Abstract
Several empirical studies have identified unique characteristics of banks that subsequently failed during the Great Financial Crisis. The notion is that by identifying these risk characteristics we are better able to monitor and regulate the risks to banks during the next crisis. A concern is bank failure is a relatively rare event, therefore inferences based on a single model specification can be sensitive to the choice of variables. We re-examine three studies (DeYoung and Torna in J Financ Intermed 22:397–421, 2013; Jin et al. in J Bank Finance 35:2811–2819, 2011; Ng and Roychowdhury in Rev Acc Stud 19:1234–1279, 2014) of bank failures during the Great Financial Crisis to determine whether these authors’ main findings are robust to accounting for uncertainty in the model’s specification. Our results indicate their results are not robust and that the causes of bank failures during the Great Financial Crisis are similar to those of past periods of crisis and are driven by traditional measures of risk.
Similar content being viewed by others
Data Availability
The data used in this study are all from publicly available sources.
Code availability
The R-package BMA (Raftery et al. 2018) is publicly available. The code and dataset necessary to replicate the results will be made available on the author’s website on acceptance.
Notes
Reliance on brokered deposits as a source of funds (liability) can create liquidity issues for a bank during a crisis due to the volatility of their withdrawal, relative to core deposits.
Harvey (2017) notes p-hacking may also involve choice of estimation method (e.g. logit vs survival model) and sample selection (e.g. observation exclusion).
The odds equal = \(\Omega (H_{0} |D) = \frac{{P(H_{0} |D)}}{{1 - P(H_{0} |D)}}\) therefore \(P(H_{0} |D) = \frac{{\Omega (H_{0} |D)}}{{1 + \Omega (H_{0} |D)}}\)
For the logit model the number of observations is equal to the sample size, whereas the Cox proportional hazards model uses the number of events, i.e. failures.
See Enkhtaivan and Lu (2021) for a thorough overview of TARP implementation.
The FDIC uses similar variables in their statistical CAMELS off-site rating (SCOR) model to predict changes in CAMELS ratings (Collier et al. 2003).
Lane et al. (1986) interpret that measures of loan composition reflect management quality. Collier et al. (2003), however, believe that management quality cannot be identified with any financial ratio. An alternative approach to identify differences in management quality is to use textual analysis to reveal differences in banks’ culture, which Luu et al., (2023) observe influences bank stability. Differences in corporate governance measures have also been shown (Alzayed et al. 2023) to influence bank stability. It should be noted that these measures of bank culture and corporate governance are only available for very small samples of the population of US banks.
Wicker (1980), for example, discusses how the failure of the investment bank Caldwell and Company, the largest in the South, contributed directly to the closing of 120 banks affiliated with the firm in a two-week period in November and December of 1930. Wicker (1980) argues other failures in the period originated from the uncertainty caused by Caldwell’s collapse. A result Wicker (1980) notes is due to Caldwell’s heavy borrowing from bank affiliates, which was used to finance the purchase of municipal securities for trading purposes.
Torna (August 29, 2019) indicated in a personal communication that they (DeYoung and Torna 2013) no longer had access to the data or the code needed to replicate exactly their sample and results.
These result appear in Appendix Table 14
The allowances for loan losses counted for risk based purposes deduct the allocated transfer risk reserve and add allowances for credit losses on off-balance sheet credit exposures.
The restrictions are a result of the construction of the control variables, which in some cases use lagged and unlagged values, to avoid dividing by zero.
See Appendix Table 16.
Roychowdhury (August 26, 2019) indicated in a personal communication they (Ng and Roychowdhury 2014) no longer had access to the data or the code needed to replicate their data and results so it is unclear how their timely measure or dataset more generally was constructed.
Using the difference in unadjusted R2 did not materially affect our results.
The marginal effect reported here and for the effect of equity are based on the authors’ calculations using Ng and Roychowdhury’s (2014) estimates and summary statistics.
These estimates are available in Appendix Table 17. Panel A has the logistic model estimates and panel B the hazard model estimates.
