Extended Shareholder Liability for Systemically Important Financial Institutions

69 Am. U. L. Rev. 967 (2020).

* Yale Law School.

** Faculty of Law, University of Oxford; European Corporate Governance Institute.

*** Sam Harris Professor of Corporate Law, Corporate Finance, and Securities Law, Yale Law School; European Corporate Governance Institute.

Regulators generally have tried to address the problems posed by the excessive risk-taking of Systemically Important Financial Institutions (SIFIs) by placing restrictions on the activities in which SIFIs engage. However, the complexity of these institutions makes such attempts necessarily imperfect. This Article proposes to address the problem at its very source, which is the incentives that SIFI owners have to push for excessive risk-taking by managers. Building on the traditional rule of “double liability,” we propose to modify the current (general) rule limiting the liability of SIFI shareholders to the amount of their initial investments in such companies. We propose replacing the extant limited liability regime with a new system that imposes additional liability over and above what SIFI shareholders already have invested in a preset amount that varies with a SIFI’s centrality in the financial network. Our liability regime has a number of advantages. First, by increasing shareholder exposure to downside risk, it discourages excessive risk-taking. At the same time, by placing a clearly defined ceiling on shareholders’ total liability exposure, it will not obliterate shareholders’ incentives to invest in the ﬁrst place. Second, the liability to which shareholders are exposed is carefully tailored to the level of systemic risk that their institution creates. Thus, our rule induces shareholders to account for the negative externality SIFIs can impose without unduly stifling such financial institutions’ role within the financial system and in the wider economy. Third, as the amount of liability is clearly defined ex ante using the rigorous tools of network theory, our rule minimizes the influence of interest groups and the impact of idiosyncratic government decisions. Last, as markets know in advance the amount of liability to which shareholders are exposed, our rule favors the creation of a vibrant insurance and derivative market so that the risk of SIFIs defaults can be allocated to those who can better bear it.

Introduction

The default of a systemically important financial institution (SIFI) imposes significant negative externalities.¹ Such a default inevitably propagates through the financial system with dramatic and fatal consequences for even the most prudently run businesses. It is understood that the risk of a national or global economic meltdown attributable to the failure of a systemically important financial institution justifies aggressive regulation as well as significant departure from ordinary and customary corporate governance norms for SIFIs.² Perhaps the most telling manifestation of the public policy implications of being considered systemically important is the entrenched policy of governments around the world to bail out financially distressed SIFIs, despite the massive costs and perverse incentives associated with these bailouts. No one has devised a functional plan to enable governments credibly to commit to refrain from carrying out such bailouts.³

It is well known that the near certainty that SIFIs will be bailed out creates acute moral hazard.⁴ Specifically, SIFI creditors, shareholders, directors, and managers, knowing that their firm will be bailed out if too many risky investments turn out badly, have incentives to take excessive risk⁵ and refrain from engaging in monitoring.⁶ To inhibit such excessive risk-taking, policymakers consistently pledge that there will be “no more tax-funded bailouts.”⁷ But such pledges are unconvincing because, unlike Ulysses, regulators and politicians cannot credibly tie themselves to the mast.⁸ Bailouts are therefore a fact of life.⁹

Regulators generally have tried to mitigate SIFIs’ proclivity to engage in excessive risk-taking by imposing strict capital requirements and placing restrictions on the activities in which SIFIs can engage.¹⁰ Yet, the high costs and sheer complexity of these regulations make such attempts necessarily imperfect.¹¹ This Article proposes an important new addition to these standard regulatory approaches by addressing the problem at its very source, which is the market-based incentives that SIFI owners have to encourage and incentivize excessive risk-taking by managers. Our proposal is to modify the current practice of limiting the liability of SIFI shareholders to the amount of their investment in such companies.

We propose replacing the extant limited liability regime with a new system that imposes a fixed and stable amount of potential additional liability, over and above what SIFI shareholders already have invested, in case the SIFI is resolved or bailed out. By increasing shareholder exposure to downside risk, our proposal strongly discourages excessive risk-taking. At the same time, by placing a fixed ceiling on shareholders’ total liability exposure, it will not eradicate shareholders’ incentives to invest in the first place.¹² The advantage of our approach is that, by realigning shareholders’ incentives to reduce SIFI risk-taking to more societally acceptable levels, it provides a measured and proportionate complement to existing, highly imperfect regulatory initiatives to reduce excessive risk-taking. Thus, our approach is unique in that it will create an operating environment in which bankers with properly aligned incentives will voluntarily engage in societally beneficial self-discipline and avoid excessive risk-taking in the first place.

The legal regime of extended liability that we propose for SIFIs is different from both unlimited and “classic” limited liability systems. Under the former, which is traditionally the rule for general partners in partnerships, partners are liable for the partnership’s unsatisfied debts, whether deriving from contractual obligations or torts, with no cap on the amount of the liability.¹³ In the classic limited liability system, which has been the rule for corporations in the last couple of centuries, shareholders are not liable for any of the unpaid corporate debts.¹⁴ Hence, their loss is limited to their investment in the company if it goes bankrupt.¹⁵ Extended liability is located in between these two extremes because shareholders stand to lose more than their investment in the company, but their downside exposure is still capped at a preset amount.

The main claim of this Article is that, for SIFIs, a carefully crafted extended liability regime is superior to both unlimited and traditional limited liability. To be clear, we do not argue that an extended liability rule can induce shareholders to internalize all the possible externalities caused by the distress of a SIFI. As we argue in Section II.B, this result is both impossible to achieve under any liability rule and undesirable. Instead, the more modest goal of our rule is neutralizing the moral hazard created by the expectation of SIFI bailouts.

Extended liability has a long and illustrious tradition in the United States. For roughly three quarters of a century, shareholders of banks faced a rule of “double liability,” which made them liable for more than they had invested in the bank.¹⁶ Scholars, including one of us with Geoffrey Miller, have argued that it is important to investigate “whether double liability—or some variant on the idea—offers promise for coping with contemporary problems in the banking industry.”¹⁷ We return to this question showing that the current features of the financial system (namely, the presence of capital ratios for financial institutions, the dominance of institutional share ownership, and the availability of well-developed insurance and derivatives markets) create the perfect conditions for implementing a special form of extended liability. Against this background, the argument advanced in this Article is then articulated in four steps. First, limited liability is inadequate for SIFIs. Second, due to the specific features of SIFIs, it is possible to avoid the theoretical and practical shortcomings that are usually associated with extending shareholder liability. Third, unlimited shareholder liability for SIFIs would be inefficient. Fourth, a carefully designed, “network-sensitive” rule of extended liability would outperform both limited and unlimited liability.

The structure of the Article is as follows. Part I explains why it is important to ensure that systemically important financial institutions do not engage in excessive risk-taking. Moreover, this part sketches the proposals that have been advanced by the literature to increase shareholders’ liability, and hence reduce their risk propensity. Part II explains why the traditional limited liability rule and a regime of unlimited liability for shareholders are both undesirable for SIFIs. Part III is the core of the Article and describes in detail the liability rule that we propose. Part IV tries to anticipate the effect that our liability rule would have on equity markets and SIFI ownership. Part V shows how traditional objections to unlimited shareholder liability either do not apply to our proposed rule or can be easily addressed and outlines how shareholders of SIFIs transitioning to the proposed regime could be compensated for the losses it imposes on them. Part VI briefly concludes by summarizing the main findings of the Article.

I. The Challenge of Taming SIFIs

To begin our analysis, we start by explaining why policymakers still face a serious challenge in providing for an effective regulatory framework for tackling SIFIs and their tendency to engage in excessive risk-taking. Next, we briefly review previous proposals to dispense with shareholder limited liability. We cover both general recommendations in that direction, namely Henry Hansmann and Reinier Kraakman’s proposal to make shareholders liable towards tort creditors,¹⁸ and the more recent, post-crisis suggestions to extend shareholder liability with respect to (some) financial institutions, explaining why each of them would be inadequate as a tool to prevent SIFIs’ excessive risk-taking. We also discuss the double shareholder liability regime which applied to banks for a considerable period of U.S. history.

