Skip to main content

The recourse rule, regulatory arbitrage, and the financial crisis


In November 2001, regulators finalized the “Recourse Rule.” The rule lowered risk weights, and therefore commercial bank holding company capital requirements, to 0.2 for holdings of AAA- and AA-rated “private label” securitization tranches, created by investment banks and securitizing commercial bank holding company subsidiaries; risk weights for A-rated holdings equaled 0.5. The rule’s aim was to encourage securitization, but not risk-taking. Regulators indicated that the rule would apply to larger holding companies, without identifying them. Using bank holding companies with subsidiaries that commented on the proposed rule-makings as a treatment variable, average treatment effects from a fully flexible difference-in-differences model indicate that treated banks increased their holdings of the highly rated tranches relative to total assets, while other holding companies, on average, did not. Holding companies with greater highly rated tranche holdings also experienced greater increases in risk after Q1 2008, which suggests that poor performance may have been unanticipated.

This is a preview of subscription content, access via your institution.

Fig. 1

Source: Based on author’s estimates

Fig. 2

Source: Based on author’s estimates

Fig. 3

Source: Based on author’s estimates


  1. The full name of the “Recourse Rule” was the final rule covering “Risk-Based Capital Guidelines; Capital Adequacy Guidelines; Capital Maintenance: Capital Treatment of Recourse, Direct Credit Substitutes and Residual Interests in Asset Securitizations.”.

  2. Tier 1 capital included common equity capital, noncumulative perpetual preferred stock, and minority interests in equity capital accounts of consolidated subsidiaries, minus goodwill, other intangible and deferred tax assets disallowed, and any other amounts deducted as determined by the bank’s federal supervisor. Tier 2 capital included cumulative perpetual preferred stock, convertible and subordinated debt and intermediate-term preferred stock, and the allowance for credit losses, such as loan and lease losses, and up to 45% of pretax net unrealized holding gains on available-for-sale equity securities with readily determinable fair values.

  3. In this sense, the results here could be consistent with claims by academics who in the years leading up to the 2007–2009 crisis observed that Basel-style regulatory capital standards create arbitrage opportunities (Merton 1995; Jones 2000; Brealey 2006). Moreover, others have found that while banks satisfied regulatory capital requirements, which rely on book values, market valuations of capital plunged well below book values during the crisis (Flannery 2014; Flannery and Giacomini 2015). It is no surprise then that risk-based capital requirements have a weaker relationship with firm performance and risk than non-risk based capital requirements, such as the so-called leverage ratio. For instance, Demirgüç-Kunt et al. (2013) find a stronger empirical relationship between stock returns and non-risk-based measures of capital than between risk-based measures. Acharya et al. (2014) find no cross-sectional relationship between risk-weighted measures of capital and market risk, while Hogan (2015) and Hogan and Meredith (2016) also find that the non-risk-based leverage ratio better predicts measures of bank risk than risk-based capital measures.

  4. See the discussion in part “I. A. Asset Securitization,” p. 59615. See also for the initial draft of Basel II.

  5. See the blogpost by Kandarp Srinivasan, which includes a link to a longer working paper.

  6. See Chicago Federal Reserve Bank Call Report Y-9C form data.

  7. See Micro Report Series Description (series name: Combined Bank Holding Company File),, 30–31.

  8. To merge the Call Report and CRSP data I use the CRSP-FRB file for 2014-3 available from

  9. See the discussion in part “VI. Regulatory Analysis,” p. 59629.

  10. See the discussion in part “I. C. The Combined Final Rule,” p. 59616.

  11. See A follow up inquiry with the Federal Reserve’s FOI Office revealed that the missing letters were destroyed. A prior FOI Act request at the FDIC turned up nothing, while a subsequent request at the OCC turned up 27 comment letters for the 1997 NPR and 26 comment letters for the 2000 NPR, and no additional banks.

  12. The list in Table 2 may understate the total number of banks that were involved in the notice-and-comment process as some comment letters were submitted by trade associations that may represent a large number of banks. For instance, the Bankers Roundtable submitted a comment for the 1997 NPR, which states “[m]embership in the Roundtable is reserved for the 125 largest banking companies in the United States.” Based on Q4 1997 total assets data, all of the largest, securitizing bank holding companies in the sample used here were among the top 125.

