## The Problem of Economic Calculation

13 April 2008

Let's say that we might make veeblefetzers, and we have three available processes:

process adamantium per unit  vibranium  per unit
A 200g 300g
B 300g 200g
C 300g 300g

Now, we have two questions:

• How many veeblefetzers should we make?
• Which process should we use to make veeblefetzers?
From our table we can answer part of the second question: If adamantium and vibranium are actually goods (rather than something otherwise harmful), then we just don't want to use process C; compared with process C, process A would save us some adamantium for other uses, and process B would save us some vibranium. So (again on the assumption that adamantium and vibranium are goods), technical considerations tell us not to use process C.

But it gets messier when we attempt to decide between process A and process B; we have to know whether we want adamantium more for other things than we want vibranium. And there's basically the same question in deciding whether to make veeblefetzers. Could the adamantium or the vibranium be put to better use than in making veeblefetzers? (It gets even more complicated if the processes don't simply scale linearly, so that doubling inputs doesn't double outputs, or new veeblefetzers get put to decreasingly important uses as we make increasing quantities, or we start to take adamantium or vibranium away from increasingly important things as we make more veeblefetzers.)

What we want are numbers or number-like things that are assigned to veeblefetzers, adamantium, and vibranium, so that we can compare those numbers or number-like things, and know whether it's better to save adamantium or vibranium, and whether the priority should be to make veeblefetzers or to do something else with the adamantium and vibranium. Those numbers or number-like things are prices. Market prices are prices assigned by a market process, but any prioritization implies a corresponding system of prices, explicit or implicit. Where there aren't any prices (explicit or implicit), there isn't a system of priorities.

If prices weren't and couldn't be set for us by a market process, then how should they be set? One student, when given a hypothetical example like this of veeblefetzers, adamantium, and vibranium, demanded to know why I didn't use familiar, real-world products and inputs. The answer is that we have a sense of how the world does price things such as watches, aluminium, and gold. Further, aluminium was once more precious than gold, and a decent watch was once unattainable. Relative priorities change, and with them prices. I wanted him to consider pricing from scratch.

In attempting to price adamantium and vibranium, part of what would obviously need to be considered would be the other uses to which one might put each. Were one going to price adamantium or vibranium rationally, then one would have to answer the same sorts of questions for those alternate uses as one wanted to answer for the production of veeblefetzers. In other words, one would have to price the alternative products, and the other inputs (besides adamantium and vibranium) in those other processes. That means looking at other products and processes for those inputs, with no direct involvement of veeblefetzers, of adamantium, or of vibranium. To price anything rationally, one would have to price everything rationally.

One might in theory construct tables of processes for other goods and services. In many cases, one could quickly discard some processes for the same reason that we discarded process C. And, were adamantium or vibranium itself something that must be produced, and not something that a consumer would want in and of itself, then perhaps one could replace it in one's calculations as a function of other inputs.

But the presumption that one could assemble all that technical information — and it must be assembled (not merely collected, but organized) before boffins and wonks can feed it to their super-computers — is truly heroïc, even if one assumes purely automated production. (More on that assumption in a bit.)

Even then, what would be produced would not be prices, but a system of equations and inequalities expressed in terms of unknowns, corresponding to relative valuations of consumer goods and services. One would need to know, at various levels of consumption, whether each participant would rather have a bit more water or a bit more warmth, a bit more warmth or a bit more clothing, and so forth, before one could compute prices rationally or (equivalently) allocate production and distribution rationally.

And, actually, this matter of human preferences cannot be postponed quite like that, because the assumption of complete automation is counterfactual. Human beings may not be factors of production in the making of veeblefetzers, but they are factors of production in quite a bit else. So, to describe production, not only would one have to know how many whoozits can be produced by a purely automated procedure; one also would have to know exactly how specific, real-world human beings (who are very much a part of real-world production) will respond to specific, real-world incentives.

Markets establish prices by concurrent processing. Knowledge is left dispersed; private data, such as preferences, are expressed in the behaviour of individuals participating in the economy. When the price for a given good or service is below an equilibrium, would-be sellers withdraw or try to negotiate a higher price while would-be buyers pursue the good or service in frustration (showing would-be sellers that they might get a higher price) and perhaps offer a higher price; when the price is above an equilibrium, these rôles are largely reversed. So prices are being pushed towards the equilibrium of the moment. Meanwhile, the prices of each good and services is effecting how much people are attempting to buy or sell of other services.

A general equilibrium would never actually be obtained in a real world, because preferences are always in flux, and resources, including technology, change in ways that cannot be particularly well predicted. Still, a market would be in constant pursuit; the process of correction beginning immediately as new information were introduced. And a notable mathematical result is the Coase Theorem: If property rights were clearly defined and respected, then in the absence of transactions costs a market equilibrium would always be economically efficient.

Socialism — the doctrine or practice that the means of production should be owned by the community per se and administered for the over-all benefit of that community — is now-a-days primarily associated with claims about fairness, but in its heyday, and to some extent still to-day, it entailed a belief that technocratic planning could outperform the market by some combination of speed, accuracy, and the avoidance of transactions costs. In the context of production which is assumed to be dramatically more efficient, it becomes easier to talk about the relationship of workers to work changing, and about different patterns of distribution. But the problem here was and remains that socialists rarely recognize the problems of efficiency beyond technical efficiency. In effect, they presume that the differences between processes A and B must ultimately be no different from those between either and C. One often even finds socialists who deny the relevance of prices to socialist planning, betraying some failure to understand the general concept of price. The great problem for socialism is that it must price but, because pricing involves more than technological efficiency, has no sensible system for pricing in the absence of markets. In the absence of sensible pricing, rather than getting wildly more efficient production, one gets horrific chaos and waste.

In practice, socialisms that attempted to avoid having their own markets largely adopted the prices of the more market-oriented economies that they could observe. They had some sense of the present relative values of aluminium, of gold, and of watches because markets elsewhere had valued them. But those prices are prices appropriate to the context in which those markets prevailed; if markets had prevailed in the imitating community, then they would set different prices corresponding to that context. Further, the technocrats can't give the game away by fully imitating the prices of market economies — that would make their own economies more overtly caricatures of market economies. So pricing in communities without their own markets, while not typically being as awful as it might be, is still markedly worse than that in less socialistic nations.

There have long been some socialists who recognized the need for market prices, and such socialists started to become relatively common in the 1980s. The problem for these attempts to combine markets with socialism is that, to the extent that the resulting system behaves like a market, it behaves nothing like socialism; and, to the extent that it behaves like socialism, it behaves nothing like a real market — it doesn't coördinate private and dispersed knowledge to form prices.

(Neoclassical economics hasn't had a good handle on these issues, because it typically assumes-away the underlying problem of information being private and decentralized.)

That's not to say that markets should be used for all economic allocation. The real world involves transactions costs. Sometimes the transactions cost of markets are sufficiently high, and the transactions costs of alternative institutional relationship — contracts, firms, &c — are sufficiently low that the difference then more than off-sets the lost informational value of market prices. But one should remember that real-world alternative institutions do entail transactions costs, and indeed sometimes these are even higher than those associated with markets.

This site uses Akismet to reduce spam. Learn how your comment data is processed.