Friday, September 25, 2009

Predicting banking crisis

When the asset bubbles were building up, the common refrain was that it would be unwise to deflate bubbles because we are not sure whether it is a bubble and the costs of such actions can be substantial. It was argued that departures of the market values (itself a tricky valuation exercise) of assets from its fundamentals was difficult to predict and therefore it was preferable to clean up after the bubble burst.

Bharat Trehan of the San Francisco Fed claims that "simple indicators based on asset market developments can provide early warnings about potentially dangerous financial imbalances", and suggests that departures of credit levels and asset prices from their historic threshold levels are good predictors of an imminent banking crisis.

Trehan points to multi-country studies by Claudio Borio and Philip Lowe and Graciela Kaminsky and Carmen Reinhart (and also this) which examined three different measures - asset prices, credit, and investment - and found that increase in credit levels above a certain threshold (they called it "credit gap") is the best predictor of a banking crisis. They defined the credit gap as the difference between the current ratio of credit to GDP and a slowly changing measure of the trend value of this ratio. They found that with a credit gap of 5% as threshold, we would be able to predict 74% of the crises that occurred subsequently.

Asset (stock and property markets) prices are the other indicator that can best approximate the imminence of a banking crisis. After examining a sample of 15 countries, Bordo and Jeanne declare an asset market to be in a boom or bust if the three-year moving average of the growth rate of the inflation-adjusted asset price falls outside a specified range, whose width is defined taking into account the historical average growth rate and volatility of the asset price. This analysis is more accurate for property than for equity markets.

As Trehan writes, the purpose is not to zero in on the ideal set of indicators, but to identify simple indicators that "would have signaled impending trouble prior to the current crisis"... simple indicators can be useful, not to fine-tune policy during normal times, but as signals of rising levels of risk in the economy."

No comments: