Calculation Problems vs. Incentive Problems

I recently started listening to Bryan Caplan and Peter Boettke's debate on the Austrian School, and something dawned on me when Caplan started talking about the socialist calculation problem. I agreed with Boettke rather than Caplan on the question of the calculation problem in the Soviet Union (and on several other points, actually), but Caplan did raise the question of the centrality of the calculation problem to the Austrian School, and I think this is the source of a lot of unnecessary conflict with Keynesians.

As I alluded to in an earlier post on Hayek and Stiglitz, the "socialist calculation problem" has never really interested me that much. It just strikes me as so obvious, and I fully recognize that a lot of that is due to my age. As far as I can tell, von Mises and Hayek made comments about the possibility of socialism in the 1920s and 1930s that were not necessarily on everybody's radar, but weren't exactly controversial for economists either. The alternative view gained a modicum of respectability only when Oskar Lange and Abba Lerner in the 1930s provided an ex post justification of socialism that essentially dependend on a benevolent dictator orchestrating a traditional neoclassical economy. The "socialist calculation debate" seems to have only really taken off when Lange and Lerner provided Mises and Hayek with someone (serious) to argue with. Others of course joined Lange and Lerner. Lawrence Klein barely even tries to hide his socialist sympathies in his writing (although I respect his exposition of Keynes). My point is that as far as I can tell, this is a debate that only started raging after my primary influence (Keynes) had made his impression on the discipline, and wrapped itself up before I was even born and with all the modern economists that I admire coming down against Lange and Lerner. Moreover, none of the economists in the interim that I admire - Phelps, Friedman, Hicks, Modigliani, Minsky, Coase, Williamson - really succumbed to the Lange-Lerner infatuation with socialism. I can certainly understand why someone who grew up during the Cold War could maintain an interest in this skirmish even into the twenty-first century, but it doesn't hold any particular interest for me, because it seems so eminently obvious and conclusively decided.

Nevertheless, Austrians still regulalry point to decentralized knowledge and the calculation problem in their critiques. When I'm speaking with Austrians I often have to remind them that we have no difference of opinion on these questions (indeed, I think that most economists have no difference of opinion with them on these questions). One of the major impediments to good dialogue, I think, is that many people confuse "calculation problems" with what I will call (for lack of a better term) "incentive problems". Most of the disagreements today revolve around "incentive problems" and we're all on the same side when it comes to "calculation problems", but a lot of people argue as if the issue were still a calculation problem.

Take fiscal stimulus. The argument has never been (so far as I can tell - correct me if I'm wrong... certainly it's not the argument that I ever make) that the government can more efficiently allocate resources than free individuals. It has always been recognized that the market economy does this better. The concern is that incentive problems introduce stable biases to market outcomes. Asymmetric information isn't a challenge to the idea of decentralized knowledge and market calculation - it's a claim (in econometrics-speak) that decentralized information is going to efficiently target a biased solution. The same with the Keynesian theory of output, which can be stable at sub-optimal levels because output is jointly determined by loanable funds market equilibrium and liquidity preference equilibrium. As Keynes writes:

"To put the point concretely, I see no reason to suppose that the existing system seriously misemploys the factors of production which are in use. There are, of course, errors of foresight; but these would not be avoided by centralising decisions. When 9,000,000 men are employed out of 10,000,000 willing and able to work, there is no evidence that the labour of these 9,000,000 men is misdirected. The complaint against the present system is not that these 9,000,000 men ought to be employed on different tasks, but that tasks should be available for the remaining 1,000,000 men. It is in determining the volume, not the direction, of actual employment that the existing system has broken down."

I think Peter Boettke makes this mistake of calling an incentive problem a calculation problem at minute 4:40 here when he tries to grapple with the idea of "market failures". To him, "market failures give rise to arbitrage opportunities", and so Stiglitz and Akerloff and guys like that (allegedly) completely miss the point of how markets clear. But this misunderstands the problem of market failures. Market failures occur precisely where arbitrage cannot solve problems. If arbitrage could solve the problem, the market wouldn't fail - it would succeed! What form of arbitrage can possibly solve a problem of negative externalities? The problem with a negative externality is not with the availability or use of knowledge, and it's not with the will to act on that knowledge in the pursuit of self-interest. The problem is precisely that entrepreneurs acting on their decentralized knowledge and seeking out arbitrage opportunities to satisfy their self-interests will converge on sub-optimal solutions as a result of the institutional structure in which they are operating.

There are two relevant questions, I think. First, is a "market failure" a genuine market failure or is it, as Boettke says, an opportunity for arbitrage? If it's just an opportunity for arbitrage then "market failure" is a misnomer. If it is a genuine market failure, we then of course have to ask ourselves what the best solution is. In some cases a public solution will be appropriate, and in some cases a private solution will be appropriate. It is the article of faith of no economist that I am aware of that a public solution is necessarily a good solution. The point for me is that none of these claims rest on the idea that Lange and Lerner were correct in the calculation problem debate. The underlying assumption is that they were wrong - the question is, does the admitted fallibility of the state preclude a public solution or doesn't it.

If the problem isn't a calculation problem at all, but an incentive problem, there may be certain areas where a public solution could be quite successful. For example, let's take the case of the negative externality of carbon emissions. Some cost-benefit analysis aficionado may think it's important to determine precisely what the social cost of carbon consumption is, but that's not really an economist's perspective. The economist simply understands that choice on the basis of decentralized knowledge would converge on the "true cost" of carbon if it had an incentive to, but that the lack of property rights to the air we breathe removes that incentive. The government doesn't have to know what the right solution is to move society in the direction of the right solution. If the property rights arrangements introduce a negative externality then a moderate correction in the direction of raising the cost of carbon will be an improvement, regardless of whether the government knows the true social cost of carbon. Of course, what constitutes a "moderate correction" is going to be a matter of debate, but if we're confident that what we're dealing with is a negative externality then doing nothing is certainly inferior to a small carbon tax.

The important thing to note here is that the carbon tax (a specific example of a "Pigovian Tax") absolutely relies on the assumption that Lange and Lerner were wrong on the socialist calculation debate. These sorts of market failure arguments are driven by the belief that free markets use decentralized information efficiently. They aren't correcting market soltuions, they're correcting institutional arrangements that distort the incentive structures that markets respond to.

There's obviously still considerable scope for disagreement here. People are going to come down all over the place on our certainty regarding market failures and on the two questions that I posed earlier that we have to ask ourselves when thinking about market failures. The point is, it's erroneous to think of this as a "calculation problem" debate. We are on the same side of that debate. It is an "incentive problem" disagreement, not a calcualtion problem disagreement.

UPDATE: Mario Rizzo has a great post on this aggravating tendancy to simply list cognitive biases (or market failures), and then take this myopic perspective that nothing must work right. My perspective is this: market failures and cognitive "biases" are very important, but (1.) a lot of them cancel each other out, (2.) a lot of them bias outcomes, but only minorly, (3.) we don't really notice these problems because we don't know what an "optimal" universe looks like, so things don't really appear that bad, and (4.) even if there is sub-optimality, a free society still progresses for the same reason that it progresses when we assume away all these biases and failures. So for me, these biases and failures are incredibly important for understanding the difference between where we're at and where we could have been - but they don't change the fact at all that generally speaking a free society is going to guarantee progress regardless of whether we could have been further along had these biases and failures not been around.