EBA欧洲银行-Presenter_2_Giuseppe-Maddaloni_31页_264kb

Summary of "Bank Risks and Liquidity Dynamics: Evidence from the Euro Area Financial Crisis"

Core Content

This document presents a study on the liquidity dynamics of euro area banks during the financial crisis, focusing on how central bank interventions, such as the European Central Bank (ECB)'s Long-Term Refinancing Operations (LTROs) and Medium-Term Refinancing Operations (MROs), influenced banks' liquidity management. The research introduces a framework for modeling liquidity risk and the ECB's response to it, using a combination of econometric models and empirical analysis.

Main Objectives

• To develop a methodology for understanding and managing banks' liquidity risks during financial stress
• To estimate the size of liquidity shocks and the required precautionary liquidity buffers under different confidence levels
• To test the dynamic equilibrium between liquidity demand and supply, incorporating central bank actions and market innovations

Key Findings

• Liquidity Sensitivity: Euro area banks' liquidity dynamics were highly sensitive to stress-induced shocks, showing long memory in conditional variances and disturbances.
• Confidence Levels: Precautionary liquidity buffers at the 0.95 confidence level should be increased by 11.3% compared to the daily expected liquidity change. For higher confidence levels (0.975 and 0.99), the required increases are 21.9% and 44%, respectively.
• GARCH Model: The GARCH (1,1) process explains 59% of the liquidity reserve changes, with the remaining 41% attributed to a long-memory process. The GARCH parameters sum to 0.99, indicating strong persistence in liquidity shocks.
• Distribution Assumption: The normal distribution hypothesis is rejected, and the Generalised Error Distribution (GED) with a leptokurtic shape (GED parameter = 0.71) is used to model liquidity shocks.

Bank's Problem

• Banks aim to balance the cost of liquidity risk (payment failures) against the opportunity cost of holding excess cash.
• The optimal liquidity choice is determined by minimizing the distance between the value of assets and the foregone yield from holding cash.
• The bank's decision is influenced by past information and innovations, with the reaction to shocks captured by the difference between the expected and actual liquidity changes.

Central Bank's Problem

• The ECB's loss function is convex and grows with the banking system's financial risks, which are not under its direct control.
• The central bank provides liquidity in response to shocks, aiming to offset the impact of financial risks.
• The relationship between the ECB's liquidity supply and the banks' liquidity risk is captured by the equation $\rho_{\theta} \Delta \theta_{s+1} = -\rho_{M} \Delta M_{s+1}$.

Liquidity Dynamic Equilibrium

• The equilibrium condition between liquidity demand and supply is captured by the equation:
  $$
  l_{s+1} = m_{s+1} + E_s a_{s+1} + E_s r_{s+1} + \sigma_{s+1}^{ar} - E_s y_{s+1} - \sigma_{s+1}^{yl} + \xi_{s+1}
  $$
• Liquidity shocks are modeled using the residuals $\xi_{s+1}$, which are estimated through a GARCH (1,1) process.
• The GARCH model is estimated using daily log first-differences of financial variables and is validated using EViews 7 software.

GARCH Estimation Results

• Variance Equation:

  • Constant: 3.65E-05
  • RESID(-1)^2: 0.053890
  • GARCH(-1): 0.936990
  • GED PARAMETER: 0.709583

• Model Fit:

  • R-squared: 0.589568
  • Adjusted R-squared: 0.588109
  • S.E. of regression: 0.109416
  • Log likelihood: 3877.850
  • Durbin-Watson stat: 2.254802

Liquidity Risk Management

• Liquidity Shortfall (LS): Defined as the smallest liquidity shortfall occurring with a probability at most $1 - \alpha$. For example, $LS(0.95) = 0.107$, $LS(0.975) = 0.198$, and $LS(0.99) = 0.365$.
• Expected Liquidity Shortfall (ELS): The expected value of liquidity shortfalls given the distribution. For the same confidence levels, $ELS(0.95) = 0.298$, $ELS(0.975) = 0.456$, and $ELS(0.99) = 0.727$.
• Confidence Level Adjustment: Precautionary liquidity buffers are adjusted based on expected liquidity changes and confidence levels, with higher confidence levels requiring larger buffers.

Policy Implications

• Precautionary liquidity buffers can complement traditional liquidity requirements.
• The model can be used for stress testing by applying GARCH-estimated shocks to banks' liquidity positions.
• Sensitivities to risk-related variables can be stressed using multiples of their GARCH-estimated standard deviation to assess the adequacy of liquidity buffers.

Methodology

• The study uses daily financial data from January 2007 to December 2016.
• Data sources include the ECB, European Money Market Institute (EMMI), and RepoFunds Rate (RFR) platforms.
• The model incorporates log first-differences and conditional expectations, assuming a random walk for the liquidity change process.
• The GARCH model is used to estimate the persistence and volatility of liquidity shocks, with the GED distribution capturing the leptokurtic nature of financial time series.

Conclusion

The document provides a comprehensive framework for understanding liquidity dynamics in the euro area banking system during the financial crisis. It emphasizes the importance of precautionary liquidity buffers and the role of central bank interventions in mitigating liquidity risks. The model's empirical results, based on a GARCH (1,1) approach, offer actionable insights for stress testing and liquidity risk management.