Do Lending Booms Lead to Financial Crises?

09/01/2001
Summary of working paper 8249
Featured in print Digest

Lending booms do make Latin American economies considerably more volatile and vulnerable to banking and balance-of-payments crisis. The probability that a banking crisis and a balance of payments/currency crisis will follow a lending boom is twice as high in Latin America than in the rest of the world.

Lending booms have been used to explain many banking crises, including Chile's in 1982, Mexico's in 1994, and Thailand's in 1997. In each case, reliance on foreign capital led to financial disturbances that combined banking crises with a balance-of-payments collapse. The experience of lending booms had led some academics and practitioners to advocate the use of controls on short-term capital inflows or on private credit growth. Even the International Monetary Fund has been moved by recent experience to acknowledge the benefits of targeted capital controls.

But are lending booms really that bad? According to Pierre-Olivier Gourinchas, Rodrigo Valdés, and Oscar Landerretche, writing in Lending Booms: Latin America and the World (NBER Working Paper No. 8249), the answer is a qualified no, with Latin America standing out as the exception.

A lending boom is defined as a period when the ratio of private credit to private gross domestic product deviates from its historical trend. During a boom, credit to the private sector increases rapidly. The danger is that as lending increases, the quality of funded projects declines, and the banking sector becomes more vulnerable. However, Gourinchas, Valdés, and Landerretche show that the presumption that lending booms generically lead to banking crises is wrong. While a lending boom may precede most banking crises, banking crises do not follow most lending booms. Financial development typically occurs in stages, with periods of intense financial deepening and increases in the level of financial intermediation by banks. Large increases in lending may represent a permanent capital deepening rather than just a transitory boom.

The authors use two measures of lending booms, a relative measure that compares the size of additional lending to the size of the banking sector and an absolute measure against the size of the economy. Their study is based on data for 91 countries, over the period 1960-96. The number of lending booms identified depends on the size of the threshold used in measuring deviation from the norm. Using a relative deviation of 24 percent or an absolute deviation of 5 percent, there are 60 and 65 cases respectively. Using relatively high thresholds of 42 percent and 8 percent, there are 23 and 33 cases respectively.

Argentina, for example, experienced two lending booms (from 1979-82 and 1992-5) with the credit/GDP ratio increasing by 100 percent in the first episode and by 70 percent in the second episode. Chile experienced a long lending boom between 1975 and 1984, where the ratio increased by 1,200 percent. Mexico experienced a lending boom from 1988 to 1994, with the credit/GDP ratio increasing by 350 percent.

The researchers find that lending booms are associated with: an investment boom and to a lesser extent a consumption boom; declines in trend output growth over the episode of over 1 percent; a large increase in domestic interest rates; a large increase in the current account deficit and a counterpart in the form of capital inflows; a real appreciation of the domestic currency; some worsening of the fiscal situation; a decline in foreign reserves; and a shortening of the maturity of the external debt.

However, the authors find no significant increase in banking and balance-of-payment vulnerability. Nor do they find any evidence that lending booms--which last an average of 6 years on the relative measure and 5 years in the absolute cases -- tend to come to an abrupt halt.

To analyze whether boom episodes are related to financial crises - and particularly whether they signal future banking troubles - the researchers compare the probability of having a banking crisis before and after a boom with the probability of experiencing a crisis during more tranquil periods. They find that the probability of a banking crisis after a lending boom is relatively low. Although the probability of a banking crisis up to two years after a lending boom is somewhat higher than during tranquil periods, the difference is not statistically significant.

Comparing Latin America's experience with that of the rest of the world, however, the researchers find that lending booms do make Latin American economies considerably more volatile and vulnerable to banking and balance-of-payments crisis. Latin America has experienced a sharp increase in lending booms during the 1990s.

The researchers show that capital inflows are more relevant before the lending booms in Latin America than in the rest of the world; this fits with the fact that a number of Latin American countries experienced capital account liberalization during the sample period. Latin American lending booms have been built around financial deregulation, capital account liberalization, large capital inflows, and failed exchange rate-based stabilization policy. The probability that a banking crisis and a balance of payments/currency crisis will follow a lending boom is twice as high in Latin America than in the rest of the world.

-- Andrew Balls