A. SIFIs and Interconnections Within the Financial Network

The recent financial crisis has been a stunning reminder of the fragility of the financial system. As its various parts are increasingly intertwined, large shocks can quickly propagate throughout the financial system and to the real economy with catastrophic consequences.¹⁹ National and supranational policymakers reacted by tightening up the regulatory framework with the aim of minimizing the risk of future financial crises.²⁰ The main targets of these regulations have been so-called systemically important financial institutions (SIFIs), that is, banks and other financial institutions—the failure of which, due to their size and interconnectedness, can bring down the entire financial system.²¹

Describing the new regulatory framework lies outside the scope of this Article,²² but two facts are worth mentioning. First, policymakers have significantly expanded the portfolio of regulatory tools to preserve financial stability. Before the crisis, most of the regulations aimed at ensuring the solvency of the individual financial institutions without paying attention to the interconnections among them.²³ Departing from this “microprudential” approach, policymakers have now introduced a new family of “macroprudential” policies that attempt to protect the financial system as a whole.²⁴ These new policies have significantly complicated the regulatory landscape²⁵ and yet have not eliminated the risk of a systemic crisis. Regardless of how carefully they are devised, ex ante regulations—be they micro or macro—cannot eliminate systemic risk: “[f]ailure is a fact of economic life.”²⁶ Moreover, regulators suffer from a chronic lack of information that impairs their ability to produce effective policies.²⁷ For instance, leading financial economists have suggested that the capital requirements for SIFIs should be greatly increased.²⁸ Capital reserves constitute a fundamental buffer that allow banks to be more resilient during times of stress, thereby increasing the stability of the financial system.²⁹ The higher the capital reserves, the greater the losses banks are able to absorb. This is especially true for tier one capital, which includes only capital elements of the highest quality.³⁰ But identifying the optimal capital requirement for each SIFI requires a level of information that regulators simply cannot have. Hence, capital requirements imposed by regulators are likely to be either too lax, in which case they fail to ensure the stability of the financial system, or too strict, in which case they impose unnecessary constraints on SIFIs’ activities.³¹ Our liability rule is an attempt to bypass these informational problems by improving the incentives of shareholders on the one hand and by enlisting markets in the monitoring of SIFIs’ solvency on the other.

Second, regulators have attempted to mitigate the moral hazard problem created by bailouts. The financial crisis has reminded SIFIs once more that they are just too big to fail, and hence regulators are forced to intervene if a SIFI is in distress.³² In this vein, anticipating that governments will bail them out in case of need, SIFIs have incentives to engage in excessive risk-taking, while their creditors have weaker incentives to monitor them.³³ To prevent this moral hazard problem, regulators have attempted to tie their own hands by introducing mechanisms to prevent future bailouts. The most important of these mechanisms are the Orderly Liquidation provisions in Title II of the Dodd-Frank Act,³⁴ which aim to facilitate the resolution of large, complex financial institutions by providing for a new bankruptcy procedure to be used for bank holding companies and their subsidiaries as an alternative to the Bankruptcy Code.³⁵ A failing institution is placed in receivership under the control of the Federal Deposit Insurance Corporation (FDIC).³⁶ The FDIC, under the Orderly Liquidation Authority (OLA), has the power to act swiftly in order to find a new owner for the “good” parts of the failing institution, with access to government money to finance the operations of a “bridge bank” until the buyer for the good parts has been found.³⁷ That should ensure continuity of operations and therefore avoid the negative effects on other financial institutions of a SIFI bankruptcy.

Yet, serious doubts have been raised as to whether the OLA would be sufficient to resolve a major SIFI, such as Bank of America, which is not only orders of magnitude larger than any of the commercial banks the FDIC usually deals with, but also active across different businesses and jurisdictions.³⁸ Even more doubtful is whether the OLA would work in the event of a systemic crisis in which the survival of not one, but many, if not all, of the existing SIFIs was at stake.³⁹ In such a case, it will be virtually impossible to find a buyer for the good parts of the failing SIFIs. Moreover, SIFIs’ operating companies may also face illiquidity, if not insolvency problems, due to their credit or balance sheet interconnections, which may make a government recapitalization the only viable solution.⁴⁰

Ultimately, committing not to bail out SIFIs is impossible. The Darwinian proclivity for survival that characterizes political behavior in democracies leads politicians and policymakers to offer bailouts no matter how tough the ex ante rules on using taxpayers’ money to prop up banks. The immediate political benefits of a bailout, namely, the avoidance of the doomsday scenario of a financial and economic meltdown, are bound to appear superior to navigating the political consequences of such an outcome.⁴¹ As noted by the Financial Crisis Inquiry Commission, “if you bail out AIG and you’re wrong, you will have wasted taxpayer money and provoked public outrage. If you don’t bail out AIG and you’re wrong, the whole financial system collapses.”⁴² In sum, “[b]ailouts are an inevitable feature of modern economies, in which the interconnectedness of firms means that the entire economy bears the risk of an individual firm’s failure.”⁴³ Therefore, realistically, policymakers should attempt to minimize the moral hazard created by bailouts instead of hoping to convince the markets that bailouts will not happen in the future. This is exactly what our extended liability rule attempts to do.

The fact that firms are tightly interconnected has another fundamental implication that economists have fully appreciated, although legal scholars have sometimes overlooked: the structure of the financial sector, and in particular the pattern of connections among financial institutions, has a fundamental impact on the stability of the financial system.⁴⁴ In order to study these inter-bank connections and the network they form, scholars of different fields have relied on network theory.⁴⁵ Within the framework of network theory, the building blocks of a network are its nodes and the connections among them. Thus, if one models the financial sector as a network, the banks and the other financial intermediaries represent the nodes, while the financial flows among them represent the connections.⁴⁶ One important finding of this strand of literature is that, besides size, the position and the level of a financial institution’s interconnections also determine its ability to impose negative externalities on the financial sector and the economy in general.⁴⁷ In fact, policymakers and economists alike have acknowledged that some institutions are too interconnected to fail.⁴⁸ For instance, the Federal Reserve rescued Bear Stearns, a relatively small financial institution due to its interconnections with other key financial actors.⁴⁹

Network scholars have devised various measures to gauge the relevance of an individual financial institution for the financial network as a whole, that is, its “centrality.” One of the most commonly used measures of centrality is known as the DebtRank, which measures the dollar value of the harm imposed on the financial system by the distress of a given financial institution.⁵⁰ Notably, the DebtRank can be normalized between zero and one, so that it captures the fraction of the financial network (in value) that would be affected by the disruption of a node. In a companion paper, two of us have argued that insights from network theory can and should be used to improve on the effectiveness of financial regulation.⁵¹ Similarly, we suggest below that measures of centrality should be a key component in the development of an extended liability regime for SIFI shareholders. Centrality allows policymakers to tailor shareholder liability based on the harm that an individual SIFI would impose on the financial system if it were to collapse.

B. Other Proposals to Extend Shareholders’ Liability

Scholars have long acknowledged that firms are able to impose negative externalities on society due to the limited liability of their shareholders.⁵² To address this concern, proposals to extend the liability of shareholders have cyclically resurfaced. This Section provides a short overview of some of the most influential of these proposals.

In the modern debate about the merits of limited liability, the first scholars to question it as a general rule for corporations were Henry Hansmann and Reinier Kraakman in an article that spurred great controversy.⁵³ They noted that shareholders are shielded from liability for two kinds of corporate debts: those having a source in contracts and those deriving from torts and argued that limited liability is appropriate for the former but not the latter.⁵⁴ In fact, they suggest that contract creditors decide to enter in a relationship with the firm and can therefore assess its creditworthiness and reliability.⁵⁵ Moreover, they can contract on the appropriate compensation for bearing the risk of default and can include clauses to protect themselves from opportunistic behavior on the part of the firm.⁵⁶ Therefore, contractual creditors would not need the additional protection deriving from unlimited shareholders’ liability. Instead, tort creditors have not consented to enter into a relationship with the firm and cannot contract ex ante about risk allocation. As a consequence, shareholders can impose consequences on creditors that are not internalized via private contracting. Building on these premises, Hansmann and Kraakman concluded that, as a matter of general corporate law, shareholders should face unlimited liability for tort losses.⁵⁷

Their proposal had the great merit of challenging the standard assumption that limited liability is an inherent feature of corporations and aimed to restore the full force of tort law to tackle externalities. It was not meant to tackle the specific problems arising from too-big-to-fail financial institutions. But it is worth noting here that the kind of externalities that tort law addresses are different from the ones deriving from the failure of a large financial institution. As we explain in Section II.B, most of SIFIs’ externalities qualify as pure economic losses, giving no rise to a tort claim. Hence, Hansmann and Kraakman’s proposal for unlimited shareholder liability in torts would have little, if any, effect on shareholders’ incentives ex ante.

Fast forward twenty years, and the financial crisis prompted scholars to advance new proposals to extend shareholder liability in order to curb SIFIs’ propensity to engage in risky activities. Peter Conti-Brown has advocated for the introduction of an “elective” unlimited liability for the shareholders of systemically important institutions.⁵⁸ The gist of his idea is that shareholders would be allowed to collectively choose whether they want to face unlimited liability for the total cost of their SIFI’s bailout or hold higher capital reserves (15 to 20 percent).⁵⁹ The problems with this proposal are two-fold. First, under such a regime, shareholders cannot figure out in advance the value of the expected liability because the “total cost of bailout” cannot be known ex ante. Hence, they are unable to set their level of precautions accordingly. Second, Conti-Brown confronts shareholders with a draconian choice: high capital requirements arbitrarily set by a regulator with limited information, or a regime of unlimited liability based on the unpredictable and idiosyncratic determinations of policymakers in times of economic tensions.⁶⁰

Steven Schwarcz’s proposal on extended liability focuses instead on “shadow banks” only, arguing that their “investor-managers” should face a liability that is a multiple of their investment in the firm.⁶¹ His proposal, however, does not engage with the questions of how to calculate the “original investment” in the firm or what the multiple should be. Second, his proposal is limited to shadow banking, which Schwarcz defines loosely as including “special purpose entities (SPEs), . . . as well as finance companies, hedge funds, money market mutual funds, nonbank government sponsored enterprises, securities lenders, and investment banks.”⁶² While shadow banks may be big taken together, they represent a relatively marginal subset of SIFIs. In fact, since SIFI designation rules entered into force, only one such entity has been designated as a SIFI, GE Capital Holdings, but its designation was rescinded three years later.⁶³ Finally, the proposal only extends the liability of “investor managers,” who are defined as equity investors who also have “significant power to control the firm’s actions.”⁶⁴ However, this limitation raises the issue of defining who has significant power to control the firm’s actions, which might not always be straightforward. With a few notable exceptions, such as hedge fund management firms and at many institutional investors, nobody within the organization will have significant power to control the SIFI’s actions.

Most recently, similar to Schwarcz but with reference to all SIFIs, Charles Goodhart and Rosa Lastra have proposed an extended liability regime for all “insiders who have both the information and capacity to influence corporate decision-making.”⁶⁵ In that category, they include, in addition to top managers, large shareholders, i.e., those with a stake higher than five percent.⁶⁶ Shareholders with a stake between two and five percent should be able to choose whether to be treated as “outsiders” or “insiders,” and in the former case, shareholders must commit to not exercise their voting rights. Large shareholders should “have double liability, i.e. for an additional twice par value of their shares,”⁶⁷ while insider shareholders (those with between two and five percent of voting rights) would “be liable to pay in an additional par value of their shares.”⁶⁸

Leaving aside the inadequacy of par value as a multiplier for extended liability in today’s environment,⁶⁹ the distinction between large and small shareholders under this proposal would seem to be a troublesome solution to a relatively minor problem. The solution would be troublesome because it would create a disincentive to invest more than the relevant thresholds, which may be either self-defeating (if no one crosses the threshold) or such that a disproportionate voting power is granted to larger shareholders (if many of the insider shareholders opt for passivity). The problem it solves is to avoid “the unfairness, and political unpopularity, of bankrupting innocent, uninformed and powerless outside shareholders.”⁷⁰ This outcome would be highly unlikely in today’s equity markets, where most individuals do not invest directly in equity shares. And those who do would be unlikely to face financial ruin so long as liability is extended but not unlimited. In addition, very small shareholders would be unlikely to be sued at all for liability because it would not be cost-effective anyway.

The closest antecedent to our liability rule proposal is in legislation: for roughly three quarters of a century, shareholders of American banks were subject to double liability.⁷¹ More precisely, they were liable for an amount in excess of their investment up to the par value of their stock.⁷² Empirical studies confirmed that double liability was effective in reducing banks’ risk-taking.⁷³ Reviving this rule “as it was” would be problematic. On the one hand, par value has become a meaningless measure of the value of a stock.⁷⁴ On the other hand, because of its one-size-fits all approach, the expected liability faced by the shareholders under a hypothetical double liability rule for SIFIs would not vary with the size of the negative externalities that their SIFI can create and would thus imperfectly mold shareholder incentives.

II. Why Extended Liability?

In this Part, we discuss why both traditional limited liability and various versions of an unlimited liability regime for SIFI shareholders are not, or would not be, optimal.

A. Why Limited Liability Is Inappropriate for SIFIs

Limited liability has been the standard for U.S. corporations for over 150 years.⁷⁵ If only for that reason, a strong justification is needed for any deviation. This Section provides support for the proposition that limited liability should not apply to SIFIs. Let us recall that a SIFI default can have dramatic consequences on the financial system and ultimately on the entire economy. The SIFIs’ shareholders will internalize only a small fraction of such losses. Moreover, in order to prevent such a fallout, and no matter what the statutes say about never again using taxpayers’ money to save financial institutions, the government will be forced to bail out the SIFI.⁷⁶ As the SIFI shareholders and creditors are aware of that, the former will favor excessively risky strategies while the latter will fail to guard against them.⁷⁷ This moral hazard problem is further worsened because shareholders are unlikely to be completely wiped out during a bailout. Hence, shareholders are likely to bear an even smaller fraction of the losses caused by the distress of their SIFI.

There are a number of reasons to believe that the expected value of a bailout for the shareholders is larger than zero. First, as a matter of historical fact, the average shareholder has received a positive payoff from bailouts. Even the bailout of AIG, which was litigated for being exceedingly harsh on shareholders,⁷⁸ actually prevented holders of AIG common stock from being wiped out.⁷⁹ Intuitively, politicians will prefer to intervene too early rather than too late. After all, they act to avoid the catastrophic effects of a systemic meltdown that may easily cause a serious recession and cost them re-election; thus, they are unlikely to wait until the very last moment when such are the stakes. Hence, they will tend to inject money into SIFIs when shareholders still own valuable shares thereby leaving them in place.

There is an even more fundamental reason to believe that the expected value of government aid for SIFI shareholders cannot be zero, namely, that alternatives to a bailout have a positive expected value for shareholders. Consider, for instance, the case of “regulation by deal,” in which a healthy firm is assisted by the government in the acquisition of a defaulting SIFI. Examples of this practice during the last financial crisis are ubiquitous and involved defaulting financial institutions, such as Bear Stearns and Merrill Lynch on one side and JP Morgan and Bank of America as their saviors, on the other.⁸⁰ An intervention of this kind clearly has a positive value for the shareholders of the acquired firm because they can sell their shares to the aided acquirer instead of being wiped out completely. Most importantly, as the shareholders know that by opposing the merger they can impose a large externality on the economy, they are likely to extract a high price for shares that might otherwise be almost worthless.⁸¹ In short, it is extremely unlikely that the expected value of government aid for SIFI shareholders can be set to zero.

Having established that the benefit of a bailout to shareholders tends to be higher than zero, it becomes even clearer why limited liability cannot be an adequate rule for SIFIs. A fundamental principle of modern corporations is that investors risk what they have invested in the company: when the expected value of government aid is positive, this principle is violated. Most importantly, it is violated exactly by those firms, like SIFIs, that have the potential to impose the greatest negative externalities on the economy. As noted above, this can result in moral hazard problems and induce shareholders (and creditors) to favor excessively risky strategies.⁸² Extended liability aims to restore this axiom and prevent this moral hazard problem.

B. Why Not Unlimited Liability?

This Section explores why shareholders should not face unlimited liability. The first step is defining exactly which losses shareholders would internalize if unlimited liability was the rule. While at first glance trivial, this task becomes very problematic in this context. To clarify this point, we divide the losses caused by the distress of a SIFI into three categories: losses to creditors, the cost of the bailout, and other externalities.⁸³

To begin with, SIFI shareholders could be liable for the losses experienced by contract creditors in addition to tort creditors. However, unlimited liability towards contract creditors is opposed even by the staunchest supporters of the idea of extending the liability faced by shareholders.⁸⁴ Unsurprisingly, to the best of our knowledge, nobody has ever explicitly advocated for an extension of liability towards contract creditors. The basic reasons are that creditors that want higher guarantees can simply contract around limited liability and that the interest rate will incorporate the level of liability chosen by the parties.⁸⁵ Nevertheless, it is important to explore this rule in order to have a complete picture of the alternatives available to the regulators.

In particular, unlimited liability towards creditors is simply inadequate to address the problems created by SIFIs. SIFIs are not bailed out because they impose losses on their creditors, but because they have the potential to impose enormous losses onto agents that have not contracted with the corporation.⁸⁶ Unlimited liability towards contract creditors would not solve this problem because neither creditors nor shareholders would have incentives to account for such externalities.

At best, unlimited liability towards contract creditors would merely constitute a transfer of resources from shareholders to creditors. Under limited liability, an insolvent SIFI imposes significant losses on its creditors, whereas under an unlimited liability regime, the fact that creditors can rely on shareholders’ assets would mitigate these losses. Therefore, making shareholders liable towards creditors might reduce the risk propensity of shareholders but would also reduce the incentives of creditors to monitor the SIFI.⁸⁷ Thus, this form of unlimited liability presents a trade-off in terms of monitoring incentives of shareholders and creditors.⁸⁸ It is unclear whether shareholders are in a better position than creditors to monitor the SIFI. Hence, it cannot be predicted whether unlimited liability towards creditors would make SIFIs safer.

At worst, unlimited liability towards creditors would simply achieve nothing. Because lending money to SIFIs would become safer, debtors would ask for lower interest rates. In turn, this would increase the returns to shareholders in good states of the world, which might offset their higher expected losses in bad states of the world (i.e., when the SIFI goes bust). The ability of shareholders and creditors to contract around interest rates to account for the level of risk might neutralize any substantial effect of extending liability towards creditors.

Against this background, increasing shareholder liability can only be effective if it forces agents with the ability to affect the behavior of the SIFI to account for externalities imposed on society at large.⁸⁹ One possible approach, suggested by Conti-Brown, is to make shareholders liable for the cost of actual bailouts as an alternative—freely chosen by the individual SIFI—to higher capital ratios.⁹⁰ However, having shareholders bear the cost of bailouts creates significant problems. To begin with, it would further politicize the debate around bailouts in a period of political and economic tension. Some politicians might seize the opportunity of earning consent by advocating for very large bailouts funded with shareholder money. The narrative of using the money of large institutional investors to cover the losses caused by the stock market is likely to gain significant traction among a certain part of the electorate. Other politicians, instead, would be exposed to the political pressure of a relatively small and cohesive group of institutional shareholders with a homogeneous interest in minimizing their expected liability and hence the size of the bailout. These conflicting political interests would generate policy outcomes that would weaken financial stability and overall social welfare at a particularly vulnerable moment.

Furthermore, for a liability rule to provide the right incentives, those facing potential liability must be able to affect the probability and the dimension of such liability by engaging in monitoring and other risk-reducing activities. However, in this case, the liability faced by the shareholders would largely depend on unpredictable political decisions outside their control. As a consequence, they would not be able to adopt the optimal precautions to minimize their liability risk. Concerns about shareholders’ limited capacity to estimate or to control the extent of their future liability prompt objections to the imposition of unlimited liability regimes.⁹¹ However, this standard objection is based on the inability of shareholders sufficiently to control the excessive risk-taking proclivities of their own agents, the managers of the corporation.⁹² In this context, shareholders’ inability to control the excessive risk-taking that could result in liability is even more profound because decisions about whether and how to bailout a SIFI depends not on the managers of the SIFI, over whom shareholders have some control, but on the government, which is an agent over which shareholders’ ability to exercise control is, and should be, attenuated. Therefore, making shareholdersliable for the portion of harm that depends on the reaction of the government would not improve financial stability. Rather, it would discourage investment in the financial sector.

The third option, namely making SIFIs’ shareholders unlimitedly liable for torts, would also be unviable. Current tort rules on causation and pure economic losses would make an unlimited liability rule largely ineffective in inducing SIFIs’ shareholders to internalize the losses caused by the distress of their firm. It would also be insufficient to couple shareholder liability with a change in tort law that stretches causation to allow for recovery of pure economic loss. Allowing the recovery of pure economic loss from shareholders would open the door to complex and endless litigation and ultimately result in uncertainty.

A significant part of the harm caused by the distress of a SIFI qualifies as what in tort parlance are called “pure economic losses,”⁹³ which are generally not recognized by the tort system. That is, compensation will generally be denied to plaintiffs that are unable to establish a direct connection between their physical injury or property damage and the acts or omissions of defendants.⁹⁴ Most of the losses that are caused by the distress of a SIFI are likely to be too remote to meet the traditional tests for causation required by the law of torts. Consider, for instance, the following case. A distressed SIFI cannot meet its obligation to pay one of its creditors, say, a carmaker. Hence, the carmaker cuts its orders from its supplier, which in turn goes bankrupt. Assume also that the carmaker could have kept a steady level of production after hearing of the SIFI’s distress by borrowing money at higher interest rates. Would the distress of the SIFI be the proximate cause of the default of the carmaker’s supplier? It is true that the supplier of the carmaker would have not gone bankrupt but for the distress of the SIFI, yet there was another event (i.e., the decision of the carmaker not to borrow more money) that likely broke the chain of causation. Given that the spillovers from a SIFI’s distress will propagate through the economy, countless instances akin to the one just described are likely to emerge that would fall outside the reach of tort law. Denying compensation for pure economic losses and for losses that are “remote” would again defeat the purpose of extending shareholders’ liability, which is inducing shareholders to account for the externalities that a default of their institution would cause on society at large. However, allowing plaintiffs to recover losses that are not proximately caused by the injurer and pure economic losses means that judges would have to develop a whole new set of rules just for dealing with SIFIs’ bankruptcies. It is easy to foresee that this process would involve costly and complex litigation and generate great uncertainty. Thus, unlimited liability seems unworkable with the existing tort rules on harm causation.

The last option would be making SIFI shareholders liable for all the liabilities of their institution, regardless of whether such liabilities are contractual in nature. In this vein, shareholders would have to cover the losses of the creditors, the cost of the bailout, and any additional loss caused by the default of the SIFI for which the SIFI itself is liable according to general tort law or other doctrines and rules. In other words, this solution would be a “pure” unlimited liability.

