Editorial note: this article is meant to be an introductory section of a chapter on banking in economic theory. I am going to make a bunch of wild assertions that are supposed to be dealt with later in the chapter. I expect that I will have follow up articles filling in details later.
One initial observation about banking debates is that banking is not popular on the two ends of the political spectrum — populists are notoriously not fans of banking. (I am a Canadian prairie populist, but I am also a political realist. I view banking as a core part of industrial capitalism, and we have to manage it as such.)
The “Rothbardian” wing of libertarianism is a major source of discontent with banks on the political right. (The extreme right has a beef with international bankers, for a relatively well-known reason that I do not want to drag into this text. Whether this infects the rest of the right’s complaints about banking is a judgement call on the part of the reader.) Fractional reserve banking is seen as an evil perversion of the beauty of gold-based money. Anything that takes the system away from the primeval simplicity of barter between individual proprietors is unwelcome. Libertarians in practice might be in bed with large corporations, but their tracts are filled with examples of shoemakers trading their wares with fishermen.
The popularity of internet Austrianism in finance means that if anyone is being wrong online about banks, they are most likely regurgitating some variant of Rothbardism.
At the other end of the political spectrum, the Left has figured out that banks are in fact part a core part of the capitalism system. This is aided by the reality that bank economists are the public face of capitalism in the media: cheering on tax cuts and the efficiency of markets, and wringing their hands about money being spent on poor people. This also leads to an adversarial stance towards the banking industry. Unlike the Rothbardians, there is not over-arching theme of theoretical disinformation about banking.
Bankers have generally figured out that they are not hugely popular, which leads to the centrist strategy of trying to avoid talk about banking. At most, nostrums from Economics 101 will be invoked. Although not publicising extreme views has some merit as a strategy, it is less tenable in the modern social media environment.
I am not going to pursue this angle, but we need to recognise that the political landscape if we want to understand the discussions around banking.
Despite the importance of banking to the economy, it is hard to fit within standard economic models. This has the effect of causing great difficulty for many economists to discuss banking — which is pretty obvious to anyone reading economic commentary.
One immediate danger of my criticism of economic theory in this matter is that the reader might assume that I am claiming to know all the answers. To make my position clear: my guess (“conjecture”) is that banking is exceedingly difficult to insert properly in a mathematical model of the economy. I will explain my logic elsewhere. The usual objective of writing a primer is to inform the reader, removing ignorance. My objective is to rearrange ignorance — yes, we know some things, but other things remain hard to explain.
Another immediate objection I see is that we need to distinguish between economists who study banking versus economic theory. The problem with banking are the mathematical economic models — which many (but not all) academic economists rely upon to develop an intuition about a subject. However, specialists in studying banking are likely to be looking at the details of the system, not trying to build a macroeconomic model. As such, they can have a perfectly decent understanding of banking. Unfortunately, specialised studies of banking are complex and probably fairly boring, so they do not get a lot of media air time.
I can now turn to the theoretical problems in banking. The starting point is what I would call “classical” economics: pre-World War II mainstream analysis of banks. The basic principles of this have made their way in “Economics 101” models of the banking system.
The advantage of the “Economics 101” models of banking is that they are simple models, and make powerful predictions. This makes them very attractive to mainstream economists. They just suffer the minor inconvenience of being utterly wrong descriptions of the system. (“Oh well, you can’t have everything,” sigh the defenders of Economics 101.)
Heterodox economists pointed out the flaws of the classical models a long time ago; a lot of the work was systematised in the 1930s. (Heterodox economics is a wide term, but I am here referring to post-Keynesians and the fore-runners to post-Keynesians, and certainly not the Austrians, who might be otherwise considered “heterodox.”) And what has happened over the past century is that mainstream economists keep repeating the same tales from Economics 101, and the heterodox economists keep repeating the same critiques. When the mainstream economists actually respond to the heterodox critics, both sides will repeat statements that have been made repeatedly since the 1930s — and both sides will claim victory in the “debate.”
Although I obviously sympathise with the heterodox critiques of Economics 101 banking models, the theoretical situation can only be described as dysfunctional. One problem is that the mainstream has managed to convince everyone that “you cannot beat a model without a model.” The models typically being pushed by post-Keynesians are simplified, and thus can run afoul of other critiques. Furthermore, the heterodox side of the debate is generally not following what the mainstream side sees as important (and vice versa). As such, both sides argue past the other. There is also a cultural issue — post-Keynesian academics enjoy cluttering their arguments with appeals to ancient papers and books that realistically will only be read by post-Keynesians. Terminology is tortured, being tied to ancient economic debates. I now understand the wisdom of the practice of throwing old notation and debates under the bus within undergraduate applied science textbooks.
