Thread by @VincentGeloso

Thread: This is a poor reading of Hayek.
The Use of Knowledge in Society (AER, 1945) does not say what Acemoglu *thinks* it says. Prices are not aggregative devices. They are knowledge-economizing devices that are cheaper than *any* other devices. #econtwitter

https://twitter.com/DAcemogluMIT/status/1660269222621069315

The problem starts with fact that there is a great deal of knowledge that is simply tacit and impossible to code. Moreover, the *value* of that tacit knowledge is impossible to code as well. For example, riding a bike is hard to express into a book/code. So is coding its worth

But, the time-price, the price of an instructor, the price of a parent's time etc. are going to express the value of using this knowledge. Prices economize on all the knowledge we need to collect about bikes. They tell us how much knowledge is necessary.

This may seem like a boring point but it changes entirely the meaning of prices as a *technology*. To plan prices, you must collect *more* information. You need to use more inputs because you arent being told which knowledge-inputs matter or how to use them efficiently.

And I am not even allowing for *errors* in planning here. I am just talking about the cost and efficient use of knowledge.

The reason why I love econometrics but hate its use for planning is that it is so much costlier to plan than the only thing you can do is use *past* observations to predict prices. Its the cheapest way to collect *all the information* (rather than the relevant one).

But its still pretty bad because you need to collect the data, weigh it, employ resources to assemble etc. And its still only going to be a rearview mirror of *past* efficient uses of knowledge.

So, its not costlier *now* (at t_1) to collect *all* the tacit knowledge in the world. Its also growing costlier at the margin because the best way to do so is still a rearview mirror approach whose productivity erodes with distance from t_1 (and productivity is tied to cost-duh)

Think of my point in the sense of a production function where A is price-determination as a technology. For Hayek, A_free is > A_planned. It scales up the production function more because it cut costs of knowledge more.

And here is the kicker that people dont get. Any new technology like coding, AI etc. scale up ALL price-determination systems by the same scalar.

If X=coding power today, then
Q_free=(A_free*X)KL
Q_planned=(A_planned*X)KL
By simple definition Q_free>Q_planned still greater by a factor of (A_free - A_planned).

This means that coding power may make central planning cheaper *NOW* relative to *YESTERDAY*. But it is *NEVER* cheaper than freely determined prices.

That is the summary of Hayek. Period.

not only costlier *now* (sorry, missing word)

@PeterBoettke @knowledgeprob @ccoyne1 @Roger_Koppl @PerBylund @petergklein @adamgmartin @Snowtini @jmhorp @Brooke_Medina_ @CafeHayek @tylercowen @ATabarrok @bryan_caplan