Banks, tail risk & agency problems

In a very good interview with the Telegraph, Mervyn King complains that we haven’t yet properly sorted out the banks:

“We’ve not yet solved the 'too big to fail’ or, as I prefer to call it, the 'too important to fail’ problem…
Every supervisor should say: 'The banks I should worry about are not only the ones that are losing money but the ones who are making a lot of money.’

For me, one issue here is that tail risk, asymmetric information and agency problems interact nastily.
Put it this way. Imagine the safe interest rate is 1%. A banker than devises a strategy which as a 98% chance of paying 3% and a 2% chance of losing 100%, and these chances are random. This strategy then has an expected payoff  in any one year of 0.94%. Logically, this is a bad investment. A rational non-risk-tolerant person wouldn’t accept it if in involved his own money.
However, over a 10 year period there’s a four-fifths chance of this gamble paying off every year. And there’s a two-thirds chance of it paying off every year for 20 years.
So, who has the incentive or ability to stop banks pursuing such a strategy?
Not bankers themselves. Given the high pay in the industry, you only need a few years in the business to make a fortune. The best case scenario - which is highly likely - sees you earning millions. The worst case scenario sees you leave the business and take up some reasonably paid job elsewhere. The balance of risks, then, pushes you to take the gamble.
Nor will shareholders  stop this strategy. If you short sell the bank, there’s a huge chance of you losing money because the risk of collapse won’t materialize. And most fund managers’ interest lies only in out-performing their peers, which they can do by slightly underweighting banks. So there’s not much selling pressure on the stock.
This means that deferred bonuses paid in stock will not necessarily rein in the bankers. Given the small risk of disaster, share prices could stay high for years, making such bonuses valuable.
Nor do regulators have the ability or incentive to solve our problem. One problem is that they face little penalty for failure: so what if a bank blows up on their watch? They don’t lose much. Another problem is that my random strategy is observationally equivalent to a skilful one. When the regulator asks: “isn’t this strategy too risky?” the bankers will show him sophisticated risk management methods to suggest that they can avoid that 2% chance. They can also, of course, do this to shareholders, which further limits the extent to which the stock market will rein in risky behaviour.
So, is the solution nationalization? Not in itself. Over the five-year term of office of a government, there’s a 90% chance of my strategy working. A government focusing only on the next election might not therefore attach a high priority to reforming the banks, and would prefer instead to simply privatise them as soon as possible to finance tax cuts.
So, what’s the answer? One possibility would be for the concept of fiduciary duty to be taken more seriously. If bankers, fund managers, governments or regulators approached the banks as if their own wealth were at stake, these problems would not arise, or at least would be greatly mitigated.
Another possibility would be for the banks not only to be nationalized, but to then be put under open, democratic control.
The chances of either happening are, however, slim.

Comments

[Keith]

Why would "democratic control" solve this problem or any merely structural reform? The problem is insoluble as all financial commitments involve risk of loss in a market economy under conditions of uncertainty. Understanding our ignorance in theory does not remove our ignorance in practice.
Voting cannot abolish the Law of gravity. Nor the laws of probability. More democracy might produce more conservative choices about investment but that means lower real returns and risk of loss never goes away. Uncertainty cannot be abolished. All approaches are moot.
And if every person owns the banks and makes policy why would the demos not prefer more risk and more returns as well? like gamblers in a casino?

[chris]

I think the case for democratic control is not so much that it eliminates the risk of fianncial crises - maybe you're right that nothing can do this. Instead, it's that it allows the people to decide how to make the trade-off between easy credit and the risk of a crisis.

[Jim]

If you gave most people the figures at the top of your post and asked what their views were, you'd get some very blank looks. 'Whatever gives me the most money now, b*gger 20 years time' is probably the answer you'd get. People taking the easy credit on offer got us into this crisis - what makes you think if the banking system was in ordinary people's hands there would be any less easy credit or people willing to take it on?