And yet, this solution is not workable either: it compounds the flaws of all the forms of unlimited liability analyzed so far while also creating four additional problems. First, due to the draconian risk associated with this form of shareholders’ liability, it will be difficult if not impossible for SIFIs to find equity investors. A lack of investment in them would damage the economy because SIFIs, like financial firms more generally, create positive externalities by making the allocation of capital within the economy more efficient and moving economic resources to more productive uses.⁹⁵ As Schumpeter noted more than a century ago, “the services provided by financial intermediaries—mobilizing savings, evaluating projects, managing risk, monitoring managers, and facilitating transactions—are essential for technological innovation and economic development.”⁹⁶ A large fraction of the benefit created by these activities is thus not internalized by financial intermediaries, and it spills over to the most productive sectors of the economy.⁹⁷ While economists have at times questioned the idea that financial development can facilitate economic growth, “[a] growing body of work would push even most skeptics toward the belief that the development of financial markets and institutions is a critical and inextricable part of the growth process.”⁹⁸

When an activity produces positive externalities, an unlimited liability rule thus results in over-deterrence, excessively discouraging parties from engaging in it.⁹⁹ Consider a simple example. Assume that by contributing to the efficient allocation of capital within an economy, a SIFI can make profits of $70 and produce a positive externality of $40, provided that it does not go bankrupt. Moreover, assume that the SIFI has a 50 percent chance of going bankrupt, and that by defaulting it would cause a systemic harm equal to $100, for which the SIFI would be liable according to general tort law rules. From a social perspective, it is optimal if the SIFI engages in its activity because its expected value is positive.¹⁰⁰ However, under an unlimited liability rule that does not account for positive externalities, the SIFI will find it unprofitable to conduct its activity. Therefore, unlimited liability would lead financial markets to shrink beyond what is optimal. In turn, this might have a negative impact on many other sectors of the economy.

A second drawback of unlimited liability is that it would create the perverse incentive of pressuring SIFIs to reduce their equity to the bare minimum required by regulation and to limit the extent of potential damage by concentrating ownership in the hands of a small number of shareholders.¹⁰¹ At present, most SIFIs hold reserves in excess of the minimum threshold imposed by the law.¹⁰² As unlimited liability increases the cost of equity vis-à-vis the cost of debt, it is less likely that SIFIs would continue to hold extra reserves if such a rule were passed.

The third problem with unlimited liability stems from the fact that bailouts take place in times of economic tensions; SIFI defaults, if a bailout is not engineered, create these tensions. During these times policymakers (and, to a lesser degree, courts) are exposed to great pressure from various interest groups and public opinion in general. As a result, political decisions taken in the proximity of a financial crisis tend to be, “to put it mildly, less than optimal.”¹⁰³ This is a significant drawback of any possible form of the unlimited liability rule because the quantification of the harm is bound to be made ex-post (i.e., during a time of economic and political tension), and hence to be driven by political considerations. In this vein, the amount of liability that shareholders will face is unpredictable ex ante and unlikely to be an accurate measure of the harm. Consequently, the deterrence effects of unlimited liability rules are highly imperfect.

Finally, perhaps the biggest drawback of a regime of unlimited liability as compared with our preferred approach of extended, but bounded, liability is that private liability insurance markets would be far more likely to emerge to protect investors in bounded liability regimes than in unlimited liability regimes. Just as liability insurance is available from private carriers in a variety of contexts, such as for officers and directors of public companies, we believe that insurance markets would generate liability coverage for shareholders facing heightened (but not unlimited) liability for the failure of the SIFIs they are invested in. As we further argue in Section IV.A, we are dubious that an active or vibrant insurance market for such a risk would exist if liability were unlimited because of the difficulty of calculating the risks of loss.

III. The Proposed Extended Liability Rule

This Part describes how we propose to shape an extended liability rule that accounts for the features of present-day financial markets and better serves the goal of neutralizing the moral hazard created by bailout expectations. With that goal in mind, we devise a liability regime specifically aimed to provide shareholders with greater incentives to monitor management so as to avoid excessive risk-taking. Its main function is deterrence rather than compensation. The regime we envisage is one of SIFI shareholder liability for up to the average share price in the period preceding the SIFI’s bailout or orderly liquidation, with the precise amount depending on the SIFI’s systemic relevance as proxied by its position within the financial network.

A. The Building Blocks of the Rule

To define the contours of our proposed extended liability rule for SIFI shareholders, we identify and answer seven questions: (1) What should be the upper bound to liability? (2) How can the extent of liability be made sensitive to the specific systemic relevance of the individual SIFI, so that shareholders have greater incentives to monitor those SIFIs that endanger the financial system more? (3) What would trigger extended SIFI shareholders’ liability? (4) Which shareholders should be liable? (5) What should be the standard for liability? (6) How would the new rule coordinate with the existing capital requirements regime? and (7) Who should recover and where would the money go?

1. The liability cap

The basic element of our proposal for an extended liability regime is the liability cap, or the maximum amount of liability SIFI shareholders should be required to bear. Under the traditional double liability rule, a bank’s shareholders were liable in excess to their investment in the bank up to the par value of their stocks.¹⁰⁴ In today’s markets, the par value is no longer a reliable proxy of the value of a firm. It suffices to notice that, as of August 6, 2019, the price of Apple shares was $193.34, whereas their par value was just $0.00001.¹⁰⁵ Hence, the stock price is over 19 million times higher than the par value. Similarly, the stock price of Goldman Sachs share on the same day was $201.68, whereas the par value was just $0.01.¹⁰⁶ For that reason, we need to introduce a different base to calculate the liability cap.

While no cap can be devised that will provide for the optimal level of shareholder monitoring over excessive risk-taking, we posit that a reasonable starting point is the average market price of a common stock during the time that goes from thirteen months before shareholder liability is triggered (t_₁₃) to one month before that (t_₁).¹⁰⁷ We call the time between t_₁₃and t_₁ the “value window.” The advantage of having a value window is threefold. First, the expected liability will not be drastically reduced by the unavoidable drop in the stock price that precedes a SIFI’s collapse. Second, a value window accounts for the fact that the monitoring of SIFIs is an ongoing process, and that a default of an institution of this magnitude is generally the result of a series of decisions taken during a considerable time interval. Recall, for instance, the movie “The Big Short.”¹⁰⁸ In describing the events that led to the financial crisis, it shows how SIFIs’ insiders engaged in reckless behavior for years before the financial system collapsed. In fact, over two years passed between the time when Michael Burry, the hedge fund manager, first discovered the problems with the housing markets and the explosion of the financial crisis. In this vein, tying the liability to a specific point in time would be unreasonable. Third, with an average of 365 data points, abnormal stock price fluctuations will not dramatically affect the expected value of the liability. For instance, assume that the price of Bank of America’s stocks is equal to $100 for the 365 days of the value window. In this case, the base to calculate the expected liability will be $100. Assume now that over a single day, the price of Bank of America’s shares doubles. In this case, the base to calculate the expected liability will be $100.27.¹⁰⁹ This stability increases predictability and protects shareholders from sudden, extreme changes in stock prices.

2. The multiplier

Double liability had the goal of preserving the stability of the single bank. Hence, it was reasonable to have a one size-fits-all solution for all banks. Our rule’s purpose is, rather, to preserve the stability of the financial system, and thus, we introduce a multiplier that reflects the risk that the individual SIFI imposes on the system. The advantage of using a multiplier is that the extent of liability can be made a function of the level of risk posed by the specific SIFI, thereby exposing the shareholders of SIFIs that can cause more (less) systemic harm to higher (lower) liability. As an outcome, shareholders of more central firms will have greater incentives to prevent excessive risk-taking.

There is ample empirical evidence suggesting that the systemic risk posed by an institution depends on its size and its position in the financial network.¹¹⁰ Therefore, the multiplier must account for the size of the SIFI and for its centrality, that is, for its systemic relevance. Larger and more central institutions pose higher systemic risk, and therefore, their shareholders should face higher liability. For these reasons, it is useful to define the multiplier in terms of a network-based (normalized) indicator of the firm’s systemic relevance, such as the DebtRank Score (DRS).¹¹¹ Imagine that the average stock price of the shares over the relevant value window for a given SIFI is $100, and that its DRS equals 0.6. Then the liability faced by the shareholder will be equal to 100 * (1 + 0.6) = $160.¹¹² Because 0 < DRS < 1, the maximal liability that shareholders can face is twice the average price of the shares during the value window.

