In a recent paper published by the Heritage Foundation, frequent AIER contributor Alexander William Salter argues that industrial policy, of the sort that is advocated by Oren Cass and other of today’s national conservatives, is not properly subject to the criticism that this policy will fail because it cannot solve “the knowledge problem.” This policy might fail for other reasons — most prominently, the inevitable noxious influence of interest-group politics — but it will not fail because of an inability on the part of government officials to access enough knowledge to make the policy a success.
It’s useful to quote Salter at length:
Because industrial policy advocates have goals other than economic efficiency, free enterprise objections sometimes miss the mark. A popular criticism of industrial policy is that it is subject to the “knowledge problem.” Formulated by Nobel laureate Friedrich A. Hayek, the knowledge problem asserts the impossibility of comprehensive economic planning because the information required to execute such a plan (the “data” of the central planning problem) is diffused throughout society. Top-down policy cannot harness it. Any attempt to supplant markets with a rationalized, hierarchical planning process is doomed to failure. This is one reason Scott Lincicome and Huan Zhu, in a Cato Institute working paper, question industrial policy. However, they misunderstand the scope of Hayek’s arguments, and, as a result, they are too quick to dismiss industrial policy as impractical.
Hayek developed his knowledge-centric approach to economics in response to the pretensions of early-to-mid-20th-century socialists. They believed they could use the tools of neoclassical economic theory to centrally plan the economy. Specifically, they held that the state could engineer efficiency by producing the quantity of goods that equated marginal cost and marginal benefit (allocative efficiency), using the combination of labor and capital that minimized average cost (productive efficiency). They were wrong, and Hayek was right.
But this debate is about the feasibility of state-led optimal resource allocation, where “optimal” takes its definition from the subjective-marginalist criteria of scientific economics. As we saw, this is not at all the goal of American industrial policy advocates. They have much narrower objectives: increasing employment and output in specific manufacturing sectors. As the author [Salter] wrote in an essay for National Review: national conservatives are forthright in their belief that economic efficiency and the national interest diverge. It’s the latter they’re trying to achieve. While their intermediate objectives differ – some want industrial policy for national-security reasons, others for supporting American families, and more still, because they think it might build a winning political coalition – they agree that manufacturing employment and output should be higher than they are now. There isn’t a Hayekian knowledge problem here. If the government wants to increase the number of factory workers or the output of domestic auto manufacturers, it can.
Salter here uncharacteristically errs. He is mistaken to argue that the knowledge problem as identified by Hayek (and, earlier, also by Ludwig von Mises) does not doom industrial policy to failure.
It’s true that national conservatives’ explicit goals initially appear modest enough to be practically achievable — goals that include (as Salter described these in an earlier op-ed) “more factory workers and more of what factory workers produce.” It’s also true that Cass has proclaimed that his proposal “has nothing to do with the most efficient allocation of resources.”
But these appearances and proclamations are misleading. National conservatives want government to engineer more factory work and more factory output only because they believe that doing so is the best means of improving the economic and social well-being of the people of the nation generally. National conservatives believe that the allocation of resources brought about by the free market is in fact a less-efficient allocation — one that is worse for the country — than is the allocation that would be brought about by their industrial policy. And so achieving their desired allocation of resources is believed by national conservatives to justify the cost of this achievement — that is, to justify whatever is sacrificed to achieve the pattern of resource use deemed best by industrial-policy proponents.
The fact that some national conservatives deny that their goal is economic efficiency reveals only that they fail to grasp economists’ meaning of efficiency. Were they to grasp this meaning they’d understand that “efficient allocation of resources” means ‘that allocation of resources that achieves the maximum possible satisfaction of human wants.’
And so Salter is correct when he observes, about the increased factory work and more factory output demanded by national conservatives, that “[d]irect subsidies, tax credits, and similar policies are fully capable of achieving this.” But he’s incorrect to suppose that the story ends there. The story ends only when we determine if these engineered increases in factory work and output are indeed worth their costs, for only if this juicing-up of the manufacturing sector is reliably determined to be worth its cost can industrial policy truly be said to yield an improved economy.
