Discretionary Credit Rating and Bank Stability in a Financial Crisis

Abstract

This article studies the incentives for discretionary credit risk assessment under current banking regulations. We use Slovenian data on the credit ratings of nonfinancial enterprises and analyze their reliability as predictors of corporate default to test whether banks in financial distress systematically underestimate credit risk. Our results show that the predictive accuracy of credit ratings deteriorated during the Great Recession both in absolute terms and relative to the benchmark econometric model that uses publicly available data only. Predictive accuracy was lowest for domestically owned banks and, within this group, for small banks. These results can be linked to incentives to underestimate credit risk due to exposure to nonperforming loans and the limitations on raising additional capital. Given that credit ratings are closely related to loan-loss provisions, our analysis indicates that underestimation of credit risk served to inflate banks’ books. These findings can rationalize the results of the comprehensive review of the Slovenian banking system in 2013, which revealed significant capital shortfalls, on average, but also significant differences in capital shortfalls across groups of banks with different incentives to underestimate risk. Robustness checks confirm the validity of our conclusions. Our findings provide a plausible explanation for the results of a similar comprehensive review in the euro area prior to the launch of the Single Supervisory Mechanism in 2014.

Keywords:

• discretionary credit ratings
• Great Recession
• probability of default
• underestimation of credit risk

JEL-Codes: G01:

• G21
• G28
• G32
• G33

ACKNOWLEDGMENTS

We would like to thank the participants of the 3rd European Banking Authority Research Policy Workshop (https://www.eba.europa.eu/news-press/calendar?p_p_id=8&_8_struts_action=%2Fcalendar%2Fview_event&_8_eventId=846258). and the seminar participants at the University of Salerno and University of Ljubljana. Special thanks to Matej Marinč for early discussions and further valuable contribution to our work.

Notes

¹ This feature is present both in banks that use the internal ratings base and banks that use the standardized approach under Basel II.

² Such incentives might be amplified if the regulators make the minimum capital requirements stricter in times of economic downturn. Such was the case when the European Banking Authority, with the aim of boosting confidence in the European banking system in 2011, set the provision that a minimum of 9% of risk-weighted assets should be held in the form of Core Tier 1 capital.

³ For details, see the Report on the comprehensive review of the banking system, Bank of Slovenia (2013b)

⁴ Such capital shortages indicate potential problems with insolvency. In fact, for two small, domestically owned, private banks, the central bank had initiated insolvency procedures before the results of the comprehensive review were published. These two banks were subsequently excluded from the stress test part of the comprehensive review.

⁵ This could be any significant financial difficulties of the debtor, default on obligations to the bank, information about potential bankruptcy, financial reorganization, decrease of estimated future cash flows, or other changes that could represent a loss for the bank.

⁶ All foreign-owned banks operating in Slovenia in the period 2006–2012 can be classified as small.

⁷ SID bank was established with the special purpose of securing international trade deals, and it enjoys an explicit government guarantee for its liabilities. During the crisis, it served as a vehicle to stimulate corporate lending through state guarantee schemes and for disbursement of loans of international financial institutions.

⁸ This appeared to have happened despite the fast expansion of lending activity by foreign-owned banks, which actually led the pace of credit expansion in Slovenia prior to the crisis. In the period 2003–2008, the total amount of loans outstanding of foreign-owned banks expanded by 372%, while those of domestically owned small and large banks grew by 274% and 207%, respectively.

⁹ Although better risk management and corporate governance and “cherry picking” represent plausible explanations for the lower share of NPL burned reported in Table 4 it is important to note that the comprehensive review revealed a significant 78% capital shortfall for foreign-owned banks also. This demonstrated that, on average, risk management and corporate governance were not at a high level in absolute terms.

¹⁰ The Financial Stability Review of the Bank of Slovenia (Bank of Slovenia, FSR 2013a) documented that throughout the crisis, smaller domestic banks had, on average, higher capital requirements for credit risk on overdue and high-risk exposures, with foreign-owned banks on the opposite side of the spectrum. Similar developments were observed regarding capital adequacy. During the crisis, it increased the most for the group of foreign-owned banks, while domestic banks, especially small ones, experienced significant difficulties in raising additional capital.

¹¹ The incentives to underestimate risk cannot be linked to differences in capital buffers formed before the crisis. In 2008, the average capital adequacy ratio of foreign-owned banks was at 10.6% below average.

¹² The second important quantity is the estimate of the loss given default, which depends on the valuation of underlying collateral. Valuation of collateral is another area where banks can apply discretion to inflate their books. Given that the focus of our analysis is the informativeness of credit rating, we focus our discussion on discretion in modeling the probability of default.

¹³ We also estimate the credit ratings model using clustering standard errors across firms. This only marginally increases the standard errors, and all the coefficients are still significant at less than 1% probability of type I error. The results are available upon request.

¹⁴ To distinguish between defaulters and nondefaulters, we use the standard 0.5 cut-off of predicted probability.

¹⁵ Slovenia slid into a recession in the fourth quarter of 2008.

¹⁶ To some extent, this can explained by the fact that with the crisis, the of firms in default increased considerably. In estimation of a bankruptcy-prediction model, this implies that a higher probability mass is accounted for by the default cases, which facilitates more precise discrimination between healthy firms and firms in default. See Brezigar-Masten and Masten (2012) for a discussion.

¹⁷ While the worst classification accuracy of large domestic banks before the crisis is consistent with moral hazard explanations of the incentives to underestimate risk, this no longer prevailed in the crisis. The fact that the large banks outperform the small banks in classification accuracy indicates a stronger effect of smaller incentives to underestimate risk due to easier access to capital and funding.

¹⁸ We also checked the robustness of our results with respect to the forecast horizon. In particular, horizons between 1 and 8 quarters were considered. Our conclusions, obtained in the previous section, are robust to different forecast horizons. Results available upon request.

¹⁹ Results for different prediction horizons are similar and available upon request.

Additional information

Notes on contributors

Arjana Brezigar-Masten

Arjana Brezigar-Masten is an Assistant Professor with the Faculty of Mathematics, Natural Sciences and Information Technology and Institute of Macroeconomic Analysis and Development at the University of Primorska, Slovenia.

Igor Masten

Igor Masten is an Associate Professor at the Faculty of Economics at the University of Ljubljana, Slovenia and Head of Research at the Bank of Slovenia.

Matjaž Volk

Matjaž Volk is an Economist at the Bank of Slovenia.