During the past few years, regulators have imposed a new capital standard loosely based on the traditional leverage ratio that compares a bank’s equity capital to its total assets. Despite the fact that this measure fails to fully account for the risk of a bank’s assets and creates perverse incentives, regulators implemented this standard based on a view that it is simple and transparent and serves as a backstop to other risk-based regulatory capital requirements. Recently, the Federal Reserve’s Vice Chairman for Supervision, Randal Quarles, gave a speech in which he indicated support for removing leverage buffer requirements from the stress test. In light of this announcement, it makes sense to revisit the rationale for leverage-based capital requirements. In this post, we examine the underlying premise for the Supplementary Leverage Ratio and discuss recent evidence documenting how the perverse incentives it creates are affecting banks and their customers.
The supplementary leverage ratio, or SLR, was adopted by the Basel Committee on Banking Supervision as part of the Basel III package of post-crisis banking reforms and is described by the committee as a “simple, transparent, non-risk based leverage ratio to act as a credible supplementary measure to risk-based capital requirements.” The term “leverage ratio” indeed brings to mind a relatively simple and transparent metric that compares the ratio of a bank’s capital to its total assets. The SLR, however, does not compare capital to total assets. Rather, the SLR compares capital to “total leverage exposure,” or TLE. So, what exactly is TLE and how simple and transparent is it?
Put as simply as possible (which isn’t that simple), total leverage exposure is a regulator-defined quantity that measures a bank’s total assets and a variety of off-balance sheet items including derivatives, repurchase agreements, and a host of other items including loan commitments, standby letters of credit and trade finance exposures. Each of these items is included in TLE through a complicated set of computations that depends on specific characteristics of the item, such as its maturity, and a regulator-defined set of weights that make some high-level and rough-and-ready distinctions about the risk of different exposures. These additional off-balance sheet items are a significant component of TLE that varies across banks and across time. Figure 1 shows the breakdown between on- and off-balance sheet items in TLE for each Forum member and in the aggregate. As shown in the figure below, the importance of these off-balance sheet items varies considerably as it accounts for between 17 and 56 percent of total TLE across Forum members.
Figure 2 further explores the variability of TLE by comparing the variability of the ratio of TLE to total assets to the variability of the ratio of risk-weighted assets (RWA) to total assets for Forum members. Proponents of leverage requirements claim that leverage measures are simpler and hence less variable than more complex risk-weighted asset measures. The results presented in Figure 2 show that the variability of TLE (1.8% per quarter) has been roughly as large as the variability of RWA (1.9% per quarter), which suggests that TLE is no simpler than RWA as a capital measure. Another way of assessing the complexity of TLE is to consider that the U.S. federal banking agencies described the TLE calculation in their final rule with more than 2,500 words, filling 10 pages. It should be fair to say that any calculation that requires 10 pages of guidance and explanation can’t reasonably be characterized as “simple.” Accordingly, the complexity of TLE broadly mirrors the complexity of RWA, which it was intended to simplify.
So, if regulators desire a simple and transparent leverage metric, why is the SLR so complicated? The answer is actually quite simple. Large banks assume a variety of risks that cannot be adequately characterized by their accounting value on the balance sheet. As a result, the SLR is actually a hybrid measure that arbitrarily applies a form of risk weights to some exposures and not others. Accordingly, the SLR does not treat all exposures equally. The net result is a complicated capital metric that does not sensibly capture risk differentials and creates perverse incentives.
While the construction of the SLR is complicated, the basic incentive problem it creates is both simple and well understood. When capital requirements do not increase with risk, banks have a clear incentive to reduce their footprint in lower-risk activities to engage in higher risk and higher return activities. Recent research from the Commodities Futures Trading Commission (CFTC) demonstrates that the new leverage requirements are reducing U.S. banks’ willingness to engage in low-risk and low-margin derivative transactions on behalf of their customers. These client clearing services are important to end-users – such as pension funds and university endowments- who don’t have direct access to derivative markets but need derivatives to manage important financial risks.
Table 1, which is taken directly from the CFTC’s research paper, shows the market share for U.S. and EU derivative market participants before and after the imposition of the SLR requirement in the United States. The EU has not been subject to the same leverage requirements as U.S. banks and so they serve as a useful control group. As the table shows in row 1, the market share for the U.S. declined by more than 10 percentage points, while the EU market share has increased by roughly the same amount. The majority of this decline came from the market share of U.S. bank customers who rely on a U.S. bank to provide access to derivative markets while the market share of EU bank customers grew by a corresponding amount, shown in row 5.
These results illustrate two important points about leverage requirements. First, leverage requirements are having a direct impact on bank behavior. The top-line results indicating a 10-percentage point decline in U.S. market share after the imposition of leverage requirements in the U.S. clearly demonstrates that these requirements affect banks’ willingness to engage in lower-risk client-facilitation activities. Second, bank clients are directly impacted by leverage requirements. As shown in row 5 of the table, U.S. bank customers relying on U.S. banks to provide derivative market access saw the biggest reduction in market share. As a result, bank clients’ access to derivatives has been restricted. While some bank clients may seek access to derivatives through EU banks or non-banks, this process is likely to be more complicated, more costly and to result in reduced choices for clients.
These changes in derivatives markets are not going unnoticed. Specifically, CFTC Commissioner Rostin Behnam recently noted in a speech that “the SLR is contributing to higher…concentration levels and reduced access to clearing.” Also, and importantly, the SLR’s impact on derivative markets should not be viewed as an isolated event. Rather, a growing body of research shows that the perverse incentives created by leverage requirements are distorting bank behavior in a variety of contexts. As an example, Professor Jeremy Stein, a former Federal Reserve governor now Chairman of the Department of Economics at Harvard University, has shown that banks whose capital levels are more tightly restricted by leverage requirements respond by assuming more risk. Additionally, a recent research paper and accompanying blog post by the Federal Reserve Bank of New York also shows that bank behavior is impacted by the SLR. Further, Professor Darrell Duffie of Stanford University has found that the SLR has negatively impacted the market for repurchase agreements. Accordingly, the evidence that leverage requirements impact bank behavior, bank clients, and markets is widespread.
Against all of this evidence on the negative effects of leverage requirements, one is brought back to the regulators’ fundamental rationale for its necessity. Regulators have maintained that leverage requirements are simple and transparent and serve as a backstop to prevent gaming behavior by banks. These claims, however, are hard to square with the facts. We have shown that the SLR and TLE, in particular, is complex and not transparent. The computation of TLE represents an arbitrary hybrid approach in which some assets and exposures are carried at their accounting value and others are valued using a complex regulatory computation. The concerns raised about gaming and the need to use the SLR as a backstop are also puzzling. In particular, large banks are already subject to both stress testing requirements and standardized and regulator-defined risk weights that generally do not depend on a bank’s own risk assessment. Indeed, the recently announced Basel IV reforms will further reduce any reliance of risk-based capital requirements on internal bank risk assessments. Accordingly, it is not clear why regulators need a leverage requirement as a backstop to risk-based capital and stress testing requirements that are either largely or entirely free from bank internal risk assessments.
During the past several years, regulators have substantially strengthened risk-sensitive capital requirements through stress testing and related reforms. At the same time, Forum members have increased their resiliency by increasing their collective capital levels by nearly $350 billion in Tier 1 capital. The SLR represents an unnecessary capital requirement that is neither simple nor transparent and has been shown to have negative consequences in a number of areas. The recent remarks made by Vice Chairman Quarles on the leverage ratio are well reasoned and helpful. Further consideration of the necessity of the SLR could also help improve regulatory efficiency without sacrificing resiliency.