Each of the banks in the United States designated by the Basel Committee for Bank Supervision (BCBS as a Global Systemically Important Bank (GSIB) is required to maintain a capital level that is higher than that required of other banks by an amount equal to its “GSIB surcharge.” There are two common misconceptions about GSIB surcharges that this post rebuts:
- The GSIB surcharges are calibrated; that is, they are based on data.
- The GSIB surcharges already reflect the effects of subsequent GSIB regulations because those changes were anticipated when the surcharges were established in 2015.
The surcharges are not calibrated
While the GSIB surcharges are based in part on data, they are mostly based simply on judgement and negotiation among international regulators. Calling the GSIB surcharges “calibrated” is like calling pills with exactly 6.75 grams of medicine and instructions to the doctor “size the patient up and prescribe as many pills as seem right” carefully dosed.
Specifically, as described in the whitepaper the Fed released when promulgating the GSIB surcharge rule, the Fed bases the surcharges on an estimate of the capital needed to equalize the expected social cost of a GSIB failure to the expected social cost of a failure of a non-GSIB reference bank.[1]
In particular, the surcharge is calculated as
Where is the surcharge, is a characteristic of the GSIB (such as total exposures under the size category), is the same characteristic of a reference non-GSIB, and is a weight multiplied by the bank characteristic to determine the bank’s social loss given default. In the U.S., the GSIB capital surcharge depends on 10 bank attributes. The sum of these 10 bank characteristics times the corresponding weight is called the systemic loss given default score, or SLGD.
Of the formula, only the number in green, 2.18, is calibrated. That number is estimated using the return on risk-weighted assets of the 50 largest banks from 1987 through 2014. The estimate allows the Fed to establish the amount of capital needed to withstand losses for a given probability.
The rest of the formula—the systemic loss given default—is not calibrated. A recent paper by Wayne Passmore and Alexander von Hafften (2017) states “The Basel G-SIB score framework is best interpreted as a judgment by bank supervisors about which balance sheet measures are correlated with systemic importance.” [2] In other words, it is made up.
Specifically, neither the Basel Committee nor the Fed conducted any analysis to determine which bank characteristics should be included in the formula, how the characteristics should be weighted, or whether the relationship between the characteristics and systemic loss given default is linear.
These judgments are not obvious. For example, the U.S. banking agencies chose to depart from the international standard because they concluded it was necessary to include a set of additional characteristics (short-term wholesale funding) and exclude another set (the substitutability category) while the rest of the countries in the Basel Committee concluded the opposite.
Nor are the choices inconsequential. The U.S. version produces GSIB surcharges that are about twice those derived under the Basel version.
Even the authors of the Fed whitepaper acknowledge the choice to model loss given default as linearly related to the characteristics is arbitrary or even wrong. They state, “…there is reason to believe that the function relating the scores to systemic LGD increases at an increasing rate…”[3] As we show in a research note assessing the Fed’s whitepaper, the GSIB surcharges would be substantially different if the relationship were nonlinear, as the Fed evidently thinks, or even if there is simply an excluded constant term.[4]
Moreover, it is not impossible to estimate the relationship between loss given default and bank characteristics. Passmore and Hafften (2017), for example, use a market-based measure of a bank’s systemic importance to estimate loss given default and derive much different estimates for GSIB surcharges than from those derived using the Fed’s method. In addition, in our research we have used a similar empirical approach to evaluate the cap on the substitutability category used in the calculation of the Method 1 GSIB surcharge. Our results also indicate that the economic magnitude of the interconnectedness category has an incorrect sign, and the size category is not statistically important in explaining systemic risk, after controlling for the remaining categories.
In sum, while the Fed did calibrate one number in the formula used to determine GSIB surcharges to data, the rest of the formula is determined by judgement alone. Such a procedure left the Fed plenty of freedom to make choices that deliver GSIB surcharges that meet prior views of what is “correct” rather than surcharges determined by the data.
The surcharges do not account for subsequent regulatory changes
As we note in our earlier research note, and frequently since (see here, here, here, and here), the surcharges also do not account for the many changes in regulation and supervisory practice since their design that have made GSIB failures less likely and less costly. Such changes include living wills (and, in particular, the adoption of the SPOE resolution strategy by all eight U.S. GSIBs), capital (RCAP/RCEN) and liquidity (RLAP/RCEN) pre-positioning requirements, TLAC, clean holding company requirements, and the elimination of cross-default clauses in financial contracts. In recent work with Sean Campbell of the Financial Services Forum, we estimated that the additional requirements for loss absorbing long-term debt and liquidity alone should lead to a 1 1/3 percentage point reduction in the GSIB surcharges on average.
A reply we increasingly hear when we make this point is that all these changes were anticipated when the GSIB surcharges were designed. While we agree that the changes were anticipated, just because it is logically possible that they were incorporated into the GSIB surcharge methodology hardly proves they actually were incorporated.
Neither the Federal Reserve’s December 2014 GSIB surcharge proposal nor its August 2015 final rule (including its accompanying white paper) describes whether or how these changes were incorporated into the surcharge methodology. Indeed, while the implementation documents note that the GSIB surcharges were part of the larger set of post crisis reforms to capital and liquidity regulations, they don’t mention at all the substantial steps that have been made to reduce the social costs of a large bank failure.
Moreover, the strongest indication that the changes were notincorporated is the one piece of calibration the Fed performed when designing the surcharges. As we noted above, the Fed’s estimate of the amount of capital needed to prevent failure is based on banks’ performance between 1987 and 2014, a period during which none these changes had come into effect.
Conclusion
Getting bank regulation right is serious business. If capital requirements are too low, then society is taking an unacceptably high chance of another financial crisis with the attendant astronomical costs. But if bank capital requirements are too high, then the cost of borrowing is higher, banks lend less and economic output and welfare is lower every daythan it safely could be.
None of the shortcomings of the GSIB surcharges we discuss here indicate the surcharges are too high or too low, just that they are almost surely wrong.
Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of the Bank Policy Institute or its membership, and are not intended to be, and should not be construed as, legal advice of any kind.
[1]Calibrating the GSIB Surcharge, Board of Governors of the Federal Reserve System, July 20, 2015. https://www.federalreserve.gov/aboutthefed/boardmeetings/gsib-methodology-paper-20150720.pdf
[2]Passmore, Wayne, and Alexander H. von Hafften (2017). “Are Basel’s Capital Surcharges for Global Systemically Important Banks Too Small?,” Finance and Economics Discussion Series 2017-021. Washington: Board of Governors of the Federal Reserve System, https://www.federalreserve.gov/econresdata/feds/2017/files/2017021pap.pdf
[3]“Calibrating the GSIB Surcharge,” p. 4.
[4]They could be higher or lower. SeeCovas, Francisco, and Bill Nelson (2016), Overview and Assessment of the Methodology Used to Calibrate the U.S. GSIB Capital Surcharge, The Clearing House, May 2016. https://www.theclearinghouse.org/-/media/action-line/documents/volume-vii/20160510_tch_research_note_gsib_surcharge.pdf
