Andrew Haldane on the arms races in banking: keeping up with the Goldmans
Andrew Haldane, who is widely regarded as one of the most perceptive analysts of the global financial crisis, has given a talk at the Institute for New Economic Thinking (INET.)
He talks of harmful "arms races" in the financial sectors, of which he identifies three: one from the past, one from the present, and one for the possible future. He said these arms races (I prefer to call them races to the bottom) are prevalent throughout human behaviour then adds:
"Finance is plainly no exception. In fact the very structure of finance means that these arms race type behaviours is even more prevalent than in other aspects of human behaviour."
The arm's race from the past relates to banks' Return on Equity (ROE). This was:
"an arms race, not just a case of keeping up with the Joneses, but of keeping up with the Goldmans. If Goldman posted a ROE of 20 percent, everyone else felt they needed to leapfrog ahead. The simple way that leapfrogging was achieved was by taking on leverage. The result of this arms race on returns on equity was that everyone converged to a high ROE and therefore high leverage and therefore a high risk equilibrium, which sowed the seeds of crisis of the last 3-4 years.
. . .
(drawing on) Philippon and Reshev, important work on returns in finance and non-finance: for investment and universal banking this was a period that isn’t even close to any historical precedent, not even the 1920s."
Strong stuff, and there is a similar dynamic going on with top bankers' pay - with the difference that whereas ROE fell off a cliff in the crisis; bankers' pay hasn't. Haldane and others have analysed this dynamic before (see this newspaper article and longer paper, for example), but it helps make his overall case here.
The next arms race, from the present, concerns high frequency trading. And here he produced some astonishing statistics:
"Those traders now dominate mainstream financial markets, accounting for between a half and three quarters of volumes transacted for example in the world’s major equity markets. One reason they dominate is that they submit huge volumes of quotes in the market, most of which are never exercised: the firms cancel them before they are ever exercised. For every order executed, 60 are cancelled."
My emphasis added. This is about stuffing the bandwidth full so that others can't trade, and aggressively taking advantage of those fleeting moments of superiority. The Flash Crash of May 2010 is one example, but there have been hundreds of mini-crashes since then, he said. Here is another case of an arm's race producing a public bad.
The third race is less intuitive, and concerns a flight to safety. In essence, investors in banks all want to have their claims secured against solid collateral, because of their fears about banks' eventual solvency. But there is only so much collateral to go around.
"It, too, is an arms race: it too comes with a cost. It results in banks’ balance sheets being progressively more encumbered. They are signing away those assets to an increasing number of investors: those assets cannot be signed away indefinitely."
I think Haldane's arguments need to be complemented with further discussions of arms' races. Financial actors deliberately encourage arms' races between jurisdictions. "Don't tax or regulate us too much or we'll fly off to London / Geneva / Singapore / Hong Kong / New York" the bankers cry, and regulators as a result fail to put in place those capital requirements or other regulatory devices to make the system safer. More on this kind of 'competition' here.
So we have the regulators failing to regulate, leading to unhealthy arms races; and the bankers growing ever more powerful, and using an arms' race between jurisdictions to tie the regulators' hands, making them yet more powerful, and . . . so on. A circular race to the bottom. Where will it ever stop? We now have regulatory fatigue, and a public ground down by the sheer weight of awfulness spewing endlessly out of the financial sector.
This is the trap the world has fallen into. We are in a downwards spiral. Who is going to break this dynamic?
The text of Haldane's speech is below. It's almost complete; there are a few minor words missing here or there, and a couple of slightly spurious sections (I've put timing marks in, so you can go and find and listen to them if you want.)
Keeping up with the Goldmans (my title, not his.)
I want to have a different take on the question of the social utilty of finance. In particular I want to focus on the possibility of equilibria in financial markets that are true Nash equilibria, but are nonetheless socially inefficient in some sense, in particular where that inefficiency derives from an arms race. We know that the classic arms race does deliver socially inefficient equilibria: the desire for individual safety results in an equilibrium where everyone is collectively less safe.
I want to talk about some examples where similar sorts of dynamics are at play in the financial sector, where competition can be a public bad.
. . . some stuff about Elephant Seals . . . Gareth Harden’s classic paper on the Tragedy of the Commons . . 3:30
We have examples from throughout econonmic theory, of relative behaviours of relative standing, be it for setting of wages or prices or consumption or employment, keeping-up-with-the-Joneses type arms races do appear to proliferate across most aspects of human behaviour. Finance is plainly no exception. In fact the very structure of finance means that these arms race type behaviours is even more prevalent than in other aspects of human behaviour.
One structural feature that delivers that is information assymmetry. Finance is founded on information assymmetry, 4:30, on people not knowing. And one response to people not knowing, acting in a world of deep uncertainty about how much risk is truly being borne; who are the good guys and bad guys in finance; one response to that uncertainty and asymmetry is to rank: is to order firms, to convey financial information in league tables. If you convey information in league tables you generate just the types of arms race dynamic that we saw from the elephant seal. That possibly will generate these socially suboptimal outcomes, the so called Red Queen races, where however fast you travel, no-one ever wins. 5:30
Let me give some examples of these arms race dynamics appearing in the financial sphere. I will talk about three: one from the past – the race in returns in the banking system that we saw pre-crisis; one from present – the speed race currently being carried out in financial markets, and the one from the potential future: the quest for safety.
