“The FSOC’s authority to designate non-bank financial institutions for ‘heightened prudential supervision’ undermines both financial stability and market discipline by signaling to market participants that the government considers the designated firm ‘too big to fail,’ and that they will be protected from losses if it ever gets into trouble.” House Committee on Financial Services, Comprehensive Summary of the Financial CHOICE Act, April 24, 2017.
In response to the financial crisis of 2007-2009, Congress created the Financial Stability Oversight Council (FSOC), a committee of the chiefs of the U.S. regulatory agencies, chaired by the Treasury Secretary, to monitor and secure the stability of the financial system. Critical to this task is the FSOC’s authority to designate nonbanks as “systemically important financial institutions” (SIFIs). The Dodd-Frank Act also authorizes the Federal Reserve to impose rigorous prudential oversight (e.g. stress tests and capital requirements) over such nonbank financial companies.
While many members of Congress would like to repeal FSOC’s designation authority altogether―this is one of the provisions in the Financial CHOICE Act passed by the House earlier this year―they seem unlikely to succeed. The Trump Administration is taking a different tack: examine discretionary changes in the implementation of the Dodd-Frank Act that would constrain the FSOC. With this in mind, on November 17, the U.S. Treasury issued a report responding to the President’s April 21 Memorandum requesting that the Secretary assess the FSOC’s designation process on the following dimensions: transparency, due process, impact on moral hazard, quantification of costs and benefits, pre-designation opportunities for firms to de-risk, and post-designation opportunities for re-evaluation.
Fortunately, Treasury does not advocate the CHOICE Act approach, which spectacularly misdiagnoses the causes of “too big to fail” (see our earlier post). Instead, it calls on the FSOC to adopt a strategy that prioritizes the regulation of activities or functions—affecting whole sectors of the financial industry—over regulation based on entity or legal form (such as the designation authority). For the most part, we find this sensible, as this focus reduces the scope for regulatory arbitrage that an entities-only approach may foster (see here).
However, we doubt that activities-based regulation alone will be sufficient to limit systemic risk—at least for firms that have the capacity to conceal risk-taking through various off-balance sheet mechanisms. Our overall conclusion is that the Treasury’s approach sets the bar for FSOC designation too high, diminishing its deterrence effect on undesignated nonbanks, much as the 2016 Federal District Court decision that reversed FSOC’s designation of MetLife has already done (see our earlier post).
In the end, a sensible focus on both entities and activities is needed to fulfill one of FSOC’s key objectives—to restore market discipline. Adopting the Treasury’s proposed framework will not meet the goal, set out in the President’s Core Principles for Regulating the U.S. Financial system (see Executive Order 13772), of preventing taxpayer-funded bailouts.
In the remainder of this post, we discuss the FSOC designation of nonbanks in light of the latest Treasury report. We distinguish between asset managers, where the case for activities-based regulation is particularly compelling, and insurers, where entity-based regulation seems likely to be a necessary supplement to address buildups of systemic risk.
Last month, Treasury published the third of four planned reports on implementing President Trump’s Core Principles for financial regulation: A Financial System That Creates Economic Opportunities: Asset Management and Insurance. (For our comments on the first two, see here for the June depositories report and here for the October capital markets report.) Although these two industries are quite different, Treasury’s view on the proper regulatory structure is virtually identical: “…entity-based systemic risk evaluations of [asset managers or their funds/insurance companies] are generally not the best approach for mitigating risks arising from [asset management/the insurance industry].” Rather, “regulators should focus on potential [systemic] risks arising from [asset management/insurance] products and activities, and on implementing regulations that strengthen the [asset management/insurance] industry as a whole” (see pages 153 and 158 here).
It is not hard to see why authorities have come to this conclusion. While institutional size is important, it is of limited value in assessing systemic risk. Very large financial firms that use little leverage, have limited reliance on wholesale funding, avoid organizational complexity, and act transparently (eschewing concealment of risk through off balance sheet exposures or through customized derivatives) likely pose little threat to the financial system. Others that may be smaller can nevertheless boost systemic risk through these means, and through common exposures that cause a collection of small firms to synchronize their behavior. It is their activities, rather than their size, that matters most. (Both the Basel Committee on Banking Supervision and the Treasury Office of Financial Research have identified a range of indicators for assessing systemic importance.)