To be sure, we do not claim that this multiplier leads to optimal deterrence. In fact, a risk-averse regulator that wants to further increase the incentives of SIFI shareholders to monitor might choose a multiplier greater than 1. And yet, even if optimal deterrence is not achieved, it is key to adopt a network measure of centrality like the DSR as a multiplier since it allows policymakers to connect the liability faced by SIFI shareholders to the centrality of their firm. This is important for two reasons.

First, the goal of the proposed liability rule is countering the perverse incentives created by the prospect of bailouts. As the expected value of the bailout is higher for firms that are more central, liability must also be higher to neutralize the effects on the shareholders. Second, the variation in the extent of liability for individual SIFIs generated by the DebtRank is especially relevant in a world in which large institutional investors own stakes in many SIFIs but have limited resources to engage in monitoring. Consider the case in which there are three systemically important financial institutions: JP Morgan Chase, Goldman Sachs, and Bank of America. Assume that a default of JP Morgan Chase would take down half of the financial system (i.e., its DRS is 0.5), whereas Goldman Sachs and Bank of America would only take down one tenth of the financial system (i.e., their DRS is 0.1). Last, imagine that mutual funds managed by BlackRock are equally invested in these three financial institutions. In a world in which a double liability is in place, BlackRock would face equal liability for the default of these three banks. Consequently, it will evenly divide the resources that it can devote to monitoring among the three institutions. However, this outcome is undesirable because JP Morgan Chase can cause much larger disruption to the economy, and therefore, it would be efficient if BlackRock concentrated a significant part of its resources on monitoring JP Morgan. Our rule would lead to this result. In fact, the tailored multiplier would cause BlackRock to face liability five times higher per dollar invested if JP Morgan Chase goes bankrupt than if either Goldman Sachs or Bank of America go bankrupt. For this reason, as desired, BlackRock would deploy more resources to monitoring JP Morgan Chase.

Defining the multiplier and the base to calculate the liability this way has another fundamental advantage compared to Conti-Brown’s proposal:¹¹³ it minimizes the role that interest groups and the political process play because the criteria to quantify the compensation that shareholders have to pay are defined ex ante (i.e., before the distress of the SIFI) and are entirely transparent. For this reason, shareholders are perfectly able to anticipate the liability they face and to set their level of monitoring accordingly.

3. The trigger

The most straightforward way to identify the exact moment of the extended liability trigger is to refer to the start of the OLA process.¹¹⁴ The OLA process has its own procedural complexities.¹¹⁵ The first step is a determination of systemic risk made by the Federal Reserve Board and the FDIC.¹¹⁶ This determination has to be ratified by the Treasury Secretary, in consultation with the President.¹¹⁷ Thus, liability could be triggered when this procedural process is completed.

There is, however, one obvious problem with this approach. As noted above, the OLA process might be inadequate to cope with the problems created by the largest SIFIs or with periods of economic tension that simultaneously endanger multiple large SIFIs.¹¹⁸ In this case, a regulator might be forced to bypass the OLA and opt for a bailout. Thus, using the OLA as trigger for the liability rule would have the paradoxical result of protecting the shareholders of the largest SIFIs from liability. A way to prevent this outcome is for liability to be triggered by an OLA determination or government intervention in favor of a SIFI in distress. Since the government has multiple ways to assist a SIFI (for instance, by issuing guarantees to facilitate the merger with a sound bank), it would be impossible to provide an exhaustive list of the kinds of interventions that should trigger shareholders’ liability. For this reason, we argue that shareholders should be liable whenever a government subsidy exceeds a de minimis threshold,¹¹⁹ independently from the form of the subsidy.

4. Which liable shareholders?

One obvious complication is that shares are frequently traded, and hence, shareholders will change over time. For this reason, it becomes important to define precisely when the liability should kick in to determine which individuals or institutions should be liable. Hansmann and Kraakman note that two corner solutions are possible. The first, which they call the “judgement” rule, attaches liability to the people that own shares at the moment in which a judgment is made.¹²⁰ In this context, it would imply that the liability attaches only to the shareholders that own the shares at the time of entry into the OLA proceeding or the bailout. The obvious problem with this approach is that when there are signals that the OLA process or the bailout is approaching, all potentially solvent shareholders will have strong incentives to sell their share at a very low cost to avoid liability. Another possible approach is the so-called “occurrence” rule, under which liability attaches at the moment in which the tort occurs.¹²¹ This approach means that whoever held shares at the moment in which the SIFI made the decisions leading to its bankruptcy should be held liable. This is an unworkable solution, however, because, in this context, it is impossible to pinpoint a single moment leading to the negative outcome.

We suggest that all the individuals and institutions that held shares during the twelve months leading to the OLA or bailout trigger (hereinafter, liability window) will be liable in proportion to the value of the shares they held, the time during which they held those shares, and the multiplier of the SIFI.¹²² Note that this window is not the same as the value window. In fact, while the value window goes from thirteen months before the trigger to one month before the trigger, the liability window goes from one year before the trigger to the trigger day. The liability window also includes the month preceding the OLA trigger because we intend to give shareholders incentives to monitor and to avoid excessive risk-taking between t_₁ and the OLA trigger.

This mechanism would work as follows: assume that a shareholder owned ten shares for the entire liability window and that these shares traded on average at $1 each during the value window. Assume also that the DRS of this institution is equal to 0.5. Her liability would then be equal to $15. Assume now that the same shareholder only held these shares for half of the liability window. Her liability would now be equal to $7.50. The advantage of this solution is twofold. On the one hand, it is hard to evade because divesting right before the OLA process is triggered has only a marginal impact on the liability faced by a shareholder. On the other hand, such a rule accounts for the fact that monitoring of SIFIs to ensure their stability is an ongoing process. In addition, it is consistent with the circumstance that a default of an institution of this magnitude is generally the result of a series of decisions taken during a considerable time interval.

5. The standard

Another fundamental aspect of the rule is which liability standard should apply. One possibility is to hold shareholders liable only if they have been negligent. The main problem with this approach, however, is that it would be very hard for courts to determine whether shareholders have been negligent in monitoring the manager. Moreover, it is perfectly possible that some shareholders have been negligent, whereas others have monitored carefully. Therefore, courts should make a separate negligence finding with respect to each shareholder. Besides being costly and time consuming, these negligence findings might undermine the deterrent effect if they are—as it is reasonable to expect—often inaccurate. On the contrary, a strict liability rule would be faster and easy to administer. All shareholders would become liable once the OLA procedure is triggered or the bailout executed, independently of the level of monitoring in which they had engaged.

The law and economics literature offers support for the choice of a strict liability over a negligence rule.¹²³ In fact, as a rule of thumb, strict liability can be desirable when the potential injurers (in our case, the shareholders) have better information than the courts on the optimal care level to be adopted.¹²⁴ In this context, it is certainly the case that shareholders, and especially sophisticated institutional investors, are more informed than the courts on how to monitor managers.

6. Extended liability and capital structure

One of the most devastating criticisms raised against the idea of increasing shareholders’ liability regards firms’ ability to adjust their capital structure.¹²⁵ Firms are interested in minimizing their overall cost of capital and will choose the combination of debt and equity that allows them to reach this goal.¹²⁶ Unlimited liability of shareholders would greatly increase the cost of equity while leaving unchanged the cost of debt.¹²⁷ As a consequence, firms would minimize their outstanding equity and switch to an even more highly leveraged capital structure.¹²⁸ To understand this point, imagine a very simple scenario—in line with the famous framework developed by Modigliani and Miller¹²⁹—in which the cost of debt and equity is identical, and hence, the firm will be indifferent between financing itself through debt or through equity. However, assume now that an unlimited liability rule for shareholders is introduced. The cost of equity will greatly increase to reflect the increased risk of liability because shareholders now face losses in excess of their investment in the firm. On the contrary, creditors’ potential losses are still limited to the amount lent to the firm. Hence, the cost of debt will remain constant. The same firm would now find it more convenient to increase its leverage and finance a larger part of its operation through debt to reduce its expected liability. However, as the value of equity shrinks, the expected liability per dollar invested in the firm by the equity holders increases. This further tips the balance in favor of debt financing, which will shrink the equity cushion even more. Thus, the paradoxical outcome of a rule aimed at extending shareholders’ liability could be a vicious circle in which firms adjust their capital structure so that there will be very little equity to which liability can be attached. And as the expected liability associated with the remaining equity will be extremely large, the share value will be much higher for judgment proof investors.