Because industrial policy necessarily ignores market prices, however, there is no way for the designers of industrial policy, or for the mandarins who implement it, to know if the value of their engineered outcomes — here, more factory work and more factory output — exceeds or falls short of the value of the goods, services, and economic opportunities that are unavoidably sacrificed to achieve those outcomes.
When government engineers more resources into the building, equipping, and supplying of the particular kinds of factories favored by industrial-policy officials, we must ask: From where do these resources come? Some almost certainly come from other would-be manufacturing operations, while others come from the service sector. But no one can know any of these details. Yet even if we did know that, say, X tons of steel and Y hours of labor were diverted away by industrial policy from the service sector (say, from the building and staffing of medical-research facilities and online-retail distribution centers), how can we know that this altered allocation of resources will redound to the country’s net benefit? How can we know that the value of the output thereby lost from these service-sector operations isn’t greater than the value of the output thereby made possible in the manufacturing sector? How can we know that the particular jobs thereby destroyed in the service sector are inferior to the particular jobs thereby created in the manufacturing center?
We can’t know. No one can. There is literally nothing that tells anyone that the net result will be economic improvement for the country. Indeed, the only real knowledge we have when the reallocation of resources is first brought about is that, at least at that time, the market puts a higher value on the service-sector outputs that will no longer be produced than it puts on the additional manufacturing-sector outputs that will now be produced. We know this to be true because, were it not true, market participants themselves would have directed those resources away from the service sector and into the manufacturing sector.
In the face of this reality, industrial-policy champions have only two possible responses if they wish to defend industrial policy as being good for the country. One response is that the market is unreliable and its knowledge distorted. To those persons who offer this response it’s important to put this question: how do you know? What source of knowledge do you have that tells you with sufficient clarity that the knowledge conveyed by market signals is so defective that a government-engineered resource reallocation will improve the welfare of the people of the country?
If you ask this question you’ll get no good answer. Proponents of industrial policy ultimately are guided only by their personal tastes, preferences, prejudices, and hunches.
A second possible response from industrial-policy proponents is to concede that market prices and asset values accurately reflect today’s relative valuations of different outputs and resource scarcities, but then to assert that these prices and asset values reflect only current preferences and knowledge; because (the response proceeds) people today don’t fully appreciate how much better the economy would be with a different pattern of resource use and mix of economic outputs, today’s prices tell us nothing about what the ‘correct’ pattern of resource allocation should be tomorrow. Industrial-policy proponents insist that when government reallocates resources in accordance with industrial-policy plans, only then will market participants come to realize how much better the new resource-allocation pattern is compared to the pattern that would arise absent government intervention.
To this response, too, it’s important to ask the industrial-policy proponent: how do you know? What source of information do you have to assure you that you know better than do your countless fellow citizens, who today spend and invest their own money, what will be best tomorrow for these fellow citizens, nearly all of whom are to you strangers? Again, you’ll get no answer that satisfies. Whatever answer you do get will, upon examination, be seen to amount only to this: “I just have a feeling that I’m right that my industrial policy will improve the country!”
Prices, asset values, and profits and losses determined in open markets in which people spend and invest their own (and only their own) money are the only sources of information in a modern economy about which resource uses are worthwhile and which resource uses aren’t worthwhile. Because of the knowledge problem, no government interventions that censor, ignore, mute, dim, or override these sources of information can credibly promise to improve the overall performance of the economy.
The only way in which the knowledge problem would not be a problem for industrial-policy proponents is if they were to say “We want to achieve our desired particular outcomes and let the rest of the country be damned! We don’t care if the result for the country as a whole is good or bad.” Let industrial-policy proponents say this, and I’ll then agree with Alex Salter that industrial policy is not subject to the knowledge problem.