Activities-based regulation limits the tendency—prevalent especially in the United States—to regulate by legal form rather than economic activity or function. Systemic resilience does not depend on whether risky activities are concentrated in chartered banks or nonbank intermediaries. Regulation by legal form—such as imposing higher capital requirements on banks solely based on their size—may simply shift the systemic risk from one part of the financial system to another without increasing resilience.
In our view, the case for activities (over entity) regulation is greatest in the asset management industry. The key reason is that the beneficial owners of the assets in mutual funds or in segregated accounts are the investors themselves, not the firms that manage them. In addition, mutual funds already face a number of restrictions on their activities. These include limits on leverage, on liquidity mismatch, and on derivatives use. Adjusting these rules—say, to limit the first-mover advantage arising from exiting a mutual fund holding illiquid assets—would be an effective way of reducing systemic risk (see, for example, our earlier post regarding open-end funds and ETFs). Furthermore, these investment vehicles are subject to rules promoting transparency (through frequent disclosure) and safety (for example, by requiring independent custodians to safeguard assets) that, when combined with the absence of off-balance-sheet accounting gimmicks, limit the potential for concealment of risk-taking.
In our view, shifting undiversifiable asset price risk from banks and insurers to mutual fund investors usually makes the financial system safer. A key reason is that the asset managers themselves don’t fail when their funds lose value.
Activities-based regulations also can reduce systemic risk in the insurance industry. Securities lending (SL) is a prime example (see our primer). Like repurchase agreements and asset-backed commercial paper, SL can be a source of short-term wholesale funding. Rules that constrain how a securities lender reinvests the cash collateral it receives can limit the potential for a damaging run. Insurers have a whole array of liabilities that are vulnerable to runs. These include guaranteed investment contracts, deferred annuities and universal life policies. McMenamin et al estimate that roughly one-half of life insurance liabilities are highly or moderately liquid, suggesting substantial short-term funding risks.
Other activities of life insurers that pose systemic risk may be amenable to activities-based regulation as well. For example, as Koijen and Yogo highlight, life insurers lower capital needs through increased reliance on shadow insurance—“affiliated reinsurance (within the same financial group) between an operating company (i.e. a regulated and rated company that sells policies) and a shadow reinsurer (i.e. a less regulated and unrated off-balance-sheet entity).” With shadow insurers facing easier standards for accounting and capital, affiliated reinsurance rose from $11 billion in 2002 to $370 billion in 2013. Koijen and Yogo emphasize that better disclosure—a classic form of activities-based regulation—would facilitate assessment of capital adequacy for shadow insurers. They also emphasize the need for finer disclosure of variable annuities, of interest-rate risk, and of derivatives exposure to allow counterparties and investors to assess insurer risk mismatch and capital adequacy.
However, the agency model of asset managers does not apply to banks or insurers, who (for the most part) issue liabilities that are a claim on the overall value of their assets, not on some segregated pool. Put differently, neither a bank depositor nor an insurance policy holder own a designated slice of a particular loan or security on the institution’s balance sheet. When assets plunge in value, leveraged banks and insurers can and do fail; as we already mentioned, asset managers do not.
We also know from painful experience that insurers have various means to conceal risk-taking. AIG is the poster child for concealment during the crisis, with much of its losses coming as a consequence of low-visibility exposures in over-the-counter derivatives and securities lending. The continued ability of life insurers to boost leverage by shifting risks off balance sheet makes them notably different from mutual funds.
To bring a small amount of data to bear on the issue, it seems that regulation to date has failed to limit the systemic risk arising from the largest insurers. Our preferred measure of systemic risk—the NYU Stern Volatility Lab’s SRISK—quantifies the capital shortfall in the financial system that would be expected in the event of a crisis (proxied by a 40-percent drop in the equity market). Since peaking in mid-2008, aggregate U.S. SRISK has plunged by nearly 80 percent, from roughly $950 billion in September 2008 to less than $200 billion today. However, as the chart below shows, the combined SRISK of the two largest insurers did not decline over this period, so that it now constitutes roughly one-third of U.S. SRISK.