Accounting Identities Not Enough
A lot of primers and arguments about banking from heterodox authors will run through bank balance sheets. This is necessary — you need to know what is possible. The problem is that balance sheet operations are not sufficient — we need to know what are the limits on behaviour. The Economics 101 models do offer an answer to behavioural limits — the most important being the limit to bank lending. The answer might be wrong, but it is an answer. Balance sheet operations do not offer that answer. Although one might point to what might limit a bank’s loan book growth over a span of a few days, there is no balance sheet constraint stopping a developed country’s bank loan book growing by 100% over year. Any short-term liquidity/capital limits can be bypassed in various ways. Meanwhile, we do not see bank balance sheet growth anywhere near that rate under usual conditions in low inflation economies. This implies that these “constraints” are not actually behavioural constraints in practice.
Within the confines of what is seen as acceptable economic debate, the only chance for progress is that post-Keynesian authors shift the terrain to the superiority of their models over neoclassical ones. However, given that neoclassicals do their best to ignore the existence of post-Keynesians, the debates never start that way. Instead, the standard format is that a mainstream economist states something out of an Economics 101 textbook, a post-Keynesian responds, and then the two sides debate the neoclassical model. I would argue that a century provides sufficient empirical evidence to see that approach will not get anywhere.
Do We Need Banks In Economic Models?
In the first half decade after the Financial Crisis, readers of popular economic debates will have run into thousands of variations of the statement “Neoclassical models are useless because they do not take into account the financial/banking system.” Neoclassicals attempted to respond by adding a “financial system” epicycle to their general equilibrium models, which was of course pathetic. However, as memories fade of the Financial Crisis, one does not hear that argument as much.
I am somewhat glad that those arguments are less prominent, since I am somewhat sympathetic to the mainstream view. Inserting a “financial sector” into a model might not make it better.
Take as an example the housing market — the major driver of private sector debt growth in the “Anglo” economies since 1990. As interest rates fell, it was possible to have a larger mortgage with the same payment. This made it possible for households to build up the price of houses — which mechanically increased mortgage debt.
To the extent that we can model house prices, interest rates and nominal income did a decent job. There was also a willingness to increase mortgage service burdens as a percentage of income. One could argue that this was the result of the financial sector loosening lending standards, but it also functionally equivalent to households being more willing to take on debt. In any event, there was generally no problem for the financial sector to allow the increased borrowing to occur.
That is, until the financial sector froze up, and borrowers were turned away. (Problems appeared in the United States in some segments of the mortgage market earlier than 2008, but in other countries, the sudden stop was in 2008.) In that case, it is clear that something happened in the financial sector, and it was not just a question of households getting a different optimal solution.
We needed to know that the financial sector was going to blow itself up going into the Financial Crisis. The problem is: can we get a mathematical model to reflect this? During the expansion, the plumbing of how mortgage debt was raised did not matter: it was always available. Once the financial sector freezes up, we are in a completely different economic environment. Can a single model capture both modes of behaviour?
- If the mode of operation just randomly switches, the model can be used to tell stories about the past, but is useless for forecasting. Given my questioning of the ability of mathematical models to forecast accurately, I cannot complain about this too loudly. However, I am in a minority — most mainstream economists would insist that models need to be predictive. (We also need predictive models to falsify them.) Furthermore, saying the lending conditions magically switch is nonsensical — the financial sector blew itself up, and many critics said that they were going to blow themselves up.
- We can have a simple model where the change of behaviour is predicted by some macro variable. This results in “tipping point” models — “if the debt/income ratio rises above some magic ratio, the system will blow up.” The problem with such models is twofold. Firstly, we only know we hit the tipping point in retrospect. Secondly, it ignores the ugly details of the credit system. For example, there is a big difference between debt increases in “prime” mortgages that can be absorbed by (upper) middle classes and “sub-prime” debt, and if we lump those two things together into “debt,” we are not getting useful information.
My feeling is that Minsky’s instincts were right: we have a simple debt accelerator model explaining the expansion, but the expansion drives loopy credit behaviour outside of the attention of regulators. Things work great until they do not. Predicting a breakdown means that you can predict profitable shorting opportunities. The adversarial nature of markets makes directional prediction difficult (“weak form market efficiency”).
I see two major areas for further discussion. The first is a short discussion of Modern Monetary Theory and banks, which will also pick up the dreaded endogenous/exogenous money debate. The second is a longer discussion of the theoretical problems for inserting banks into mathematical models, which will mainly consist of wild assertions on my part.