  13. Information concerning TARP funds comes from Only Union Bank California, which as of the writing of this paper exists as MUFG Union, does not appear to have received TARP funds, and it was the only BHC in the list that was not securitization active.

  14. For instance, the Global Capital Index,, lists Bank of America, BNY Mellon, Citigroup, State Street and Wells Fargo as US G-SIBs.

  15. Still, adding additional controls yields qualitatively similar results. The additional variables include one-quarter lagged regulatory capital slack, the fraction of assets held with a risk weight equal to 0 and the fraction of assets held with a risk weight equal to 1. A negative coefficient estimate for the regulatory capital slack variable would suggest that BHCs on average purchase more highly rated tranches when their capital buffer is closer to the regulatory minimum. Negative coefficient estimates for the latter two variables could be consistent with BHCs substituting out of higher or lower risk-buckets in favor of the highly rated tranches.

  16. The differences between the treated BHCs and the securitization active BHCs likely reflect the fact that many of the treated BHCs have a dominant position in securitization, while the securitization active group merely includes any BHC that reports securitization activities, and therefore includes many that may not have a dominant position in the market; unlike the treatment variable used here, the securitization active BHC variable varies over time. The Q2 2001–Q4 2006 median volume of reported securitization activities for BHCs with commenting subsidiaries equaled $14.86 billion, and for securitization active BHCs equaled $350 million, which would lend some support to the idea that the treated BHCs tended to have a more dominant position in the market.

  17. As an additional robustness check, in unreported results I find that the median ratio of highly rated tranche holdings relative to total MBS holdings for BHCs in the treatment group rises from about 0.15 in Q2 2001 to about 0.31 by Q4 2007; for the control group the median ratio is close to zero and exhibits no trend until the end of the sample, when it rises slightly. These findings would be consistent with the Recourse Rule encouraging holdings of highly rated tranche holdings, relative to other MBS, for the treatment group.

  18. The number of observations used in the estimation equals 11,225, and the R-squared equals 0.03. Adding in additional controls yields similar estimates of the treatment effects and also lowers the p value for the associated test of common pre-treatment dynamics to zero. In terms of the additional controls, on average BHC holdings were not sensitive to regulatory capital slack, which suggests that BHCs that were at the regulatory minimum did not subsequently adjust their highly rated tranche share of total assets, even though greater holdings might be used to lower the required minimum amount. BHCs on average also had lower holdings if they had greater allocations to the risk buckets with risk weights of 0 or 1, which reflects the substitution away from the assets with risk weights of 0.2 or 0.5. Still, the interpretation of the results changes in ways that are less of interest here.

  19. Rejecting the hypothesis would not invalidate the applicability of the fully flexible model, but would mean having to make assumptions about whether identification rests on the assumption of parallel paths or parallel growths, since the treatment effects would no longer be identical.

  20. As a robustness check, in unreported results, I estimate average treatment effects for lower rated residual holdings for treated BHCs, and the time paths are relatively flat and zero is always included in the 95 confidence interval. Test statistics indicate that I cannot reject the null hypothesis of common pre-treatment dynamics but do reject the null hypothesis of common post-treatment dynamics. The 95% confidence intervals for all of the average treatment effects under the parallel paths and parallel growths assumptions include zero.

  21. For all models, statistical tests fail to reject the null hypothesis that all bank holding company fixed effects equal zero. Therefore, I do not include them in the specifications, as the results are qualitatively similar.

  22. Laeven and Levine (2009) and Erel et al. (2014) measure the denominator using the standard of lagged return on assets. Taking first differences of the natural log of the z-score would generate a series computed from overlapping samples. Instead, by using intraquarterly daily data, I can generate a “naïve” measure of asset volatility similar to one proposed in the working paper version of Correia et al. (2018). Whereas Correia et al. divide market value of equity by the sum of market value of equity, book value of short-term debt, and book value of long-term debt, I use the book value of assets, which must equal the sum of the book values of equity and all liabilities.

