EBA欧洲银行-Session-II-Olivier-de-Bandt_39页_1mb

Summary of "Stress Testing, Financial Networks and Behavioral responses"

Core Content

This document discusses the integration of financial networks into stress testing frameworks, focusing on the interconnections between banks and insurance companies, as well as the modeling of contagion and liquidity shocks. It presents a three-round stress testing model inspired by the ACPR's Mercure (2015) and Battiston et al. (2016), with applications to the EBA 2016 stress test.

Main Points

1. Definitions and Contagion Model

• The document defines three rounds of contagion:
  1. First Round: Direct shock to a bank's equity or assets.
  2. Second Round: Contagion through interbank exposures using the DebtRank method.
  3. Third Round: Fire sales and liquidity effects on asset prices.
• The DebtRank method is introduced, which uses a leverage matrix and a transmission function to model how a shock to one bank propagates through the network.
• The fire sale model is used to simulate the impact of asset sales on market prices, considering both linear and non-linear effects.

2. Transmission Function

• The function $ f $ is used to model how the distress of one bank affects the obligations of others.
• $ f $ is calibrated using the correlation of Probability of Default (PD) changes between banks.
• In times of stress, the correlation between PD changes increases significantly.

3. RWA Adjustments

• Banks adjust their balance sheets to maintain a target CET1 ratio, selling assets in distress.
• The model accounts for:
  • Expected Loss (EL) based on PD and LGD.
  • Risk-Weighted Assets (RWA) depending on the risk category of the assets sold.
• Asset categories are classified as:
  • Securities (can be sold in distress)
  • Cash (no RWA impact)
  • Loans (can be repriced but not easily sold)
  • Others (tax assets, tangible and intangible fixed assets)

4. Application to EBA Stress Test 2016

• The model is applied to the EBA 2016 stress test, focusing on 4 French Global Systemically Important Banks (G-SIBs).
• The CET1 ratio of these banks is reduced through:
  • First round: -24.9%
  • Second round (interbank contagion): -0.002%
  • Third round (fire sales): -1.07%
  • RWA adjustment (closing 25% of the gap): -18%
  • RWA adjustment (targeting 8% CET1 ratio): -5%
• After 1000 iterations of the second and third rounds, the CET1 ratio for the 4 French G-SIBs is reduced to 9.333%.

5. Liquidity Spillovers

• Liquidity effects are analyzed, showing evidence of liquidity hoarding after a shock.
• The EURIBOR-TBILL Spread is used to assess the impact of liquidity spillovers.
• The document emphasizes that liquidity effects need to be further integrated into the stress testing framework.

6. Non-linear Impact on Asset Prices

• A non-linear market impact function is introduced, based on the depth of the market and the volume of assets sold.
• The function is defined as:
  $$
  \frac{\Delta S^{\mu}}{S^{\mu}} = -\Psi_{\mu}(q)
  $$
  with:
  $$
  \Psi_{\mu}(q, S) = \left(1 - \frac{B_{\mu}}{S}\right) \left(1 - \exp \left(- \frac{q}{\delta_{\mu}}\right)\right)
  $$
  where $ B_{\mu} $ is a floor on the price decrease (set at 50%).

7. Robustness Checks

• The conclusions about RWA adjustments being counterproductive remain unchanged even with a linear price impact model.
• The study highlights the need for sensitivity analysis due to the parametric assumptions in the model.

Key Information

• French Financial Network: The network includes 27 G-SIB banks, with interconnections between banks and insurance companies.
• CET1 Ratio Impact: The CET1 ratio of French G-SIBs drops from 12.7% to 9.3% after three rounds of stress testing.
• Fire Sales: Have a significant impact on CET1 ratios, potentially more than direct exposures.
• Liquidity Effects: Liquidity hoarding is observed after a shock, suggesting the need to include liquidity spillovers in stress testing models.
• Data Sources: COREP and FINREP data are used for modeling RWA adjustments and asset categories.

Conclusion

• Closing the CET1 ratio gap during stress is counterproductive due to negative market reactions.
• The model highlights the importance of considering both capital and liquidity spillovers in stress testing.
• Further research is needed on the market impact of fire sales, liquidity mechanisms, and dynamic balance sheet adjustments.