2006年-世界发展银行全球_International_Financial_Integration_through_the_Law_of_One_Price_43页_745kb

Summary of "International Financial Integration through the Law of One Price"

Core Content

This paper explores international financial integration through the lens of the Law of One Price (LOOP), using the cross-market premium as a key measure. The cross-market premium is defined as the percentage difference between the price of a stock in its domestic market and the price of its corresponding depositary receipt (DR) in the U.S. market. The study applies Threshold Autoregressive (TAR) models to analyze the behavior of this premium, highlighting the importance of non-linearities in price convergence and the role of transaction costs in segmenting financial markets.

Main Points

1. Cross-Market Premium as a Measure of Financial Integration

• The cross-market premium reflects the deviation between prices of identical assets (stocks and their DRs) in different markets.
• If markets are fully integrated, the premium should be close to zero.
• Transaction costs and capital controls influence the width of the no-arbitrage band—the range within which price deviations do not trigger arbitrage.
• Crisis episodes are associated with higher volatility but not with more persistent deviations from LOOP.

2. Non-linearities in Price Convergence

• The paper argues that non-linear convergence processes are better captured by TAR models than by linear AR models.
• Non-arbitrage bands are narrow, and deviations outside these bands are quickly arbitraged away, faster than in goods markets.
• The convergence speed is slower in linear models and faster in TAR models, with the difference proportional to the band width.

3. Liquidity and Financial Integration

• Market liquidity enhances financial integration: more liquid stocks have narrower no-arbitrage bands and faster convergence.
• Liquidity is measured by the trading volume of both the underlying stock and the DR.
• The results suggest that financial markets exhibit lower transaction costs and less persistent price deviations compared to goods markets.

4. Capital Controls and Market Segmentation

• Capital controls, when binding, widen the no-arbitrage bands and increase the persistence of price disparities.
• This implies that capital controls weaken financial integration by creating barriers to arbitrage.

5. Methodology and Data

• The paper uses daily cross-market premium data from 1990 to 2004 for 76 stocks from nine emerging economies.
• The selected countries include: Argentina, Brazil, Chile, Indonesia, Korea (South), Mexico, Russia, South Africa, and Venezuela.
• Data is sourced from Bloomberg, and for some countries, opening prices in New York are used due to time zone differences.
• The TAR model is adapted to account for serial correlation and GARCH effects.

6. TAR Model Specification

• The model is defined as TAR(k,2,d), where:
  • $k$ is the autoregressive length.
  • 2 represents the number of thresholds.
  • $d$ is the threshold lag, set to 1.
• The model includes:
  • A Band-TAR specification that allows for asymmetric convergence.
  • A GARCH correction for volatility clustering.
  • Monte Carlo simulations to derive critical values for the likelihood ratio test.

7. Empirical Results

• The average cross-market premium is close to zero across all countries.
• The mean premium for the pooled data is 0.15%, with standard deviation of 1.65%.
• The largest average premium is observed in Korea at 1.59%.
• No-arbitrage bands range from 0.05% (South Africa) to 0.18% (Mexico).
• Half-life of convergence is typically less than a day in financial markets, much faster than in goods markets.
• The TAR model outperforms the AR model in capturing the speed and non-linearity of price convergence.

Key Findings

• Financial markets are highly integrated, with small and short-lived deviations from LOOP.
• Non-linearities in price behavior are significant and best captured by TAR models.
• Liquidity is a key determinant of financial integration.
• Capital controls contribute to market segmentation by increasing the width of no-arbitrage bands and the persistence of price deviations.
• Crisis periods are associated with greater volatility, but not with more persistent deviations from LOOP.

Conclusion

The cross-market premium is a robust and accurate measure of international financial integration. It reflects the presence of transaction costs and capital controls, and its behavior is best modeled using TAR models. The results support the hypothesis that financial markets are more integrated than goods markets, and that liquidity and the removal of capital controls enhance this integration. The paper also highlights the need for further research on the effects of crises on financial integration.