Some commenters still seem confused about the locus of disagreement in this week’s back-and-forth with Paul Krugman. I post today not to beat a dead horse, but to clarify the issues for those who are interested in understanding them. Please keep any discussion both civil and on-topic. I’ve numbered the points below for easy reference.
- Government expenditures have benefits, and taxation has costs. Paul Krugman, I, and all other economists agree that the most efficient policy regime is one that equates all of those benefits and costs at the margin. Here’s why: If the marginal benefit of, say, cancer research, exceeds the marginal benefit of, say, national defense, then you ought to be transferring dollars from defense to cancer research. If the marginal cost of a tax on, say, onions, exceeds the marginal cost of a tax on grapes, then you ought to be switching the burden of taxation from onions to grapes. If the marginal benefit of cancer research exceeds the marginal cost of taxing grapes, then you ought to raise the grape tax to fund more cancer research. And so on, until all of the marginals are equalized.
Call this the equimarginal principle. (This is what Krugman calls the Ricardian analysis.)
(Notes: “Marginal” means that we’re looking at the effects of small changes. The marginal cost of a tax includes the disincentive effects of that tax.)
- Suppose you’ve successfully set up policies that equalize all of these marginals. Call the marginals A, B, C, D and E, where A is the marginal benefit of disaster relief. Now a disaster comes along, which increases that marginal benefit. Then you want to spend more on disaster relief, which means either lowering some other expenditures, or raising some taxes, or both. This will change the relevant marginals; if B is the marginal cost of taxing onions and if you change the tax rate on onions, then B will change. But in view of the equimarginal principle, you want to keep all of these marginals equal. If they’re all equal to begin with, and if you’re going to change one of them, and if you want to keep them equal, then you’ve got to change all of them. Thus in this case the equimarginal principle dictates changing all categories of government expenditure and changing all taxes — essentially, spreading the cost of disaster relief as widely as possible.
- Eric Cantor has said that he wants any new disaster relief to be funded by cuts in expenditures, without any cuts in taxes. On Tuesday, Krugman raised the question: Is there any reasonable worldview under which Cantor’s position makes sense?. This is not the same as the question: Do we approve of Cantor’s policy position?, and I hope commenters will keep that mind.
- Krugman’s post was quite ill-considered (I commented here), but fortunately for him, the majority of his commenters (both pro and con) seem not to have understood it. They thought Krugman was arguing against Cantor’s position, but he wasn’t: He was arguing something much stronger, namely that there are no legitimate assumptions under which Cantor’s position makes sense, and that this was true as a matter of theory.
- Krugman proceeded by explaining points 1 and 2 above (perhaps a little less clearly than I’ve explained them here). He then attributed to Cantor the belief that all policies are currently optimal, and concluded, by invoking points 1 and 2, that Cantor’s position makes no sense.
- I blogged on Wednesday to the effect that if you attribute to Cantor a different belief — namely that current policies are very far from optimal — then his position can make perfect sense. Namely, if you believe that some programs are very wasteful, those are the programs you should cut first. I cannot imagine how anyone can fail to see that this is a logically consistent position.
- A few commenters popped up to say “Well, Krugman was giving Cantor the benefit of the doubt by assuming he had employed a sophisticated economic theory”. But that’s a grossly misleading picture. Krugman assumed both that Cantor had employed a sophisticated economic theory and that Cantor believed something (namely the optimality of current policy) that neither Cantor, I, Krugman, nor anyone else I know of believes. The gist of his column was “If Cantor believes that all policies are optimal, then his position on funding disaster relief makes no sense.” One might as well point out that “If you believe that the tooth fairy will cure all your dental problems, then going to the dentist makes no sense”. This is not a good way to prove that going to the dentist is irrational, especially when nobody has ever suggested that there might be a tooth fairy.
If you want to give Cantor the benefit of the doubt, as those commenters claim Krugman was doing, then you start by assuming that Cantor means what he says: That a lot of current spending is wasteful and should be cut. Then you note that whether or not you agree with that position, it is certainly not an incoherent one.
