Tropical Bubbles : Asset Prices in Latin America , 1980 - 2001 1 Santiago

In this paper we test for the existence of asset price bubbles in Latin America in the 19802001 period, focusing mainly on stock prices. Based on unit root and cointegration tests we cannot reject the hypothesis of bubbles. We arrive at the same conclusion using Froot and Obstfeld’s intrinsic bubbles model. We identify periods of significant stock price overvaluation to examine empirical regularities of these bubble episodes in the region. We quantify the relative importance of different factors that determine the probability of bubble occurrence, focusing on the contrast between the country-specific variables and the common external factors. We included in the country-specific variables (a) both the level and the volatility of domestic credit growth, (b) the volatility of asset returns, (c) capital flows to each country, and (d) the terms of trade. As common external variables, we considered (a) the degree of asset overvaluation in the stock market and the real estate market in the U.S., and (b) the term spread of U.S. Treasury securities. To quantitatively assess the relative importance of each factor, we estimated a Logit model for a panel of five Latin American countries from 1985 to 2001. In general, we found that the marginal probabilities of both common and country-specific variables were roughly of the same order of magnitude. This contrasts with previous studies that found that real asset returns in Latin America were dominated by local factors. Finally we explore the main channels through which asset prices affect real economic activity, with the most important being the balance-sheet effect and its impact on bank lending. We show how the allocation of bank lending across different sectors responded sensitively to real estate prices during the boom years in countries that experienced banking crises. Thus, asset price bubbles have long-lasting consequences in the financial sector, and, through this channel, on growth. Another channel through which asset prices—in particular stock market prices—affect log-run growth is through their effect on investment. We found (a) a strong positive association between stock prices and investment, and (b) a negative effect of stock price volatility on investment. An additional motive for the central bank to monitor asset prices is the general coincidence of the crash episodes identified in this paper with currency crises experienced in the region in the last two decades.