How Do Global Systemically Important Banks Lower Capital Surcharges?

Because these data started on a quarterly basis only after surcharges were introduced, we cannot use them to determine whether surcharges caused these drops or whether these drops already existed before surcharges. However, assuming that indicators subject to window dressing driven by surcharges would necessarily exhibit larger drops in the fourth quarter at GSIBs in the post surcharge period, we use this initial evidence to screen the indicators that might be subject to manipulation.

Data that start before surcharges were introduced—the FR Y-9C data—are not available for every systemic importance indicator. Also, even for the indicators with some FR Y-9C data available, the definitions of the data collected in the FR Y-9C and the FR Y-15 often differ. For these reasons, we perform the initial screening of systemic importance indicators with the FR Y-15 data exist for all indicators but start on a quarterly basis only in the second quarter of 2016.

See also De Genaro et al. (2021), who study the effects of capital surcharges for domestic systemically important banks in Brazil on the interest rate spreads that banks charge on their loans.

The banking literature has studied other bank characteristics that are not included in the GSIB framework, but are related to categories of systemic importance indicators from the framework. For example, (Goldberg and Meehl 2020) examine how measures of organizational, business, and geographic complexity of large U.S. banks have changed over time and Correa and Goldberg (2022) analyze how these measures affect systemic risk.

The GSIB framework and the U.S. GSIB rule were introduced in publications from the Basel Committee on Banking Supervision and the FRB, respectively. Basel Committee on Banking Supervision (2013, 2018) describe the Committee’s GSIB framework and includes an empirical analysis supporting the calibration of surcharges. Basel Committee on Banking Supervision (2014) provides more details about how GSIB scores are calculated in that framework. Federal Register (2015) contains the U.S. GSIB rule, which adopts the Basel Committee’s framework with some changes. Board of Governors of the Federal Reserve System (2015) discusses the rationale and the calibration of surcharges for U.S. banks. See also Passmore and von Hafften (2018), who argue that GSIB surcharges are too small because the framework “underestimates the probability of bank failure, wrongly disregards short-term funding, and excludes too many banks."

For this calculation, the value of the substitutability score is capped at 100.

D’Errico and Roukny (2021) estimate that about three quarters of the notional amount of OTC derivatives in a sample of contracts from firms in the European Union can be eliminated with compression.

Based on a similar reasoning, (Aramonte and Huang 2019) argue that the increase in clearing by central counterparties across various types of OTC derivatives likely contributed to a decrease in notional amounts in recent years. Because central counterparties can observe the positions of their clearing members, they can also identify the notional amounts that can be compressed.

The FRB introduced the FR Y-15 in 2012 to calculate GSIB surcharges, to monitor the systemic risk profile of BHCs subject to enhanced prudential standards under section 165 of the Dodd-Frank Act, to identify other institutions that may present systemic risk, and to analyze the systemic risk implications of proposed mergers and acquisitions.

The assets threshold was $50 billion at the start of the sample period and the Economic Growth, Regulatory Relief, and Consumer Protection Act of 2018 increased this threshold to $100 billion in the second quarter of 2018. Since the start of the sample period, 8 U.S. banks were GSIBs: Bank of America, Bank of New York Mellon, Citigroup, Goldman Sachs, JP Morgan Chase, Morgan Stanley, State Street, and Wells Fargo. The 18 non-GSIB banks in the sample are Ally, American Express, BMO Financial, Capital One, Citizens Bank, Discover, Fifth Third, HSBC, Huntington Bank, Keycorp, M &T Bank, MUFG, Northern Trust, PNC, Regions Financial, Santander, Truist, and US Bank. BB &T acquired SunTrust in December 2019 and changed its name to Truist. In our empirical analysis, we assume BB &T and Truist are the same bank.

The asset thresholds for the collection of STWF data are as follows: assets greater than $700 billion, assets between $250 billion and $700 billion, and assets between $50 billion and $250 billion, where banks in each threshold are subject, respectively, to the aforementioned reporting dates. For a complete description of STWF reporting thresholds, see Schedule G of the FR Y-15 reporting instructions (Board of Governors of the Federal Reserve System 2019).

