We’re reading Hayek’s famous paper, “The Use of Knowledge in Society,” in our monthly reading group tomorrow. I’ve never been convinced by Hayek’s argument, and get less convinced every time I re-read the paper. I don’t have time to work out a full response to the paper, so here, for whatever it’s worth, is a quick laundry list of objections to be developed at some later date.

1. Hayek takes his paper to be a critique of centralized government planning in a mixed or socialist economy, but as Kevin Carson deftly argues, the argument applies with equal cogency to centralized corporate planning in a capitalist economy. To the extent that Hayek’s argument is, by design, an argument against socialism, it’s unwittingly and unintentionally an argument against capitalism as well. Ultimately, it’s an argument against large-scale economic planning as such–not at all what Hayek intended.

2. The argument proceeds as though “local knowledge”–that is, “knowledge of the particular circumstances of time and place”–always (or often) trumps non-local knowledge. Hayek gives no reason for thinking this, and it strikes me as pretty implausible.

(a) Moral knowledge as non-local. General moral truths about rights, responsibility, virtue, and justice are not cases of local knowledge (just the opposite), but arguably trump judgments based purely on local knowledge. A manager may have better local knowledge than a judge or jury of a given worksite. Yet a judge or jury may reasonably be given the moral and legal authority to nullify the manager’s decision if the decision violates someone’s rights.

(b) The myopia of the local. Even as a purely epistemic matter, there’s no reason to think that the person with greater local knowledge knows more or better than the person with less. The possession of greater local knowledge is perfectly consistent with being the victim of the epistemic and pragmatic equivalent of myopia; sometimes the person farther away from the scene knows more and better.

Example: An OR janitor spends more time in the operating suite than the hospital’s Director of Surgical Services, and knows the physical terrain of the suite far better than she. But it doesn’t follow that, all things considered, an OR janitor knows how to run an OR better than the Director. It doesn’t even follow that a given OR janitor knows more about the janitorial side of the OR than a given Director. The Director may be able to integrate facts at the right or relevant level of abstraction in a better way than the janitor mired in the epistemically hyper-local.*

In short, locality is epistemically valuable, but locality comes in degrees, and most local doesn’t map directly onto most valuable. The relevant epistemic value is relevant locality. But having relevantly local knowledge is often a matter of being able to combine local knowledge with considerations unavailable at the local level. There’s more to local knowledge than mere locality. 

(c) Epistemic proficiency as multi-dimensional. Though Hayek says correctly that not all relevant knowledge is statistical, some is. When statistical data is relevant to a given decision, statistical knowledge comes to have overridingly high epistemic weight. In such contexts, those who lack statistical competence have less of relevance to add to the decision, no matter how proximate to the scene they may be. 

Take a case in which two persons, Smith and Jones, are looking at a spreadsheet relevant to some business decision. Suppose that Smith is well-versed in statistics but operates at considerable physical distance from the target location where the decision is to be made; Jones, by contrast, is innumerate but physically proximate to the target location. Hence Smith is superior to Jones along one epistemic dimension, the statistical, and Jones is superior to Smith along another, the physical. Hayek says nothing to preclude the possibility that there are decisions such that Smith is better positioned to decide than Jones. But it’s obvious that there are.

Friedrich Hayek (1899-1992); photo credit: Wikipedia

A very clear instance of the preceding problem is risk. Suppose that the spreadsheet just mentioned contains probability figures for risks that might arise from a decision made at a given physical site. Suppose, however, that the risks in question would arise at a location physically remote from the site rather than at the site itself. In this context, it’s beside the point that Jones has better physical knowledge of the site than Smith. If Smith understands probability theory, and Jones doesn’t, and Smith can grasp how decisions at the site affect people and places beyond it, but Jones doesn’t, then Jones’s vaunted local knowledge not only doesn’t matter, but constitutes an epistemic impediment to good decision-making. Nothing in Hayek’s paper acknowledges this, but it’s crucial, not only to economic decision-making, but to pricing, the idée fixe of Hayek’s argument.

The underlying problem in both (2b) and (2c) above is that Hayek hasn’t taken seriously the possibility that epistemic proficiency is a multi-dimensional affair; given that, he’s said nothing about how to weight competing dimensions in a given context. What he’s done instead is to suggest by indirection that local knowledge has overriding weight in an unspecified set of situations without being able to explain how, where, or why. 

3. Hayek’s identification of “local knowledge” with “particular circumstances of time and place” is overly narrow and overly physical in focus. He seems to have in mind the knowledge that might be possessed by a capitalist manager of the physical features of a factory or warehouse, by contrast with the knowledge possessed by a government bureaucrat overseeing facilities of that kind. But the relevant issue is not physical locality; it’s epistemic proximity. For any given decision, some information will be relevant, and other information not. The question is not who is most physically proximate to the place where the decision is to be made?, but who is most epistemically proximate to the information relevant to the decision? Hayek’s answer to the first question is not a reliable answer to the second. But the second is far more important than the first.

To illustrate this, consider a common decision in health care. Suppose that a hospital performs a set of procedures on a patient and bills the patient’s insurance $10,000. The insurance company then denies the charges on the grounds that the procedures were medically unnecessary. The hospital in turn initiates an appeals process with the insurance company, eventually exhausting the options within the company’s internal appeals process, so that the decision then has to be sent to a third party for review.

Assume that the third party mediator is physically farther away from the hospital than either the representatives responsible for the insurance company’s denials/appeals process, or the hospital’s own denial managers. But suppose that the third party auditor has greater impartiality than either the insurance company reps or the hospital’s denial managers. In that case, the auditor may lack physical proximity, but has greater epistemic proximity to the facts: it tracks relevant truths more reliably than more physically proximate actors. Hayek ignores this, but it sits in tension with a major theme of his paper, i.e., the epistemic supremacy of the local. In other words, Hayek’s fixation on physical locality ignores the fact that physical locality is sometimes irrelevant to economic decision-making. 

4. Finally, Hayek seems to be equating the “right” decision with the most economically efficient decision, so that relevant features of any decision must of necessity be captured by a free market price system (see in particular section VI of the paper, which explicitly makes this point). But it seems obvious that some economic decisions involve values that are either beyond price or otherwise not susceptible of symbolic representation by prices (e.g., honesty, fairness, professional integrity, the refusal of complicity in injustice). I don’t see how Hayek’s argument handles any of these considerations, but his failure to handle them seems highly problematic.

In short, Hayek’s argument strikes me as a lemon–with some, but relatively little, lemonade to be made from it.

*I don’t mean to be offering these examples as generalizations about janitors and directors, but as live possibilities that, when actualized, are counter-examples to Hayek’s thesis. Some of what Hayek says is true: front-line workers often know more about front-line processes than bureaucrats operating at a distance from them. But this is as true of capitalist managers as socialist planners; the epistemic differences between them are often just nominal. And it’s a truth counter-balanced by the equal and opposite truth that there are times when front-line workers don’t know more than people at a distance from front line operations, but think they do simply because they are on the front line.

For useful discussion, see David Kelley, “Rand Versus Hayek on Abstraction,” Reason Papers 33 (Fall 2011), particularly sections 3-5 (pp. 15-24). Though Kelley doesn’t explicitly discuss practical reasoning, his discussion makes clear that physical proximity is neither necessary nor sufficient for abstracting in such a way as to facilitate practical reasoning. The relevant issue is whether the reasoner tracks the right similarities relevant to the decision, a skill not guaranteed (or even necessarily probabilized) by proximity to a given physical location.