EBA欧洲银行-Paper-Session-4.-Ingrid-Stein-28Deutsche-Bundesbank29_41页_713kb

报告摘要

Summary of "Risky Banks and Risky Borrowers"

Core Content

This paper investigates how bank distress affects the probability of default (PD) of firms in their loan relationships, particularly in the context of a systemic banking crisis. It focuses on the German corporate loan market, utilizing detailed micro-data to explore the transmission of risk from banks to firms. The study distinguishes between two channels: the bank risk channel, where changes in lending practices and conditions of banks influence firm risk, and the firm risk channel, where firm-specific factors like industry, economic conditions, and idiosyncratic risk affect PD.

The research aims to understand whether distressed banks adjust their loan portfolios in ways that increase or decrease firm risk, and whether the behavior of relationship-oriented banks differs during normal times versus systemic crises.

Main Points

Bank Distress and Risk Transmission: The paper finds that there is a risk pass-through from distressed banks to firms. This effect depends on the firm's idiosyncratic risk and the relationship orientation of the bank.
Distressed Banks Behavior: Distressed banks may either tighten lending conditions, increasing perceived firm riskiness, or loosen credit standards, potentially "evergreening" riskier borrowers to avoid losses.
Systemic Crisis Impact: The study differentiates between normal times and systemic banking crises (e.g., 2008-2010). It suggests that the effects of bank distress may be more pronounced during systemic crises due to regulatory and market pressures.
Data and Methodology: The authors use the Mannheim Enterprise Panel (MUP), a comprehensive dataset of German firms, combined with Deutsche Bundesbank data on bank balance sheets and regulatory actions. They apply nearest-neighbor matching to create a control group of banks not in distress, allowing for a comparison of firm outcomes between treated and control banks.
Propensity Score Matching: The matching is based on observable characteristics of banks in the year before the treatment (capital injection). The matching criteria include:
- Same year of treatment
- At least 3 years of data before and after the treatment
- Same bank type (savings bank, cooperative bank, private bank)
- Same location (Bundesland)

Key Findings

Treated Banks (Distressed Banks): Treated banks tend to be smaller, with lower growth in risk-weighted assets (RWA) and returns on equity (ROE) compared to control banks.
Firm PD Changes: After a capital injection, treated banks experience a decline in the number of distressed customers and a reduction in non-performing loans (NPLs), indicating a possible "evergreening" strategy to reduce losses.
Impact of Bank Type and Location: Banks with similar types and locations are matched to ensure comparable macroeconomic and regulatory environments. The results suggest that the behavior of distressed banks can significantly influence firm PD, especially for those with high idiosyncratic risk.
Limitations of the Analysis: The methodology excludes cases where both capital injection and bank mergers occur, as these events are not independent. This limits the scope of the analysis but ensures a more controlled comparison.

Conclusion

The paper contributes to the literature on financial intermediation and risk transmission by showing that bank distress has a measurable impact on firm PD, and that this impact varies depending on the firm's risk profile and the bank's relationship orientation. The results suggest that during systemic crises, the effect of bank distress on firms is more complex, with potential for both risk mitigation and amplification. The use of detailed micro-data and advanced matching techniques provides a robust framework for understanding how banks and firms interact during financial stress.

Key Information

Sample Period: 2000–2012
Data Sources: Mannheim Enterprise Panel (MUP), Deutsche Bundesbank
Methodology: Nearest-neighbor matching, propensity score estimation
Treatment Definition: Banks receiving capital injections (initial capital support)
Control Group: Banks not in distress, matched on characteristics
Outcome Variable: Firm probability of default (PD)
Key Variables Analyzed:
- Bank characteristics: Total assets, total loans, NPL ratios, reserves ratio, hidden liabilities
- Firm characteristics: Number of customers, share of distressed customers, single relationship customers, main bank customers, regional concentration