Estimates from the models that consider for inclusion provisions as a share of assets are available in Appendix Table 18. Panel A has the logistic model estimates and panel B the hazard model estimates.
The marginal effect is based on our calculation using estimates and summary statistics reported by Jin et al. (2011), where we compare the difference in probabilities evaluated at the variables’ mean values for a bank with and without a Big 4 auditor and first quarter data.
The measures and call report series we used in their construction are included in Appendix Table 19.
Summary statistics for the variables are available in Appendix 20.
The link table is available at https://www.newyorkfed.org/research/banking_research/datasets.html.
The series (RCFDB705, RCFDB706, RCFDB707, RCFDB708, RCFDB709, RCFDB710, RCFDB711) are used to construct the numerator, which is scaled by total assets.
Series RCON5571 is defined in the Federal Reserve’s Micro Data Reference Manual as “amount currently outstanding of commercial and industrial loans to U.S. addressees (in domestic offices) with original amounts of $100,000 or less”. Source: https://www.federalreserve.gov/apps/mdrm/data-dictionary. The other two series (RCON5573, RCON5575) are also related to commercial and industrial (C & I) loans with loan amounts of more than $100,000 to $250,000 and more than $250,000.
Charge-offs, for example, were almost twice as high in the fourth quarter 2007 than in 2006.
These estimates appear in Appendix Table 23.
References
Alali F, Jaggi B (2011) Earnings versus capital ratios management: Role of bank types and SFAS 114. Rev Quant Financ Acc 36:105–132
Alzayed N, Eskandari R, Yazdifar H (2023) Bank failure prediction: Corporate governance and financial indicators. Rev Quant Financ Acc 61:601–631
Bank for International Settlements, 2009. Revisions to the Basel II market risk framework
Beatty A, Liao S (2011) Do delays in expected loss recognition affect banks’ willingness to lend? J Account Econ 52:1–20
Berg T, Koziol P (2017) An analysis of the consistency of banks’ internal ratings. J Bank Finance 78:27–41
Berger AN, Bouwman CHS (2013) How does capital affect bank performance during financial crises? J Financ Econ 109:146–176
Bessler W, Kurmann P (2014) Bank risk factors and changing risk exposures: capital market evidence before and during the financial crisis. J Financ Stab 13:151–166
Chen Z, Liu FH, Opong K, Zhou M (2017) Short-term safety or long-term-failure? Empirical evidence of the impact of securitization on bank risk. J Int Money Financ 72:48–74
Cleary S, Hebb G (2016) An efficient and functional model for predicting bank distress: in and out of sample evidence. J Bank Finance 64:101–111
Cole RA, Gunther JW (1998) Predicting bank failures: a comparison of on- and off-site monitoring systems. J Financ Serv Res 13:103–117
Cole RA, White LJ (2012) Deja vu all over again: the causes of U.S. commercial bank failures this time around. J Financ Serv Res 42:5–29
Cole RA, Cornyn BG, Gunther JW (1995) FIMS: a new monitoring system for banking organizations. Fed Reserve Bull 81:629–667
Collier C, Forbush S, Nuxoll DA, O’Keefe J (2003) The SCOR system of off-site monitoring: Its objectives, functioning, and performance. FDIC Bank Rev 15:17–32
Cont R (2006) Model uncertainty and its impact on the pricing of derivative instruments. Math Financ 16(3):519–547
Demirguc-Kunt A, Huizinga H (2010) Bank activity and funding strategies: the impact on risk and returns. J Financ Econ 98:626–650
DeYoung R, Li L (2019) Publicly traded versus privately held commercial banks: Sensitivity to growth opportunities. J Financ Serv Res 56:39–71
DeYoung R, Torna G (2013) Nontraditional banking activities and bank failures during the financial crisis. J Financ Intermed 22:397–421
