2013年-IMF国际货币组织全球_International_Evidence_on_Government_Support_and_Risk_Taking_in_the_Banking_Sector_36页_751kb

Summary of "International Evidence on Government Support and Risk Taking in the Banking Sector"

Core Content

This working paper investigates the relationship between government support and bank risk taking using an international dataset of banks across multiple countries for the periods 2003-2004 and 2009-2010. The authors aim to determine whether government support increases bank risk taking through the market discipline channel or the charter value channel, and whether bank regulation can mitigate this effect.

Main Findings

1. Government support increases risk taking:

   • Banks with higher levels of government support are associated with greater risk taking, as measured by the z-score (a composite indicator of credit and market risk).
   • This relationship is stronger during the financial crisis (2009-2010) compared to the pre-crisis period (2003-2004).
   • The effect is robust to endogeneity concerns, suggesting that government support directly influences risk-taking behavior.

2. Regulation mitigates the moral hazard effect:

   • Restricting banks' range of activities (e.g., limiting participation in securities markets, insurance, real estate, and non-financial firm ownership) reduces the magnitude of the moral hazard problem caused by government support.
   • Capital requirements alone were insufficient to curb risk taking during the crisis, but activity restrictions had a notable impact in limiting risk.

3. Market discipline is the dominant channel:

   • The authors find that market discipline (i.e., reduced investor monitoring due to government guarantees) is the primary mechanism through which government support affects risk taking.
   • They do not support the charter value hypothesis, which suggests that government support lowers funding costs and increases bank rents, thereby reducing risk.

Key Concepts and Definitions

• Bank risk taking is measured using the z-score, which is calculated as:
  $$
  \text{z-score} = \frac{\text{ROA} + \text{CAR}}{\sigma(\text{ROA})}
  $$
  where ROA is return on assets, CAR is capital to asset ratio, and $\sigma$ is the standard deviation of ROA. A higher z-score indicates lower risk.

• Government support is measured as the difference in rating notches between a bank's financial strength rating (BFSR) and its deposit rating. This is based on Moody's and Fitch ratings, with Moody's measure showing stronger predictive power for bailouts.

• Control variables include:

  • Bank-specific: revenue growth, size (log of total assets), liquidity (liquid assets to liquid liabilities), ownership structure (cash flow rights of large shareholders, ownership type).
  • Country-level: per capita income, inflation, investor protection, contract enforcement, GDP growth.
  • Regulatory indicators: capital stringency, official supervisory power, activity restrictions.

Methodology

• The study uses cross-country bank data from 54 countries and two time periods: 2003-2004 (pre-crisis) and 2009-2010 (during the crisis).
• The authors test two main hypotheses:
  1. Government support affects risk taking.
  2. Regulation moderates the impact of government support on risk taking.
• To address endogeneity, they use:
  • Saturated regressions to control for omitted variables.
  • Instrumental variables (IV), using the average GS of other banks in the same country as an instrument.

Policy Implications

• Strengthening market discipline is essential to counteract the moral hazard caused by government support.
• Increasing transparency and disclosure can help investors monitor banks more effectively.
• Activity restrictions (e.g., limiting banks to core financial activities) can reduce the risk-taking incentives of banks with government support.
• Simple regulatory rules, such as those in the Glass-Steagall Act, may help in curbing risk-taking behavior in the presence of government support.

Comparison with Previous Studies

• Earlier studies have shown mixed results, with some supporting the charter value channel and others the market discipline channel.
• The authors challenge the charter value hypothesis and emphasize the dominant role of market discipline.
• They use a broader measure of risk (z-score) and additional controls that differentiate their findings from prior research.

Conclusion

The paper concludes that government support encourages risk taking in the banking sector, particularly during financial crises. However, restricting bank activities can mitigate this moral hazard effect. Therefore, enhancing market discipline and implementing activity restrictions are important policy tools to manage the risks associated with government support. The results also suggest that pre-crisis capital requirements were ineffective in reducing the moral hazard, highlighting the need for more comprehensive regulatory measures.