2005年-ECB欧洲央行_Financial_Market_Contagion_8页_173kb

Summary of B Financial Market Contagion

Core Content

The document explores the concept of financial market contagion, focusing on its identification, measurement, and policy implications. It outlines the main channels through which contagion spreads, including physical exposure and asymmetric information, and presents various empirical methods used to detect it. The study also reviews evidence of contagion across different regions and asset classes, emphasizing that while contagion is a relevant phenomenon, it is relatively rare and often limited in scope.

Main Channels of Financial Contagion

1. Physical Exposure

   • Occurs when a financial crisis in one market affects traders' portfolios in other markets.
   • Example: A crash in one market reduces traders' wealth, leading them to rebalance portfolios and sell assets in other markets, even if those markets are not fundamentally related.

2. Asymmetric Information

   • Traders may misinterpret price movements in one market as signals about their own market, leading to excessive price spillovers.
   • Reinforces the transmission of shocks through portfolio rebalancing.

Policy Relevance

• Contagion can act as an externality, leading to inefficient risk allocation.
• Ex ante policies (e.g., regulation) and ex post interventions (e.g., stabilizing measures) are needed to mitigate its effects.
• Widespread contagion can destabilize the financial system and harm economic growth, thus justifying macroeconomic stabilization policies.

Identification of Market Contagion

The literature proposes five main criteria to identify contagion:

1. Decline in asset prices leading to declines in other asset prices.
2. Increased interdependence during crises compared to normal times.
3. Excess co-movements beyond what is explained by economic fundamentals.
4. Negative extremes, such as market crashes, indicating crisis situations.
5. Propagation over time, rather than simultaneous effects from common shocks.

Different methods yield varying results, and there is no consensus on which approach is the most effective.

Empirical Approaches

1. Conditional Spillover Probabilities

   • Limited dependent variable estimations: Eichengreen et al. (1996) and Bae et al. (2003) use probit and multinomial logit models to estimate the likelihood of financial crises spreading.
     • Findings: Contagion is more likely between Latin America and Asia than between Asia and the US. Europe appears relatively sheltered.

2. Quantile Regressions and Co-Movement Box

   • Cappiello et al. (2005) use quantile regressions to estimate conditional spillover probabilities.
   • A co-movement box is created to visualize co-movement probabilities across different return thresholds.
   • Evidence: Strong contagion between Argentina and Brazil, but not statistically significant in other cases.

3. Extreme Value Theory (EVT)

   • Longin and Solnik (2001) and Hartmann et al. (2004) apply EVT to estimate extreme market spillovers.
   • Focus on extreme negative returns (crashes) rather than regular returns.
   • Findings:
     • Contagion within the G5 countries is relatively low.
     • The flight to quality is observed, where stock market crashes are accompanied by government bond market booms.
     • The US government bond market acts as a safe haven for other countries.

Selected Evidence

• G5 Countries:

  • Domestic and cross-border contagion risk is generally low, with the US bond market serving as a safe haven.
  • Table B.1 shows that contagion probabilities vary across countries and asset classes, with Japan and France showing higher contagion risks.
  • Flight to quality is more frequent than contagion in some cases.

• Europe:

  • European stock markets are relatively insulated from large overseas downturns.
  • However, they are more exposed to Latin American shocks than to those from Asia or the US.
  • Chart B.1 illustrates the conditional probability of spillovers from Asia, Latin America, and the US to Europe.

• Emerging Markets (EMEs):

  • Contagion is observed among some Latin American EMEs during crises like the "Tequila" crisis (1994), "Asian flu" (1997), and "Russian virus" (1998).
  • However, it is not statistically significant in all cases, indicating limited and occasional spread of contagion.

Key Findings

• Financial market contagion is relevant but rare, especially in severe forms.
• It is more common within the same asset class than across different ones.
• Standard correlation coefficients are not reliable indicators of contagion.
• Extreme value theory and conditional spillover probabilities provide more accurate measures of contagion during crises.
• Flight to quality is a significant phenomenon that limits the breadth of contagion.
• Distance does not always shelter countries from contagion, especially in highly integrated capital markets.

Conclusion

• While several methods exist to identify financial market contagion, no single method is universally accepted.
• Contagion is best detected through extreme market events and conditional spillover probabilities.
• The prevalence and breadth of contagion are generally limited, and it is less frequent across asset classes than within them.
• Policymakers should focus on preventing extreme contagion and maintaining financial stability through international surveillance and regulatory standards.