For SIFIs, however, the situation is drastically different because the regulator has imposed minimum capital requirements. That is, the regulator has set a minimum size of the equity cushion in the form of capital ratios. Therefore, SIFIs cannot reduce equity beyond a certain point.¹³⁰ Yet, under a rule of unlimited shareholder liability, shareholders would have even greater incentives not to go above the regulatory floor, which, according to leading economists, is inadequate.¹³¹ In other words, by increasing the liability exposure of equity capital, higher shareholder liability would make financing through debt even more convenient for SIFIs.

On the contrary, our proposed extended liability regime would induce SIFIs to reduce their leverage, without forcing onto them the straitjacket of one-size-fits-all capital requirements devised by a regulator that is bound to have imperfect information.

In fact, with a minor tweak, our proposed liability rule can reduce the cost of equity vis-à-vis the cost of debt and hence incentivize SIFIs to reduce their leverage. In particular, the shareholders should be allowed to reduce their liability by increasing the SIFI’s equity cushion. That is, shareholders should have to bear the full extended liability only if their capital exactly matches the mandated capital ratios. Any increase in tier-one capital held in excess to the capital requirement would generate a proportional reduction of theextended liability. For instance, assume that the capital requirement of a SIFI is 8 percent, and that the SIFI holds exactly 8 percent in tier one capital. The SIFI shareholders would then have to cover the extended liability in full. If the SIFI has an extra 4 percent of equity cushion, then the extended liability will be halved. In this vein, the shareholders will return to a regime of single liability if their SIFI has an equity cushion that is twice as large as the one mandated (in this example, 16 percent).

It is easy to see why this feature of our rule allows shareholders to reduce their liability by reducing leverage. Returning to the example, assume that now the SIFI—with an equity cushion of 12 percent—needs fresh money and must decide whether to issue bonds or raise new equity. In the former case, the basis on which liability is calculated would increase. On the opposite, if new equity is issued, the basis on which liability is calculated would decrease and the expected liability would be discounted by the amount of capital held in excess over the capital requirement.

With this tweak, our proposal would share with Conti-Brown’s the feature of granting shareholders the right to decide on the extent of their own liability.¹³² However, under his proposal, shareholders can only decide between a very high capital requirement or unlimited liability. Instead, our rule grants shareholders a much wider choice because it allows them to choose the combination of liability and capital requirement that they prefer within the parameters set by the regulator.

7. Collecting from shareholders

The goal of our proposed rule is deterrence. Hence, who gets the money is a second order problem. However, one obvious possibility is that the money collected goes to, and is collected by, the FDIC, as a reserve to be used to fund either payouts to depositors of failed banks or bridge financing within OLA proceedings.¹³³ Another possibility is that the funds are collected directly by the government to (partially) cover the cost of the bailout.

IV. Market Adaptions to an Extended Liability Regime

Having laid the foundations of our proposed extended liability rule in the previous section, in this Part, we explain how the market would adjust to the new liability rule, in particular by focusing on: (1) how the risks of incurring extended liability for SIFI failures would be shifted to insurance companies and derivative counterparties because insurance and derivative products would emerge to shoulder the risk of such extended liability in exchange for compensation for doing so; (2) how our rule is going to affect the composition of SIFIs’ shareholder base; and (3) how our liability rule would ameliorate the problem of firms being too big to manage.

A. Insurance Markets for Extended Liability Exposure

In his provocative article advocating for an unlimited liability regime for SIFIs’ shareholders, Conti-Brown argued that the “greatest feature” of this regime is that it would result in the creation of a new form of derivative instrument that would effectively shift the risk of unlimited liability from shareholders to derivative counterparties.¹³⁴ Specifically, Conti-Brown envisages the introduction of a particular derivative—the shareholder liability swap (SLS)—that would work, much like a credit default swap (CDS), as follows:

The issuer would guarantee to pay the holder of equity . . . enough to cover any losses following a taxpayer bailout. In this sense, an SLS is similar to a credit default swap (CDS), which pays a bondholder the value of a bond in the event the issuer of the bond defaults.¹³⁵

The advantage of a SLS would be twofold. First, the value of the SLS would serve as a signal of the solvency of the institution. Currently, the two most accredited indicators of a firm solvency are the ratings issued by credit rating agencies (CRAs) and CDSs. CRAs tend to react to mutated circumstances at a much slower speed than the market.¹³⁶ CDSs, in turn, are insensitive to the risk of a bailout because a bailout usually prevents a default; in other words, CDSs, unlike SLSs, underestimate the solvency risk of too-big-to-fail SIFIs. Building on this, Conti-Brown argues that variations in the price of SLS would serve as a useful tool to “determine the likelihood of a government bailout.”¹³⁷ Second, shareholders could distribute the risk posed by the default of their SIFI by buying SLS from agents that are in a better position to bear that risk.

The extended liability regime that we propose improves on this important ancillary element of Conti-Brown’s proposal. The price of a SLS under Conti-Brown’s unlimited liability rule would not depend only on the likelihood of a default, but also on the estimated value of the liability given default. The latter, however, is the result of the idiosyncratic political reaction to a situation of crisis and hence is hard to estimate for market participants. To put it differently, the probability of default of an institution is a risk (i.e., a form of “randomness whose probabilistic nature is extremely familiar and can be characterized with objective probabilities”¹³⁸), whereas the cost of a SIFI bailout and hence the extent of the expected liability, is uncertain (i.e., characterized by “randomness whose probabilistic behavior is extremely unfamiliar, unknown, or even unknowable”¹³⁹). Thus, the price of a derivative that accounts for the estimated value of the liability given default is bound to be an extremely noisy signal of the probability of SIFI default. Similarly, because it would be based on an event characterized by uncertainty, the SLS would be a poor mechanism to distribute risk in the market.¹⁴⁰

On the contrary, under our proposed extended liability regime, the value of a SLS would depend only on the probability of default because the amount of liability is clearly defined ex ante. Therefore, under our rule, the SLS would be more effective at distributing risk and would constitute a more reliable and transparent signal of a SIFI’s solvency.

Moreover, under a liability regime like the one we propose, an insurance market and/or a derivatives market for shifting extended liability risk would be highly likely to emerge. Such markets would have structural characteristics similar to those of the robust municipal bond insurance market or the CDS market that currently exist.¹⁴¹ Were such markets to develop, then either the SIFI would contract and pay for insurance that would pay out the additional money owed by investors in case of default or individual investors could buy their own policies or derivative contracts.

Moreover, as with municipal bond insurance, once an investor’s extended liability is insured, the performance of the SIFI will be closely monitored by its insurer through a process known as “surveillance” in the municipal bond context.¹⁴² The insurance company would be specialized in monitoring SIFIs’ behavior and therefore in a good position to identify excessively risky conducts. SIFIs would therefore be penalized for engaging in excessively risky activities because their insurance premiums would go up. Further, the regulator itself would have an additional market signal to act upon before it is too late.

B. Extended Liability and SIFIs’ Ownership

In general, by increasing the liability faced by the shareholders, our proposed rule increases the incentives of market participants to engage in monitoring.¹⁴³ However, it is important to understand in greater detail how our proposal would affect asset managers, given their key role in today’s equity market. The first question to address is: who should be liable? The funds that hold the shares (e.g., Fidelity Magellan Fund), the advisor that manages the funds (e.g., Fidelity), or both? We suggest that the advisor and the funds should be held jointly liable. The reason is that the former has the skills and the competences to engage in monitoring, whereas the latter has the resources to cover the liability. If the liability were placed only on asset managers, then they would be incentivized to hold less assets, thus potentially creating a judgment-proof problem. At the same time, funds themselves do not have the resources and the expertise that are necessary to monitor their portfolio firms. In this vein, making them liable might produce limited benefits in terms of increased monitoring.