  23. See

  24. I include only post-2007 results because Table A2 of the working paper version shows that the results for highly rated tranches are only economically and statistically significant for the Q1 2008–Q1 2009 period, but not generally before then, and also because the scant data on CDO tranche holdings in the call report data is only available from Q1 2008–Q1 2009. Table A2 of the working paper version reports estimates from eight specifications, four for each measure of changes in BHC risk estimated over the Q1 2002–Q1 2009 period. The specifications include only highly rated residuals, lower rated residuals and agency MBS in the 2nd and 6th columns, similar to the 2nd and 6th columns of Table 4, but without crisis dummy variable interactions. The 3rd and 7th columns include all variables used in the specifications reported in Table 4 estimates, with the exception of CDO tranch holdings, similar to the 3rd and 7th columns of Table 4, again without crisis dummy variable interactions. The 4th and 8th columns report estimates of specifications reported in the 2nd and 6th columns of Table A2 after interacting the variables with a dummy variable for Q2 2007–Q4 2007 period when the crisis began to unfold and a dummy variable for the Q1 2008–Q1 2009 period, after the TSLF program was created. The 5th and 9th columns add crisis dummy variable interactions to the specifications reported in the 3rd and 7th columns of Table A2 after including the same crisis dummy variable interactions. To understand the choice of crisis dummy variables, ratings downgrades began in Q2 2007 during June, while the Federal Reserve announced the creation of the TSLF to purchase securitizations, including highly rated, private-label MBS in Q1 2008 during March (see the website of the Federal Reserve Bank of St. Louis,

  25. Only Bank of America, Citigroup, J.P. Morgan, PNC, Suntrust and Wells Fargo report CDO holdings. While I omit the results, estimates of the same model for Q2 2008–Q1 2009 after dropping the CDO tranche holdings from the specifications generate similar coefficient estimates for the highly rated residual. This suggests that the smaller coefficient estimates for the highly rated tranche holdings arises from omitting Q1 2008 rather than from including CDO tranche holdings, which were not available prior to 2008.


  • Acharya, V., Engle, R., & Pierret, D. (2014). Testing macroprudential stress tests: The risk of regulatory risk weights. Journal of Monetary Economics, 65, 36–53.

    Article  Google Scholar 

  • Acharya, V., Schnabl, P., & Suarez, G. (2013). Securitization without risk transfer. Journal of Financial Economics, 107, 515–536.

    Article  Google Scholar 

  • Admati, A., DeMarzo, P., Hellwig, M., & Pfleiderer, P. (2014). Fallacies and irrelevant facts in the debate on capital regulation. In C. Goodhart, D. Gabor, J. Vestegaard, & I. Erturk (Eds.), Central banks at a crossroads: Europe and beyond (pp. 33–50). London: Anthem Press.

    Google Scholar 

  • Admati, A., & Hellwig, M. (2013). The Bankers’ new clothes. Princeton, NJ: Princeton University Press.

    Google Scholar 

  • Barth, J., & Miller, S. M. (2018). Benefits and costs of a higher bank “leverage ratio”. Journal of Financial Stability, 38, 37–52.

    Article  Google Scholar 

  • Brealey, R. A. (2006). Basel II: The route ahead or Col-de-sac? Journal of Applied Corporate Finance, 4, 34–43.

    Article  Google Scholar 

  • Cordell, L., Huang, Y., & Williams, M. (2012). Collateral damage: Sizing and assessing the subprime CDO crisis. Federal Reserve Bank of Philadelphia Working Paper No. 11-30.

  • Correia, M., Kang, J., & Richardson, S. (2018). Asset volatility. Review of Accounting Studies, 23, 37–94.

    Article  Google Scholar 

  • Demirgüç-Kunt, A., Detragiache, E., & Merrouche, O. (2013). Bank capital: Lessons from the financial crisis. Journal of Money, Credit and Banking, 45, 1147–1164.

    Article  Google Scholar 

  • Erel, I., Nadauld, T., & Stulz, R. M. (2014). Why did holdings of highly rated securitization tranches differ so much across banks? Review of Financial Studies, 27, 404–453.

    Article  Google Scholar 

  • Flannery, M. (2014). Maintaining adequate bank capital. Journal of Money Credit and Banking, 46, 157–180.

    Article  Google Scholar 

  • Flannery, M., & Giacomini, E. (2015). Maintaining adequate bank capital: An empirical analysis of the supervision of European banks. Journal of Banking & Finance, 59, 236–249.