Edwards W, Lindman H, Savage LJ (1963) Bayesian statistical inference for psychological research. Psychol Rev 70:193–242
Enkhtaivan B, Lu W (2021) The effect of TARP on lending: evidence from the lead bank’s share in syndicated loans. Rev Quant Financ Acc 57:1169–1193
FDIC (2006) Quarterly banking profile (Fourth quarter 2006). Retrieved from https://www.fdic.gov/bank/analytical/qbp/qbpmenu.html
Griffin JM, Kruger S, Maturana G (2021) What drove the 2003–2006 house price boom and subsequent collapse? Disentangling competing explanations. J Financ Econ 141(3):1007–1035
Harvey CR (2017) Presidential address: the scientific outlook in financial economics. J Finance 72(4):1399–1440
Jeffreys H (1961) Theory of probability, 3rd edn. Oxford University Press, Oxford
Jin JY, Kanagaretnam K, Lobo GJ (2011) Ability of accounting and audit quality variables to predict bank failure during the financial crisis. J Bank Finance 35:2811–2819
Kerkhof J, Melenberg B, Schumacher H (2010) Model risk and capital reserves. J Bank Finance 34(1):267–279
Kwan SH (2004) Risk and return of publicly held versus privately owned banks. Fed Reserve Bank New York Econ Policy Rev 10:97–107
Lane WR, Looney SW, Wansley JW (1986) An application of the Cox proportional hazards model to bank failure. J Bank Finance 10:511–531
Li J, Wei L, Lee C, Zhu X, Wu D (2018) Financial statements based bank risk aggregation. Rev Quant Financ Acc 50:673–694
Luu HN, Nguyen LTM, Vu KT, Nguyen LQT (2023) The impact of organizational culture on bank stability. Rev Quant Financ Acc 61:501–533
Mason JR, Imerman MB, Lee H (2014) Self-reporting under SEC Reg AB and transparency in securitization Evidence from loan-level disclosure of risk factors in RMBS deals. J Risk Finance 15(4):334–384
Ng J, Roychowdhury S (2014) Do loan loss reserves behave like capital? Evidence from recent bank failures. Rev Acc Stud 19:1234–1279
Preston HH (1933) The banking act of 1933. Am Econ Rev 23:585–607
Raftery AE, Madigan D, Hoeting JA (1997) Bayesian model averaging for linear regression models. J Am Stat Assoc 92:179–191
Raftery AE, Hoeting J, Volinsky C, Painter I, Yeung KY (2018) Package BMA. Retrieved from https://cran.r-project.org/web/packages/BMA/BMA.pdf
Raftery AE (1995) Bayesian model selection in social research. In: Marsden PV (ed) Sociological methodology 1995. Blackwells Publishers, Cambridge, pp 111–195
Volinsky CT, Madigan D, Raftery AE, Kronmal RA (1997) Bayesian model averaging in proportional hazard models: assessing the risk of a stroke. Appl Stat 46:433–448
Whalen G (1991) A proportional hazards model of bank failure: an examination of its usefulness as an early warning tool. Econ Rev 27:21–31
Wicker E (1980) A reconsideration of the causes of the banking panic of 1930. J Econ Hist 40:571–583
Yasuda Y, Okunda S, Konishi M (2004) The relationship between bank risk and earnings management: evidence from Japan. Rev Quant Financ Acc 22:233–248
Acknowledgements
This research was supported by a summer research grant from the Nistler College of Business and Public Administration at the University of North Dakota. I would like to thank Margie Tieslau and conference participants of the 56th Missouri Valley Economic Association annual meeting for their comments. In addition, I thank the two anonymous referees and Cheng-Few Lee (editor) for their insightful suggestions, which helped to improve the paper. Gökhan Torna, Justin Jin, and Sugata Roychowdhury were also helpful in responding to my questions in regard to their studies.
Funding
Not applicable.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author declares that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Goenner, C.F. Robust lessons learned from bank failures during the Great Financial Crisis. Rev Quant Finan Acc 62, 449–498 (2024). https://doi.org/10.1007/s11156-023-01213-9
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11156-023-01213-9