This article is brief as I was somewhat side tracked last week by a mild case of COVID and then Canadian Thanksgiving. I am looking at commenting on another article on banking, but it is not ready yet.
The problem with the post-Keynesian/neoclassical arguments that I saw was that people were going on about “intermediation” — with both sides claiming victory. The best approach to that argument is to ignore both sides. As I pointed out in my “banking debate” series, bank loans consists of two fundamental components: funding, and credit risk. Funding flows are circular within the macroeconomy, and so “intermediation” is not happening.1 However, credit risk is not circular, so banks act as credit risk intermediaries between depositors and entities that borrow from the bank.
The announcement by the Nobel* committee was somewhat off-putting, as it claimed that the winners (Diamond, Dybvig, and Bernanke) had advanced our understanding of financial crises. (I assume the contributions refer to the Diamond-Dybvig banking run model, and Bernanke’s study of the Great Depression.) My concern with this is that these contributions pre-date the 2008 Financial Crisis. If we knew all this before hand, why exactly was the mainstream utterly flat-footed by the crisis — including Bernanke (“Great Moderation” and “subprime is contained”)?
The explanation is that neoclassical macro is not an internally consistent set of “physical laws” like Newtonian mechanics and electromagnetic laws within the pre-quantum/relativistic physics syllabus, rather it is a set of anecdotes in the form of mathematical models that have to take an accepted form. Choose the wrong model for a set of circumstances, whoops, your predictions are totally wrong.
Since I fell asleep trying to read Bernanke’s pseudo-monetarist discussion of the Great Depression, I will just comment on Diamond-Dybvig. Their contribution is that they developed a model for what are allegedly bank runs within a framework that is utterly unsuitable for modelling credit relationships.
To explain my previous assertion, we need to look at the theoretical core of neoclassical models. The model has agents that start with an “endowment” that is a vector of “commodities” (including labour hours as a commodity), and then it bargains with other agents so that everybody ends up optimising some utility function. Within this structure, money is a lump of something, almost indistinguishable from a lump of gold.
Credit does not fit within the model, since it creates something out of thin air — the debt owed between agents would be a new “commodity” that is created by a transaction. This does not fit the model of finding exchange ratios (relative prices) between commodities with agents with fixed endowments. So, we need to do something else to come up with something that looks like a bank run.
Although the Diamond-Dybvig model might be seen as useful by neoclassicals, we still need to deal with credit in order to grasp the investment accelerator mechanism that Minsky emphasised. Borrowing to invest generates profits in the business sector in aggregate — which provides a justification to increase investment.
A pretty standard argument is that banks need to provide against the liquidity outflow that is the expected risk of a loan being made. Needless to say, the people making this argument will be happy to chide others for “partial equilibrium thinking.” If we take that advice, we see that a deposit outflow from a bank — other than in the edge case of withdrawal of currency — implies an unexpected inflow to another bank. This means that there is balanced source and sink for funds created in the funding markets that allows for an offset for the transfer. The only thing stopping this from spiralling off to infinity is the need for liquidity and capital buffers — and the capital buffers are there to deal with the credit risk component of the loan transaction.