Network News

X My Profile
View More Activity

Costs and consequences

In late March, Andrew Haldane, the Bank of England's executive director for financial stability, delivered an interesting talk on the right ways to think about the risks posed by the modern banking system, and the options on offer for dealing with them. To excerpts I particularly liked:

The narrowest fiscal interpretation of the cost of crisis would be given by the wealth transfer from the government to the banks as a result of the bailout. Plainly, there is a large degree of uncertainty about the eventual loss governments may face. But in the U.S., this is currently estimated to be around $100 billion, or less than 1% of U.S. GDP. For U.S. taxpayers, these losses are (almost exactly) a $100 billion question. In the U.K., the direct cost may be less than £20 billion, or little more than 1% of GDP. Assuming a systemic crisis occurs every 20 years, recouping these costs from banks would not place an unbearable strain on their finances. The tax charge on U.S. banks would be less than $5 billion per year, on UK banks less than £1 billion per year. [...]

But these direct fiscal costs are almost certainly an underestimate of the damage to the wider economy which has resulted from the crisis – the true social costs of crisis. World output in 2009 is expected to have been around 6.5% lower than its counterfactual path in the absence of crisis. ... Moreover, some of these GDP losses are expected to persist. Evidence from past crises suggests that crisis-induced output losses are permanent, or at least persistent, in their impact on the level of output if not its growth rate. ... Put in money terms, that is an output loss equivalent to between $60 trillion and $200 trillion for the world economy and between £1.8 trillion and £7.4 trillion for the U.K. As Nobel-prize winning physicist Richard Feynman observed, to call these numbers “astronomical” would be to do astronomy a disservice: there are only hundreds of billions of stars in the galaxy.


Glass-Steagall was simple in its objectives and execution. The Act itself was only pages long. Its aims were shaped by an extreme tail event (the Great Depression) and were explicitly minimax (to avoid a repetition). It sought to achieve this by acting directly on the structure of the financial system, quarantining commercial bank and brokering activities through red-line regulation. In other words, Glass-Steagall satisfied all three robustness criteria. And so it proved, lasting well over half a century without a significant systemic event in the U.S.

The contrast with Basel II is striking. This was anything but simple, comprising many thousands of pages and taking 15 years to deliver. It was calibrated largely to data drawn from the Great Moderation, a period characterised by an absence of tail events ... Basel II was underpinned by a complex menu of capital risk weights. This was fine-line, not redline, regulation. In short, Basel II satisfied few of the robustness criteria. And so it proved, overwhelmed by the recent crisis scarcely after it had been introduced.

You can read the whole speech here.

By Ezra Klein  |  May 17, 2010; 11:23 AM ET
Categories:  Financial Regulation  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: The conflict between Zionism and liberalism
Next: Think Tank: The $100 billion question; the future of food; and can Google save the news?


I'm wondering why crisis-induced recessions more stubborn than other recessions. Is it partly because the lead-up was a bubble, which artificially inflated the economy, so the recovery only rises to pre-bubble levels.

Regarding Glass-Steagall, its simplicity may be its virtue. Any complex system is going to have failures, and they are going to be unpredictable.

Posted by: jduptonma | May 17, 2010 11:52 AM | Report abuse

If I'm not mistaken (and I very well could be) Basel II uses some sort of Value at Risk measure for their capital requirements. I don't know how standardized it is, but the results of such measures value dramatically based on what risk model you use and the distributional assumptions you make regarding heterskedasticity, skewness, and kurtosis. In other words, the models of extreme tail events depend heavily on the assumptions one makes about the shape of the tails.

Posted by: nylund | May 17, 2010 12:10 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company