2001年-世界发展银行全球_Tropical_Bubbles___Asset_Prices_in_Latin_America_1980-2001_64页_2mb

Summary of "Tropical Bubbles: Asset Prices in Latin America, 1980-2001"

Core Content

This working paper by Santiago Herrera and Guillermo Perry investigates the presence of asset price bubbles in Latin America from 1980 to 2001, with a particular focus on stock prices. The authors challenge the notion that asset prices in the region are primarily determined by local factors, suggesting instead that both country-specific and common external variables play a significant role in determining asset price bubbles.

Main Findings

• Bubbles Exist: The paper concludes that asset prices in Latin America, particularly stock prices, cannot be rejected as being influenced by bubbles. This finding contrasts with previous studies that emphasized the dominance of local factors.
• Bubble Testing: The authors use unit root and cointegration tests to examine the presence of bubbles. They find that observed stock prices and dividend yields are generally non-stationary, indicating potential bubble behavior.
• Bubbles Are Persistent: Bubbles are identified as periods where asset prices significantly deviate from fundamentals-determined levels. These deviations are found to be persistent and frequent, with notable exceptions in Argentina after 1992.
• Common External Factors: The U.S. term spread and U.S. asset overvaluation are identified as the most important common external factors influencing asset price bubbles in Latin America.
• Country-Specific Factors: Domestic credit growth, its volatility, asset return volatility, and terms of trade are highlighted as key country-specific determinants of bubbles.
• Impact on Economic Activity: Asset price bubbles have significant and long-lasting effects on the financial sector and real economic activity. The balance sheet effect and its impact on bank lending are identified as the primary channels. Additionally, stock prices are positively associated with investment and negatively correlated with stock price volatility.
• Currency Crises: The authors note that asset price crashes often coincide with currency crises, suggesting that monitoring asset prices is crucial for central banks to anticipate economic instability.

Key Variables and Models

• Bubble Model: The intrinsic bubbles model by Froot and Obstfeld (1991) is used to separate the present value component from the bubble component of asset prices. The model is defined as:

  $$
  P_{t} = P_{t}^{PV} + B_{t}
  $$

  where $P_{t}^{PV}$ is the present value of expected future dividends, and $B_{t}$ is the bubble term expected to grow at the real interest rate.

• Unit Root Tests: Using the Augmented Dickey-Fuller (ADF) test, the authors find that most Latin American countries have non-stationary stock returns and dividend yields, which suggests the presence of bubbles. However, the results are not robust to the inclusion of a deterministic time trend in some cases.

• Cointegration Tests: The authors apply three cointegration tests (Johansen, RALS, and ARDL) to examine the long-run relationship between stock prices and fundamentals. The results generally reject the no-bubble hypothesis, with Argentina being the exception.

• Logit Model: A logit model is estimated to assess the relative importance of different factors in predicting bubble occurrences. The results show that both common and country-specific variables have similar marginal probabilities in determining bubble behavior.

Policy Implications

• Central Bank Monitoring: Central banks should monitor asset prices to anticipate potential economic crises, as asset price crashes often coincide with currency crises.
• Financial Sector Impact: Bubbles can have long-lasting effects on the financial sector, particularly through the balance sheet effect and changes in bank lending behavior.
• Investment Behavior: Stock prices are positively associated with investment, while stock price volatility has a negative impact on investment levels.
• Global Linkages: The paper highlights the importance of global factors, such as U.S. asset overvaluation and term spreads, in shaping asset price dynamics in Latin America.

Conclusion

The study provides evidence that asset price bubbles are not unique to Latin America but are influenced by a combination of local and global factors. It underscores the need for policymakers to consider both domestic and international variables when assessing asset price behavior and its implications for economic stability and growth.