SRISK of MetLife and Prudential (level in billions of dollars, line, left; share of aggregate U.S. SRISK in percent, stacked bar, right; end-year observations), 2001-2017
Perhaps this should not be surprising. Industry-wide activities regulation is difficult to implement precisely for the reason that it is attractive: it focuses on economic function. Thus, while entity regulation may prompt criticism from a small number of firms (those that could be designated as systemic), activities regulation invites resistance from entire industries (or groups of industries) that can form a powerful lobbying force or even capture regulators. Moreover, the limited federal role in U.S. insurance regulation makes it very difficult to impose consistent nationwide activities rules, as more lenient states will tend to attract business from other states, encouraging a race to the bottom.
Consequently, while we expect there to remain very few designated nonbank SIFIs (there is only one today!), we do not share Treasury’s apparent view that designation should be reserved for the “rare instance, such as the historical case of Fannie Mae and Freddie Mac, where it was clear that individual institutions could pose a threat to financial stability, but a primary regulator has not taken or cannot take adequate steps to address the risk.” Put differently, where activities-based regulation fails to limit the buildup of systemic risk, designation should remain a viable option. Not only will this sustain designation’s deterrence value, but it will also allow more intensive oversight of those firms that are reducing the resilience of the system.
Our somewhat lower bar for designation would be consistent with Treasury’s aim to encourage transparency, to promote a rigorous standard for analysis in designation, to provide firms with a clear and detailed rationale for any designation, both before and after designation, and to allow firms to de-risk either to preempt designation in the first place, or to remove the designation that has been applied. In our view, the FSOC de-designations of GE Capital in June 2016 and of AIG in September 2017 are consistent with this approach, provided that FSOC continues to monitor the systemic risks of large nonbanks. Indeed, we hope that FSOC is prepared to re-designate a firm that responds to de-designation by adopting a riskier business model.
Before concluding, we should note that we are profoundly skeptical of Treasury’s decision to assign a central role to cost-benefit analysis in the SIFI designation process. First, the proper cost is not the “cost of any determination or designation on the regulated entity.” It is the social cost. Because intermediation can shift out of SIFIs with few systemic consequences, the private cost surely exceeds the social cost―probably by quite a bit. Second, the benefit from designation—when applied rigorously and transparently so that it serves as a deterrent to non-designated firms—is likely quite large. To see why we say this, consider a case where the cost of a crisis equals 65 percent of one year’s GDP (this is the midpoint of Atkinson et al.’s estimate for the 2007-2009 crisis). At today’s GDP of nearly $20 trillion, the expected cost of a crisis is about $13 trillion. If the designation process lowers the annual probability of a crisis by 0.05 percent, say from 1.05 percent to 1 percent per year (implying a reduction of the frequency from once every 95 years to once every 100 years), then the expected annual benefit would be $6.5 billion per year. Since MetLife’s average annual pre-tax income over the past five years was $4.3 billion, the private cost of designation is probably only a fraction of that, with the social cost even smaller. In other words, by any reckoning, it is hard to see how designation can fail a reasonable cost-benefit test.
Separately, if Treasury were really so eager to cut regulatory costs, the obvious path would be to streamline the existing regulatory framework. This is particularly clear in the case of insurance, where the primary regulators are the 50 states (plus 5 territories and the District of Columbia). This extreme balkanization reduces competition and adds enormously to cost redundancies for firms that wish to offer products across state lines. Yet, in its October report, Treasury “endorses the state-based regulatory model for the U.S. insurance industry and recommends narrowing the scope of federal involvement” (page 105). We have a hard time seeing how this promotes regulatory efficiency or meets a cost-benefit test. It seems blatantly obvious that empowering a government agency—like the Federal Insurance Office—with the mandate to standardize national insurance regulation would both reduce the cost of compliance and improve the resilience of the system. It also would enhance the influence of U.S. policymakers in international standard-setting bodies, in line with Treasury’s stated goal.
So, what’s the bottom line? Treasury argues compellingly for the priority of activities-based regulation over entity-based regulation. At the same time, unlike the Financial CHOICE Act, Treasury accepts the logic that FSOC designation is a useful device for lowering systemic risk and preventing taxpayer-funded bailouts. Unfortunately, in its continued zeal to appease the financial industry, and to protect political sacred cows (like state-based insurance regulation), President Trump’s Treasury seems intent on setting the bar on FSOC SIFI designation too high to secure these goals. The resulting risk of regulatory arbitrage—and rising systemic vulnerability—is all too clear.
Acknowledgment: The authors are grateful for very helpful discussions with their friend and NYU Stern colleague, Professor Ralph Koijen.