Saturday, 18 December 2010

The psychology of bank bonuses

The FSA is expected over the next few days to publish published yesterday its new rules on bank bonuses, broadly in line with the guidelines announced by CEBS, the pan-European committee of regulators. It's likely that, fFrom 1 January, banks will only be allowed to pay a third 40% of bonuses in cash, with the rest paid as deferred claims of one kind or another - debt, preference shares or equity - which can be drawn down over three to five years.

This is meant to reduce the incentive for bankers to take risks: with a high proportion of their wealth tied up in the company they work for, they will want to ensure its survival. However, there are two questions it leaves open, as I just managed to squeeze in on Radio 5 this afternoon before they decided the 6 o'clock news was more important.

The first question is: do banks - and bankers - actually know whether their actions are risky? It certainly didn't seem that way in 2008. Investments in property that they thought were incredibly safe turned out to be a disaster. The problem leading to the current crisis was not the mismatch of incentives or the principal-agent problem which the new regulations are meant to solve - but a lack of awareness of risk. And more importantly, a lack of awareness of the correlation of risk across the whole banking sector.

Admittedly, people who work within a bank are probably better placed than most others to see the risks to that bank's capital. And by making company-wide risks more salient and more important, the regulations will help by encouraging people to spot the problems that might happen. But two specific interventions might make it more effective.

First, a transparency principle. Within a bank (if not outside it) the bank's detailed positions on all assets and debts should be publicised. Interested insiders would have a strong incentive to spot risks and flag them up, even if the people taking the risks missed them - wilfully or otherwise.

Second, the ability to identify correlations between different institutions. A market solution is probably the best way to do this: sell correlation-related instruments such as swaps between the assets of different banks, or combined put/call options with a spread. Banks whose assets are tightly correlated would have a low price for such swaps; two uncorrelated banks would have a high price (positive or negative) for the swaps, and a non-zero price for the option. Monitoring the market prices of these instruments would provide information to the regulator on potential systemic risks. And allowing, even encouraging, insider trading in these specific instruments, would lead to information being released quickly. Critically, banks whose assets are more highly correlated (to each other, or to the market average) should hold higher capital reserves. This will encourage banks to find new kinds of assets which are more systemic-failure-proof. As a beneficial side-effect, it will create more diversification of investment - good for the economy in general.

The second question is this: can rational incentives explain bankers' behaviour anyway? There is plenty of psychological research into decision-making under uncertainty, attitudes to risk and so on - much of which shows that people do not respond in the expected way to incentives. Instead, framing of risks (e.g. as gain versus loss), the combination of different risks into a single decision (e.g. the Allais paradox) and apportionment of risks into corner cases (very high or very low probabilities) are all known to change people's perceptions and their actions under conditions of risk.

There's a general assumption that experienced financial traders are likely to be less biased than average investors, and that's probably true. But it's very unlikely that traders can be completely rational and objective, especially when dealing with novel situations or making very quick decisions. There's a well-known piece of research showing that traders take more risks on days when they have higher testosterone levels. That evidence isn't conclusive, but it is a heroic assumption to imagine that a trader can eliminate all biases other humans are subject to. These effects need to be understood in order to know what kind of regulation is appropriate.

Despite all that, the FSA's move is a good one - and follows an interesting principle, that of trying to put some parts of regulation back into the hands of people who work in the institutions. Regulating from outside - the US approach - is always likely to leave loopholes. Aligning the interests of bank employees, investors and society has a real chance of working.


PunditusMaximus said...

This is all ridiculous -- the really egregious stuff was all built on enormous frauds. Why not just prosecute people for lying about money? That's the kind of rule people should be able to follow.

Anonymous said...

Why would traders be less biased than investors? They have much less skin in the game.

Also, I don't think the underestimation of risk that caused the banking crisis was due to traders making bad decisions quickly. CDOs and RMBSs were developed and marketed over a longer time period, with plenty of opportunity to consider and price in systemic risks.