Profits as evidence of bank competition or collusion?
In New Zealand there are often complaints that bank profits are exorbitant – especially given that profitability rose during a massive financial crisis. While there is an argument that banks are backed by a lender of last resort, and that they should pay for this (such as through a tax) this doesn’t tell us too much about profitability … remember a tax on bank transactions, or margins, will in part be passed on to borrowers and lenders through an increase in gross margins! Furthermore, the increase in profitability can be seen as banks taking into account the fact that lending was now more risky (after all, the LOLR only functions when the bank collapses – making a bad loan and losing some money is still a cost that is fully faced by trading banks!).
Still, I’m not actually going to discuss any of this. I’m hear to say that the increase in banks margins during the crisis and the corresponding drop as financial conditions have improved can be analysed in the same we economists analyse any sector. And it may well point to a special type of “tacit collusion” in the banking sector.
We have talked about tacit collusion on the blog before, it is a fascinating issue – note that it is’t explicit collusion, it is just an illustrate of how the individual choices of banks regarding their strategy is setting price may in some situation mimic collusive outcomes. The key example I’ve provided was the price of Nurofen but there are many examples. If we have the type of competition modeled by Rotemberg and Saloner, then firms will tacit collude during periods of “low demand” and this collusion will break down during periods of high demand – when the size of the “pie up for grabs” is larger. Given that banks compete on price, have a very clear idea of when a shift in demand is for their firm or for the market as a whole, and don’t face significant capacity constraints, this type of argument is actually pretty relevant!
So what does this tell us about bank lending behaviour? Well in a situation where we do have multiple banks, lending behaviour during booms may act as if the industry is competitive, but during slow downs it will act as if it is more of a monopoly – as a result, the quantity of credit provided will fall more sharply during a slowdown and the price of credit will be higher compared to the case when the industry is just magically always competitive. Interesting stuff!
Two comments… firstly the ROE reported for NZ banks in the most recent PWC report was about 14% – hardly exciting stuff in terms of equity returns and hardly evidence of excessive profits..
Secondly, banks credit policies tend to move together – so they tend to tighten at the same time in a downturn hence the charge of collusion… but the tend to relax their credit policies in an upturn at different times – so the appearance of competition…
And to boot – banks are heavily regulated – so the regulator may drive credit policies in a downturn to some extent – which is a sort of insurance policy against ensuring that banks remain solvent, perhaps….
Indeed – I agree all those points are relevant, and I would normally think they are the full explanation. However, this form of tacit collusion is widely used as a model across the economy – it is interesting to apply it to bank competition.
It would be encouraging to actually have an analysis of the industry that tests these different hypotheses around pricing by banks, I suspect you could use the timing of decisions and well as the levels in order to help figure out what is going on.
It depends on which part of a bank’s operations you are talking about… in the wholesale operations (forex for example) I expect that competition is pretty vigorous all the time.. in large scale lending banks may be joined together in ‘panels’ or syndicates – not collusion per se, but ‘co-operation’ for specific very large borrowers.
At the retail end there is a lot of apparent competition for mortgages (check the value of mortgage approvals compared to value house sales) but in deposit and chequing accounts the competition is likley to be practically zero.
As all competition happens at the margin one measure of it may be the number or % value of customers that switch banks or alter their banking relationships over a given period and see how that links to the credit cycle. I don’t know of any data published on this so I expect it is down to surveys of well defined and discrete customer groups…
Maybe David Tripe knows?
The key thing with “tacit collusion” is that people don’t know about it – they are following decision rules in the “competitive game” that are optimal, but aren’t akin to the competitive outcome.
In no way does there have to be any communication or desire to collude, it is just that the structure of the market allows rules to exist that can help to replicate collusion. In the aforementioned case, collusion actually breaks down as economic conditions brighten up – which can help to explain the procyclicality of bank competition (although other factors such as cyclical risk pricing will look the same way, and are in some ways more compelling to me).
So if that is the case is ‘tacit collusion’ a bad thing? There are certainly deadweight losses in banking, but that’s generally because of regulation…. it would only be a problem if the ‘collusion’ created a deadweight loss bigger than the benefits…
Indeed, I’m not trying to say anything is bad or good – or what the impact of regulation is. I’m just describing that this is relevant then:
“So what does this tell us about bank lending behaviour? Well in a
situation where we do have multiple banks, lending behaviour during booms may act as if the industry is competitive, but during slow downs it will act as if it is more of a monopoly – as a result, the quantity of credit provided will fall more sharply during a slowdown and the price of credit will be higher compared to the case when the industry is just magically always competitive.”
You would expect lending to be even more cyclical on the basis of this. Whether there is a role for policy is a whole different question 🙂