    Article  Google Scholar 

  • Friedman, J., & Kraus, W. (2011). Engineering the financial crisis: Systemic risk and the failure of regulation. Philadelphia: University of Pennsylvania Press.

    Google Scholar 

  • Haldane, A. (2012). “The Dog and the Frisbee” (speech by Andrew Haldane, executive director, financial stability, Bank of England, and Vasileios Madouros, economist, Bank of England, at the Federal Reserve Bank of Kansas City’s Economic Policy Symposium, “The Changing Policy Landscape,” Jackson Hole, Wyoming, August 31, 2012).

  • Hogan, T. L. (2015). Capital and risk in commercial banking: A comparison of capital and risk-based capital ratios. Quarterly Review of Economics and Finance, 57, 32–45.

    Article  Google Scholar 

  • Hogan, T. L., & Meredith, N. R. (2016). Risk and risk-based capital of U.S. bank holding companies. Journal of Regulatory Economics, 49, 86–112.

    Article  Google Scholar 

  • Jones, D. (2000). Emerging problems with the Basel capital accord: Regulatory capital arbitrage and related issues. Journal of Banking & Finance, 24, 35–58.

    Article  Google Scholar 

  • Kapstein, E. (1991). Supervising international banks: Origins and implications of the Basle Accord. Essays in international finance, No. 185. International Finance Section, Princeton University.

  • Kapstein, E. (1994). Governing the global economy. Cambridge, MA: Harvard University Press.

    Google Scholar 

  • Laeven, L., & Levine, R. (2009). Bank governance, regulation, and risk taking. Journal of Financial Economics, 93, 259–275.

    Article  Google Scholar 

  • Loutskina, E., & Strahan, P. E. (2011). Informed and uninformed investment in housing: The downside of diversification. Review of Financial Studies, 24, 1447–1480.

    Article  Google Scholar 

  • Mabel, C. (2001). Bank capital requirements for retained interests in securitizations. North Carolina Banking Institute, 5, 233–264.

    Google Scholar 

  • Merrill, C. B., Nadauld, T. D., & Strahan, P. E. (Forthcoming). Final demand for structured finance securities. Management Science,

  • Merton, R. (1995). Financial innovation and the management and regulation of financial institutions. Journal of Banking & Finance, 19, 461–481.

    Article  Google Scholar 

  • Mora, R., & Reggio, I. (2017). Alternative diff-in-diffs estimators with several pretreatment periods. Econometric Reviews.

    Article  Google Scholar 

  • Roy, A. (1952). Safety first and the holding of assets. Econometrica, 20, 431–449.

    Article  Google Scholar 

Download references


I thank Hester Peirce, Chris Koch, Jeffrey Friedman, Darryl Getter, Thomas Hogan, Ricardo Mora, Lourdes Hilbers at the Federal Reserve’s Freedom of Information Office, Kevin Satterfield at the Office of the Comptroller of the Currency’s Communications Division, Natasha Smith at the Federal Deposit Insurance Corporation’s FOIA/Privacy Act Group, Jerry Ellig, Kristine Johnson, Susan Dudley, Brian Mannix, Sofie Miller, Chad Reese, Ben Klutsey, Youjin Hahn, Hee-Seung Yang, Joe Reid, Tyler Cowen, Thomas Stratmann, participants at the 2015 Western Economic Association Meetings, especially Jin-Ping Lee, participants at the 2017 Southern Economic Association Meetings, especially Weici Yuan, participants at the 2018 Public Choice Conference, especially Julia Norgaard and participants at the research seminar series at Auburn University, especially Jim Barth, for helpful discussions. I also thank an anonymous reviewer and the Editor, Menahem Spiegel, for valuable suggestions.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Stephen Matteo Miller.

Ethics declarations

Conflict of interest

The author declares that they have no conflicts of interest.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Miller, S.M. The recourse rule, regulatory arbitrage, and the financial crisis. J Regul Econ 54, 195–217 (2018).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Difference-in-differences
  • Financial crisis
  • Regulatory capital requirements
  • Securitization
  • Unintended consequences

JEL Classification

  • E02
  • F33
  • G01
  • G18
  • G28