Method 2 scores from GSIBs start in the fourth quarter of 2016 because this is the first quarter when GSIBs had to report STWF data, which are necessary to calculate method 2 scores. Also, method 2 scores from non-GSIBs start in the second quarter of 2018 because this is the first quarter when all non-GSIBs had to report STWF data.

Non-GSIBs’ average normalized method 2 score rose more than GSIBs’ over this period to some extent because of mergers and acquisitions involving non-GSIBs. These events include BB &T’s acquisition of SunTrust in December 2019 (when BB &T changed its name to Truist), Huntington Bank’s acquisition of TCF Financial Corporation in June 2021, PNC’s acquisition of BBVA USA in October 2021, and M &T Bank’s acquisition of People’s United Financial in April 2022. The Truist deal, together with a four-fold increase in Regions Financial’s STWF systemic importance indicator, contributed substantially to the jump in the non-GSIBs average score in the fourth quarter of 2019, the largest change over the sample period.

The list of banks that met the criteria to be subject to the GSIB surcharge did not change during the sample period. As a result, the GSIB status is fixed at the bank level throughout the sample period and the variable \(GSIB_{i}\) does not need a subscript t.

These contracts do not include credit default swaps (CDS) because the FR Y-9C started covering CDS data a few years after the start of our series.

This comparison is based on mean method 1 scores reported in Table 1 of our paper, Table 2 of Behn et al. (2022), and Table 3 of Garcia et al. (2023).

Munyan (2017) shows evidence that foreign banks window dress their repo amounts at quarter-end but domestic banks do not, and that this difference in behavior can be attributed to reporting requirements.

Differences in reporting requirements may induce different behavior across jurisdictions even if regulatory requirements are the same. For instance, banks may be less inclined to window dress balance sheet items that they must report as quarterly averages and as quarter-end values because these two numbers together may help regulators estimate the impact of window dressing.

Favara et al. (2021); Degryse et al. (2023), and Behn and Schramm (2021) use credit registry data, which allow them to control for borrower and loan characteristics to estimate effects of GSIB surcharges and designation on corporate lending. However, these papers do not analyze year-end variation driven by surcharges.

The FRB may have anticipated banks’ strategic behavior. Thus, the FRB may have set the parameters of method 2 scores and surcharges higher to account for this behavior. In fact, one objective of surcharges was to motivate GSIBs to reduce their systemic footprints, which are measured with GSIB scores (Basel Committee on Banking Supervision 2013; Board of Governors of the Federal Reserve System 2015; Yellen 2015). Therefore, this behavior may have been not only expected, but also an intended outcome. However, the seasonal variation in measures of systemic importance that we document was most likely unwanted because they have lowered surcharges. In addition, surcharges could not account for GSIBs’ strategic behavior for two main reasons. First, the FRB calibrated surcharges in a way that “a GSIB should hold enough capital to lower its probability of failure so that its expected impact is approximately equal to that of a non-GSIB" (Board of Governors of the Federal Reserve System 2015). Thus, the FRB calibrated surcharges to induce GSIBs to hold just enough capital to meet this expected impact, which does not include a buffer to accommodate their strategic behavior. Second, when the FRB used data from GSIBs and non-GSIBs to calibrate surcharges from the second quarter of 1987 to the fourth quarter of 2014, GSIBs’ year-end reactions to surcharges could not be observed yet.

The Dodd-Frank Act reformulated the regulation of OTC derivatives regarding execution, clearing, and reporting. U.S. regulatory agencies were in charge or implementing those changes, which they did over the following years. For instance, even as late as in 2020, the Securities and Exchange Commission approved a rule requiring counterparties to engage in compression “periodically" (Federal Register 2020). Thus, reforms following the principles of the Dodd-Frank Act have been gradually implemented over time, instead of at a single point in time.

For this calculation, the value of the substitutability score is capped at 100.

In fact, when the U.S. rule of GSIB surcharges was proposed, commenters argued to the Federal Reserve Board that an approach with a fixed denominator would provide more certainty to BHCs about the actions that they could take to reduce their GSIB surcharges (Federal Register 2015).