- I want to be crystal clear here about the locus of the disagreement: Krugman says that under no assumptions does it make sense to cut expenditures without raising taxes. I said that under some assumptions it makes perfect sense —- for example, itt makes perfect sense under the assumption that some expenditures are very wasteful. That’s the main issue. This was not an argument about the validity of the assumptions, but about what follows from them.
- I made the side point that sometimes it makes sense to hold even a good policy hostage in order to get action on other good policies. I stand by that position, but I (slightly) regret bringing it up, since it distracted attention from the main point. We can reasonably disagree about this side issue, but I do not believe we can reasonably disagree about the main point.
- Yesterday, Krugman acknowledged that I was correct regarding the main point. Specifically, he wrote:
If you think the government’s priorities are all wrong, then theory doesn’t tell you much about what should happen.
Bingo. That’s most of what I was saying, exactly. It’s not a deep point; it’s completely obvious, and the fact that Krugman overlooked it at first only goes to show that we all make mistakes. His acknowledgement of that mistake is what I’d expect from any honest academic, and makes me glad that I’m in one of the few professions where we can expect that kind of honesty.
- Krugman deflected attention from this concession by objecting to my subisidiary point about holding policies hostage. (He goes so far as to say I “missed the point”, though this point never arose in his original blogpost.) That’s a separate discussion. But note that Krugman’s position on the subsidiary point directly contradicts his (now abandoned) position on the main point. He writes:
He’s denying disaster relief to people hard hit by a hurricane. That is, he’s holding suffering Americans hostage to his goal of smaller government.
By making this argument, Krugman tacitly acknowledges that any debate about disaster relief (and how to fund it) has to be based on the specific benefits of disaster relief (and other programs), not on general theoretical principles. That, again, was the whole point.
- As a side note, Krugman objected to one of my metaphors:
Where Landsburg really goes where he shouldn’t, though, is by comparing Cantor’s proposal to denying someone goodies unless he shapes up elsewhere — he uses the example of a teenager who won’t be allowed to go to the prom unless he does his chores.
Though I hate to be more combative than necessary, this strikes me as pretty weaselly (to use Bob Murphy’s formulation). A metaphor is good or bad depending on how it is used. I used this metaphor only in the sense that disaster relief and going to the prom are both good things. Their levels of importance or frivolity has no bearing on how the metaphor was employed.
- Some commenters have said: But it is an important part of the metaphor! If you were going to the hospital to fix a broken leg, you wouldn’t stop to debate how you were going to pay for it! Those commenters miss the point that whether it’s a prom, a fracture, or a mission to save the world, it’s still got to be paid for, and the issue of how to pay for it is still going to arise. And the issue of whether to spread that cost over multiple activities is still the same issue.
Moreover, as soon as you say “what if it’s a broken leg?”, you’ve conceded the whole main point. The whole main point — for the 99th time — is that pure theory can’t dictate policy; your beliefs about the values of different expenditures have to dictate policy. If you’re arguing about the importance of fixing a broken leg, you’re arguing facts, not theory. Welcome to my side.
Finally, let me say this: I’ve been blogging now for almost two years, and have been constantly delighted and astonished by the quality of the comments around here. Try looking at Krugman’s blog to see how much worse it can be. One of the things I value most in these discussions is that we generally try to stay focused on the topics at hand. In this case, the main topics at hand were:
1) Are there or are there not any assumptions under which Cantor’s argument makes sense? (Krugman said no, I said yes, Krugman now says yes.)
2) Did Krugman’s initial argument implicitly assume that the initial policy regime is ideal? (I said yes, Krugman now says yes.)
3) Does this make Krugman’s initial argument pretty much entirely inapplicable? (I said yes, Krugman now says yes.)
There are also, as always, subsidiary issues worth discussing. But “Is disaster relief called for?” is an entirely separate question, and I’m glad we’ve largely managed to avoid getting distracted by it. It’s an interesting question, though. Maybe I’ll blog about it soon.