≡ Menu

Taking the Knowledge Problem Seriously

Here’s a letter to Ian Fletcher, who e-mailed me yesterday in response to this earlier post of mine.


Thanks for your e-mail in response to my request that you, as a proponent of industrial policy, explain how government officials would get the information they need in order to outperform markets. You respond that “There are a number of answers to this question,” and you then offer five specific ones. Unfortunately, only two answers are real; the other three are red herrings. For example, the fact that “the present international environment” features “aggressive mercantilism” is no answer to the question of how government officials will acquire the knowledge necessary to carry out industrial policy successfully.

One of your two real answers is that governments have access to information “from the Commerce Department” about the size of the trade deficit. That you believe that knowledge of the trade deficit is the sort of knowledge that government officials must access in order to outperform competitive free markets means that you’re unaware of the nature of the knowledge problem that must be solved if industrial policy is to work as advertised.

The knowledge about which I ask is knowledge of which particular goods and services should be produced and, more importantly, also of how best to produce these goods and services. How many computer chips to produce relative to potato chips (and to every other good and service)? And how should each of these goods and services be produced? How much labor to use relative to capital and to land? And which specific materials are best to use today in their production? Which kinds will be best to use tomorrow?

You perhaps think that the other of your real answers to my question suffices as a response – namely that these decisions will be made, as you put it, “by applying the logic of technology.”

Alas, this answer of yours is precisely 100 years behind the times. In 1920 Ludwig von Mises’s published a famous paper showing that, because nearly every good or service can be produced using many different methods and with countless different resource combinations, learning which are the economically best uses of resources is not a problem of technology. Instead, it’s a problem of relative economic valuation. It’s a problem of acting in accordance with the relative scarcities of innumerable possible inputs, from asphalt to asparagus and from zinc to zippers. Mises explained that an economy that fails to act in reasonable accordance with these relative scarcities can satisfy no more than a tiny handful of human desires. He showed also that market prices are indispensable conveyers of this information: prices lead producers to use fewer of those inputs whose prices have risen and more of those inputs whose prices have fallen.

Insofar as it operates, industrial policy replaces competitively set market prices as guides to resource allocation with bureaucratic commands. Industrial-policy advocates presume that government officials not only can learn all the relative resource scarcities that are at every moment taken account of by producers responding to market prices, but can acquire knowledge of these resource scarcities that is greater than is the amount of knowledge used in markets.

What you, Oren Cass, Henry Olsen, Marco Rubio, Elizabeth Warren, and other advocates of industrial policy must explain is how this knowledge – what Hayek called “knowledge of the particular circumstances of time and place” – knowledge that is inconceivably vast in volume, ever-changing, and dispersed across the globe and amongst billions of minds – will be gathered by bureaucrats and used by them productively. So far, neither you nor any other advocate today of industrial policy has shown even an awareness of the nature and reality of this problem. And you certainly haven’t taken the first step toward offering a credible explanation of how you propose to solve it. Until you do, no proposal you offer for industrial policy should be taken seriously.