priority for the GOP, with 70 percent of Republicans in a new survey by the Pew Research Center saying the federal government should focus on reducing the deficit, not new economic stimulus. And in many cases, more Republicans now support cuts than did so two years ago.

But across 18 areas of federal spending, a majority of Republicans support decreasing spending in just one: aid to the world's needy.

This is stupid. For one thing, most (public) foreign aid does not go to the "needy poor", but rather to the military budgets of Israel and Mubarak's Egypt. There are perfectly good reasons for a pretty big consensus on cutting some of that spending. But more fundamentally, it is perfectly rational for a majority of individuals to support deficit reduction, while simultaneously opposing all means of doing so. Karl Smith points out one reason that is the case:

Suppose that you have three people. Adam, who believes in cutting spending to balance the budget. Bill who believes in raising taxes to balance the budget and Chris who believes that state balanced budgets are a pro-cyclical economic destabilizer that should be alleviated by federal transfers, or as he likes to say, “lame.”

Now we are going to ask a few questions.

First we ask: Should the state stick to its balance budget requirement? Adam and Bill say yes. Chris says no. We confidently conclude that the public wants balanced budgets.

Second we ask: Should we cut spending? Adam says yes. Bill and Chris say no. We confidently conclude that the public doesn’t want to cut spending.

Third, we ask: Should we raise taxes? Adam and Chris say no. Bill says yes. We confidently conclude that the public doesn’t want to raise taxes.

But wait a minute! Is the public insane! How can we balance the budget if we don’t cut spending or raise taxes.

No the public as individuals are completely sane, but when aggregated into a whole they become irrational. And, importantly there is no clear way to make them rational, since each person is answering truthfully and with complete knowledge of the facts.

That, as Krugman should know, is arithmetic. We should probably even expect this. When we ask a Yes/No question, we are almost guaranteed to get a majority in favor of one or the other. (Depends on whether or how strong of an option "Don't Know" is). In this case, that's a strong "Yes" in favor of deficit reduction. But if we then follow that up with a string of options of how to do that (18 in the WaPo survey of Republicans linked above, 13 in the Pew survey Krugman linked to), the broad-category majority will often -- even usually -- dissipate into a minority in each narrow category.

Here's a simple analogy to drive home the point: If you ask Americans if they like baseball, a majority will say yes. If you then ask them what their favorite team is, no team will get a majority. However, it would be daft to conclude from this that peoples' preferences over baseball are irrational.

Sometimes those options are nested, and when combined end up a majority (or even a supermajority, as in this example from Tyler Cowen). Sometimes they are not, or will not. But differences in the preferred means used to achieve an agreed-upon end is not a sign of individual irrationality, or even innumeracy.

You can sometimes get around this by changing the questions asked. Instead of asking "Yes/No" on a broad category then "Yes/No" on a series of smaller categories, for each category ask some variant of "If cutting Program X was the only way to balance the budget, would you want to do it?" I'll bet you find majorities pretty quick on many of those questions.

Final note: The Pew table that Krugman reproduces shows that in every single program, more people preferred cuts in 2011 than 2009.


Patrick said...

While I completely agree with your overall point, there's certainly a good bit of research that shows that lots of people don't know what they're talking about when it comes to the taxes/cuts question (or certainly that they know less than Adam, Bill, and Chris).
Recent evidence: the piece by Suzanne Mettler on public ignorance about state benefits (

Kindred Winecoff said...

Patrick, while I certainly agree with your overall point, I found this table to be a bit misleading when it was first printed. For example, in the most extreme cases (basically the top half of the Mettler table) we're talking about tax deductions. Many people do not associate tax deductions with a government spending "program", instead thinking of it as "letting me keep more my own money". I actually agree with that: a spending program is qualitatively different from a tax deduction, and from my experience (who knows if it's representative) the only people who consistently don't are political scientists who focus on American politics.

My broader point was really that we shouldn't expect polls to align the way that Krugman, e.g., seems to think they should.

Aggregating Preferences




Add to Technorati Favorites