First, the race in returns was a key driver, generator, propagator of the crisis. With hindsight, returns to both financial capital and financial labour were extraordinary by any historical metric. The LHS picture here looks at the return on equity for UK banks over 100 years. Returns on equity to UK and other banks were at historcally unprecedented levels. The 1920s is the closest we might remotely come. 7.00 The right hand panel blows up returns on equity for all global banks during the runup to crisis from 1990 onewards. It is a steady ascent. One of the interpretations of that ascent is that it was an arms race, not just a case of keeping up with the Joneses, but of keeping up with the Goldmans. If Goldman posted a ROE of 20 percent, everyone else felt they needed to leapfrog ahead.
The simple way that leapfrogging was achieved was by taking on leverage. The result of this arms race on returns on equity was that everyone converged to a high ROE and therefore high leverage and therefore a high risk equilibrium, which sowed the seeds of crisis of the last 3-4 years. That is why ROE returned to earth, indeed below earth, in 2007-8. 8:30 That pattern in returns to financial capital has been mirrored in returns to financial labour. LHS picture taken from Philippon and Reshev, important work on returns in finance and non-finance: for investment and universal banking this was a period that isn’t even close to any historical precedent, not even the 1920s. The RHS looks at bank CEO compensation. It, too, ascends pretty much in line with the ascent in returns on equity. Some of that is east to explain, because in this period, CEOs, senior executives in the banks were being remunerated on the basis of equity-based metrics. The only difference between CEO pay and ROE: while the ROE fell off a cliff, that is not obvious with CEO pay.
10:00 Second: the speed race has been playing out, is still playing out, over last 10 years or so. Back up 20 years, and trade execution times were measured in minutes. Back up 10 years, and those trade execution times were measured in seconds. Back up five years, and those trade execution times were measured in milliseconds; today it is measured in microseconds. Tomorrow it will be nanoseconds. Iit has given rise to this phenomenon of high frequency trading.
Those traders now dominate mainstream financial markets, accounting for between a half and three quarters of volumes transacted for example in the world’s major equity markets. One reason they dominate is that they submit huge volumes of quotes in the market, most of which are never exercised: the firms cancel them before they are ever exercised. 12: 00 For ever order executed, 60 are cancelled. What is going on here? One thing: although there are loads of quotes on the screen, if you try and hit them, they disappear before you can transact. There is a mirage of liquidity. A second feature is that some have said that this practice of putting in multiple quotes, “quote stuffing”, is an externality because bandwitdth is a common good: it is finite, so by quote stuffing you are using up bandwidth and you are slowing down everyone else. This might matter a lot if prices are moving around a lot.
13:15. Take for example 6th May 2010, the well known flash crash, the mirage of liquidity was shown to be just that. .. bandwidth message traffic went through the roof: those who could not keep up were unable to trade … .the flash crash was no one-off. In the period since, though not well publicised, there have been hundreds of mini flash crashes, the sources of those was this speed race in financial markets, that brings the externality of fragile liquidity and of order cancellation.
Third, the final race is a flip of the first one I mentioned. Over the last year or two, there has been not so much a quest for risk, but a quest for safety. That has been felt acutely by the banking system. Investors increasingly want the security of collateral. Investors are much less willing to invest in them on unsecured terms than in the past. Everyone wants to be senior, to be first in the queue, to have first claim on the assets of the bank. We are engaged in a race to safety.
Not everyone can be first, can be senior. 15:30 Central banks -- in the liquidty against collateral that they have provided -- have been part of this race. For example, look at the LHS picture. That looks at the refinancing requirements of Euro area banks this year, which total around US$1.1 trillion. Most of that that needs refinancing is the blue bar, which is unsecured: money taken out on unsecured terms in the past. So far this year, the unsecured fraction, the blue bit, has been miniscule. It has almost all been secured on collateral: a good chunk of course through the ECB’s LTRO operations. 16:20 The unsecured market has found itself squeezed as investors have sought, through this arms race, security, safety. This is the pattern to the RHS picture that we have seen playing out not just this year, but for every one of the last three or four years. It, too, is an arms race: it too comes with a cost. It results in banks’ balance sheets being progressively more encumbered. They are signing away those assets to an increasing number of investors: those assets cannot be signed away indefinitely. This is BarCap’s estimates of how much of banks’ balance sheets in different countries are encumbered. Those numbers are almost certainly an understatement.
17:26 This equilibrium is self-fulfilling, because if I am an unsecured creditor, why would I refinance on unsecured terms right now: why should I let everyone else be ahead of me in the pecking order? What we are seeing is everyone converging on this high-security, high-safety but ultimately self-defeating equilibrium.
What do do about it?
If you buy this story, these arms races which proliferate across finance, then there is a clear case for public intervention, 18:05 both to protect the system and to protect individuals within it. Here is a rare case where both the regulators’ and the regulateds’ incentives are in the same place. For that intervention to be effective, it needs not to focus on any one business, but to coordinate across the system as a whole. I needs to be explicitly macroprudential, otherwise you intensify the competitive scramble.
What interventions would help?
On the returns’ race, the answer is simple. With hindsight, we would have capped leverage and prevented forestalled this ROE race. We might even have capped both. We’d have sought to put performance measures that made this race in returns less likely. On the speed front, we might stop it at source, as some U.S. regulators are now doing, this act of quote-stuffing, by placing some restrictions on order cancellation; to require traders to commit to provide liquidity; we might require circuit-breakers that sit across the system as a whole. On the last race, the safety race, this is the one where regulators might have most to do: right now there are no restrictions on how much of their balance sheets banks are allowed to encumber. In future we might need that.