Centralization and Automation

Tim Fernholz recommends the “compelling Hayekian argument about underwriting efficiency and financial regulation” from Amar Bhidé that Tony Jackson summarized yesterday for the FT, though Fernholz says he doesn’t agree with all the conclusions Bhidé draws.

I don’t think I find the analysis very compelling:

The head of Intel, for instance, can run an empire of semiconductor plants, because the product is centrally determined and raw materials consistent. But can the head of a big bank run an empire of bank branches in the same way?

Plainly not, Prof Bhidé says. Issuing a mortgage or a small business loan involves precisely the kind of on-the-spot knowledge and judgment Hayek described. But the banks have centralised those judgments. Why is that?

Because that is where the growth is. Proper banking requires an army of trained workers. That slows things down. But if you reduce everything to mechanistic models and harness them to modern computing, it goes turbocharged.

I dunno. It used to be that proper shirt-making required on-the-spot knowledge and judgment. But then apparel manufacturers found ways to centralise production and generate a finite number of standardized sizes. Custom made products are still superior in quality, but the lower prices made possible by standardization and mass production “turbocharged” the industry and produced large gains in human welfare. Why shouldn’t lending be the same? Well:

Misallocation of capital, meanwhile, is inevitable. Capital goes to sectors that can be mechanised. Mortgages qualify, so resources are poured into housing. Small business loans are less amenable, so small business lending shrivels.

Again, this strikes me as entirely typical of how the entire economy works. The system of interchangeable parts for mass production didn’t arrive suddenly from heaven and strike every industry. It was worked out in specific contexts that made the production of certain goods much more efficient. That led capital to flow to successfully automized lines of work which, crucially, inspired people to work on successfully automating other lines of endeavor. It’s not as if we just said “well, this works for firearms but not for sewing machines so let’s forget about sewing machines.” Instead decades of work went into figuring out how to mass produce different kinds of things. But it wasn’t “misallocation of capital” for capital to flow initially to fields of endeavor where efficiencies had been discovered — it was natural.


Meanwhile, I don’t think there’s anything mysterious about why a disproportionate quantity of capital flows into housing in this country — we have all these policies aimed at encouraging capital to flow into housing. That’s what Fannie and Freddie and the home mortgage interest tax deduction are for. We can and should have a debate about whether this constitutes a misallocation of resources (I agree that it was and is) but if it is it has to be tackled directly.