[CharlieMcMenamin]

I'm not utterly convinced that the principal-agent prism is the only looking glass available to examine this problem - it could equally as plausibly be recast as a Burnhamite problem of executive control of resources and organisations trumping (formal) ownership. It's only a hop, step and a jump from that view to seeing the kind of power exercised in the resource extraction practised routinely by the higher executives and market traders in the financial section as a new kind of de facto ownership.

Whether you call this power 'ownership', or just control or simply a hugely successful form of agency rent seeking I don't think it can be easily broken. Certainly public ownership in and of itself doesn't seem to have made very much difference to the behaviour of various British financial institutions. & the idea of 'workers(sic) control' in the financial sector makes me shudder with fear, though I'm generally a supporter of the idea in other fields.

None of this is to disagree with your suggestion that public ownership and democratic control of financial organisations may be necessary parts of any sensible solution. But they are not sufficient on their own - and in the case of 'democratic control' we don't have even a vague idea of what a workable model might look like, unless one goes back to some variant of 'Pension Fund Socialism' as advocated by folk like Robin Blackburn and as once dreamt of in mid-1970s Sweden.

[Metatone]

Another option to reduce systemic risk is to:

a) Separate investment and retail banking.
Why? Because while it only reduces the political pressure for bailouts of investment banks by a little bit, every little helps.

Also, a public option (TSB?) for savings banking and small business loans would give a staid, bureaucratic, inflexible but safe harbour for those who value such things.

b) Break up the banks - a greater number of smaller banks should make it easier to allow bad decisions to be punished by failure.

[twitter.com/gappy3000]

OK, this point has been made ad nauseam, and it is essentially a folk example, even though Andrew Lo made it in writing first in his "Capital Decimation Partners" (CDP) example (Financial Analyst Journal 2001). But not too much should be read into it. Empirically, it explains little. An ideal candidate for Dillow's behaviour would be a hedge fund. It has weak disclosure requirement, and an even greater incentive to take risk, make profits quickly, and if everything fails, default and start anew. However, hedge funds have experienced far lower active losses than banks and institutional investors during the crisis (15% industry average, vs. a comparable SPX loss of 46%), no bailouts, and relatively few defaults. If the CDP story doesn't explains their behaviour, I don't know how it could explain that of a large bank. Moreover, Most of the anecdotes I am aware of show bank executives and decision makers heavily invested until the bitter end in their employers, thus losing a fortune. Setting aside jail time, from their point of view that's equivalent to financial reparations paid by executives. So, incentives were aligned. Strike fiduciary duty down, please. And nationalization of banks? I don't think they have a good track record, but I don't know enough. I'd love to hear historical examples, rather than abstract proposals.

In conclusion, I think that the CDP has appeal: simple yet clever, it apparently explains a "stylized" fact. But the causes and remedies of "tail risk" (aaarghh) must be found elsewhere.

[webcontrarian]

A major stumbling block is that politicians repeatedly insist that reform of companies is the responsibility of shareholders. This view is bolstered up by the incorrect claim that shareholders "own" companies.

In fact the root of all the conflict of interest issues is the legislation that allows companies to exist and determines the various rights and responsibilities. Certainly it makes good sense to question whether limited liability should continue to take its present form, or whether it should be modified so that an insolvent bank would bankrupt its directors.

Giving the directors a strong interest in keeping the bank solvent would, assuming an absence of government guarantees, result in a need for more equity capital. Attracting more equity capital would reduce profits and do away with many of the perceived problems.

Other changes may be required to achieve a generally more equitable balance between directors of large companies and other members of society.

Chances of significant change? Low!

[fresno dan]

Nice insightful post.
good comments too, I especially liked "Understanding our ignorance in theory does not remove our ignorance in practice."

[Lil'D]

Re the CDP / hedge fund example:

Most HF managers have substantial personal wealth invested alongside their LPs, thus align with the downside as well as the upside. There is little (though some) incentive for a manager to short vol