2013年-IMF国际货币组织全球_Near_33页_1mb

报告摘要

Summary of "Near-Coincident" Indicators of Systemic Stress

Core Content

This paper by Ivailo Arsov, Elie Canetti, Laura Kodres, and Srobona Mitra explores the development and evaluation of "near-coincident" indicators to serve as early warning signals for systemic financial stress in the United States and the euro area. These indicators are designed to capture the likelihood of systemic events by analyzing market-based data and institutional exposures.

The paper is part of the IMF Working Paper series and addresses the G-20 Data Gaps Initiative, which calls for standard measures of tail risk (equated with systemic risk in this context). It emphasizes the complexity and interconnectedness of financial systems and the difficulty of predicting systemic crises with precision.

Main Points

1. Systemic Risk and Tail Events

Systemic risk is defined as the potential for disruption in the flow of financial services that could have serious negative consequences for the real economy.
Tail events are systemic events that represent extreme financial stress.
The paper proposes a systemic financial stress index (SFS) based on abnormal equity returns of financial institutions relative to the benchmark index of their host country.

2. Types of Indicators

The paper evaluates 11 near-coincident indicators, which are based on market data, balance sheet data, or a combination of both.
Key indicators include:
- Equity market-based: VIX, VSTOXX, Credit Suisse Fear Barometer
- Debt market-based: JPoD, Diebold-Yilmaz
- Balance sheet and equity market-based: Systemic Contingent Claims Analysis (SCCA), Distance-to-Default (DD), Time-Varying Conditional Value at Risk (T-CoVaR)
- Macro-based: Yield curve slope, LIBOR-OIS spread, Systemic Liquidity Risk Indicator (SLRI)

3. Performance Evaluation

The performance of these indicators is evaluated using three tests:
- Test 1: Granger Causality – to determine whether the indicators can forecast systemic stress.
- Test 2: Predicting Extreme Events – to assess the ability of the indicators to predict extreme systemic stress (SFS ≥ 0.25).
- Test 3: Early Turning Point – to identify the timing of shifts from tranquil to volatile conditions.
Two indicators perform particularly well in both regions:
- Time-Varying CoVaR (T-CoVaR)
- Distance-to-Default (DD)
Other simple indicators, such as the LIBOR-OIS spread and the yield curve slope, also show strong performance across several criteria.

4. Key Findings

Institution size does not necessarily correlate with systemic risk contribution. Some smaller institutions may have a higher spillover risk than larger ones.
Interconnectedness among financial institutions plays a significant role in amplifying losses during crises.
Market-based indicators can be more effective than exogenous events (e.g., official interventions) in signaling systemic stress, as they reflect market perceptions and responses in real time.
The SFS is used as a proxy for actual systemic stress events, capturing the fraction of institutions with large negative abnormal returns.

5. Robustness of Results

The results are robust to changes in test specifications.
The order of some indicators can change depending on the test, but the top two indicators remain consistent across both regions.

Key Indicators and Their Performance

Indicator	U.S. Performance	Euro Area Performance
T-CoVaR	Top performer	Top performer
Diebold-Yilmaz	Top performer	Top performer
Distance-to-Default (DD)	Top 3 in U.S.	Top 3 in Euro Area
LIBOR-OIS spread	Strong in predicting extreme events	Best overall near-coincident indicator in Euro Area
Yield curve slope	Strong in predicting systemic stress	Strong in predicting systemic stress
VIX	Performs better than complex indicators	Performs better than complex indicators
JPoD	Best at predicting extreme events	Performs relatively poorly in U.S.
SCCA	Complex model, performs well in U.S.	Performs well in U.S.
SLRI	Better in Euro Area than in U.S.	Better in Euro Area than in U.S.
Credit Suisse Fear Barometer	Purely equity-based	Purely equity-based
VSTOXX	Euro Area equivalent of VIX	Euro Area equivalent of VIX

Policy Implications

Near-coincident indicators can provide early warning signals for policymakers to prepare for financial crises, even if only a few weeks or months in advance.
These indicators can be used to identify the need for recapitalization or release of capital buffers.
The lack of exposure data between institutions is a major data gap that could be addressed to improve early warning capabilities.
Market-based indicators are useful in assessing systemic risk and spillover effects, especially when interconnectedness is high.

Conclusion

The paper concludes that while systemic risk is difficult to predict, near-coincident indicators offer a practical and effective way to monitor the likelihood of systemic stress.
These indicators are particularly useful for early intervention and policy planning.
The focus on market-based signals and interconnectedness is crucial in understanding the amplification of shocks in financial systems.