But how would this liability rule change the incentives of asset managers? To answer this question, leaving aside the problem of how to transition to the new regime,¹⁴⁴ one must consider that under our extended liability regime, the price of SIFI shares will be affected by exactly the same factors that are influencing prices in its absence, but with one significant exception: changes in the stability of the SIFI will have a larger impact on share prices. In this vein, investors that have the ability to monitor their portfolio companies and more specifically to influence their management so as to prevent excessive risk-taking will find SIFI shares attractive. In fact, by increasing the safety and soundness of the SIFIs they invest in, such investors could reduce insurance premiums and ultimately increase share prices.

Of course, we recognize that not all shareholders are equally well equipped to engage in the monitoring of SIFIs. Indeed, we believe that there will be vast heterogeneity among the potential shareholding population with respect to their monitoring capabilities. In particular, under our proposal, passivity will be riskier for investors (including but not limited to those who invest in index funds as well as professionally managed mutual funds that have made a determination to remain passive). We believe that this increased risk is actually an attractive feature of our extended liability proposal for two reasons. First, as noted in the previous section, the existence of derivatives and the emergence of private insurance coverage for extended liability will shift most, and possibly all, of the risk of our proposal onto derivative and insurance company counterparties. This risk-shifting will ameliorate the risk to passive investors that hold stock in SIFIs directly or through mutual funds, or investors that otherwise choose to remain rationally ignorant and passive about the levels of risk-taking actually going on within particular financial institutions. However, we recognize that under our proposal, investors would still experience extended liability for SIFI losses if the SIFIs in which they invested failed and their insurers and/or derivative counterparties also failed. Because of this risk, we acknowledge that our proposal would make investing in SIFIs less attractive for passive investors. But at the same time, it would make investing in SIFIs more attractive for active, sophisticated investors because it would increase the expected returns associated with locating arbitrage opportunities in SIFIs that are less risky than they are perceived to be in the market. The added risk of incremental liability from our proposal would have the effect of magnifying the available arbitrage opportunities beyond what they would be under the current system of limited liability and anticipated government bailouts.

C. Too Big to Manage

According to many commentators, large banks are not only too big to fail but also too big to manage.¹⁴⁵ The basic argument is that it is very hard for executives, boards, and shareholders to oversee large banks with hundreds of thousands of employees.¹⁴⁶ The main issue is that the activities of banks, and especially of the largest institutions, are incredibly complex and opaque.¹⁴⁷ Therefore, even assuming that they have the best intentions, managers and shareholders might not be able to manage a SIFI’s risk exposure.¹⁴⁸ The problem significantly worsens when one considers that managers’ incentives are further skewed toward risk-taking by the typical compensation contracts in the banking sector.¹⁴⁹

Against this background, one could argue that shareholders would have incentives to push toward shrinking the size of their SIFI in order to minimize the risk of losses. However, as noted in Section II.A, SIFIs’ shareholders enjoy an implicit subsidy, which protects them from possible losses because their firm is too big to fail. Thus, they have no incentives to break up their SIFI, despite the fact that this would facilitate their monitoring. Our liability rule would address this problem as well. By increasing shareholder exposure to downside risk, our proposal gives shareholders stronger incentives to ensure that the SIFI becomes of the “optimal” size. That is, sufficiently big to enjoy the relevant economies of scale that are associated with high volumes of activity but also not so large that it becomes impossible to monitor.

V. Why Standard Objections to Unlimited Liability Do Not Apply

Since Hansmann and Kraakman’s seminal article,¹⁵⁰ any proposal to extend shareholder liability must wrestle with five standard objections: (1) firms would just finance themselves through debt instead of equity; (2) shareholders do not have the ability to monitor; (3) an extended liability rule would result in shares being held by judgment proof shareholders; (4) it is hard to collect from offshore investors; and (5) it is hard to define when the liability attaches. We have already dealt with points (1) and (5) in Section III.A. The fourth objection loses most of its bite in the face of the explosion of international arbitration. Here, we also show that the second and third objections are, respectively, no longer valid following the reconcentration of ownership in the hands of institutional investors that has taken place in the last thirty years and is easy to address. Moreover, it is important to remark that extending liability will impose a loss on the existing shareholders. And while it is appropriate that the cost of equity incorporates the risk posed by the firm,¹⁵¹ it is still desirable to compensate shareholders that hold shares during the transition period. We discuss how this compensation can be carried out in Section V.C.

A. Shareholders’ Monitoring

One of the most convincing arguments against unlimited liability is that shareholders would not be able to effectively monitor the behavior of their firm independently from the amount of liability to which they are exposed simply because they lack the required competences.¹⁵² However, such an argument has lost most of its bite due to the transition from the Berle-Means corporation, characterized by diffuse ownership, to present-day institutional ownership.¹⁵³ In fact, unlike small retail shareholders, large and sophisticated institutional investors, such as Fidelity, Capital Research, and T. Rowe Price, have the resources and the know-how to monitor management provided that they are given sufficient incentives to do so.¹⁵⁴

A related critique of unlimited liability is that an investor would have to monitor the wealth of other shareholders because if the latter is insolvent, the investor would be exposed to higher liability.¹⁵⁵ However, this problem only arises if shareholders are jointly liable,¹⁵⁶ whereas under our rule, each shareholder can only be asked to pay in accordance to the formula described in Section III.A. Therefore, the expected liability of one shareholder is not affected by the wealth of the other.

B. Judgment Proof

The key argument behind the idea of unlimited liability is that it will induce shareholders to monitor the firm more carefully and will lower stock prices to reflect the expected value of liability.¹⁵⁷ Yet, increasing shareholders’ liability cannot have these effects if the shareholders are judgment proof. At the extreme, a shareholder that has invested all her assets in a corporation will be indifferent between an unlimited liability rule and the current regime of limited liability. While it is extremely unlikely that the judgment-proof problem affecting shareholders is this severe, the key idea that high levels of liability cannot influence the behavior of shallow-pocket shareholders remains. However, this problem would be significantly less severe under our rule than under unlimited liability, especially given the current ownership structure of SIFIs.

In particular, under this rule the maximum liability would be twice the shareholder’s investment (when DRS = 1). It is implausible that investors are undiversified to the point of placing more than half of their assets in a single SIFI. Nowadays, the vast majority of the shares are held by large institutional investors. Our proposed extended liability rule will not bankrupt these investors because they can hedge this risk via a SLS.¹⁵⁸

In any event, because the liability cap is clearly spelled out ex ante, it is easy to prevent deep-pocketed investors from passing their shares to shallow-pocket ones before liability is triggered. Assume that there are two individuals, A and B, and that A is a deep-pocketed shareholder, whereas B owns no assets. A might have incentives to park A’s shares with B in order to escape liability. The following rule may prevent this outcome. If B reports the scheme, A will have to pay treble damages, two-thirds of which will go to B. Let us return to the example presented in the introduction in which the expected liability equals $160. If B reports the scheme, A will be required to pay $480, of which B will pocket $320. The optimal strategy for B would then be accepting to hold A’s shares and then report A. Thus, A will have no incentives to transfer shares to B in the first place.

One obvious problem is that it would be possible to create limited liability companies with limited assets to hold SIFI shares.¹⁵⁹ To prevent this, it is important that these companies’ corporate veils can be pierced, but only up to a value sufficient to cover the liability obligation of the shareholders that derive from the default of a SIFI. In other words, assume that Apple owns shares in JP Morgan Chase. To avoid liability, Apple might create a company with limited assets—call it Rotten Apple—to which it could contribute its JP Morgan Chase’s shares. While for any other liability we argue that the standard veil-piercing doctrines should apply to Rotten Apple, for the liability associated with JP Morgan Chase’s default, the veil should automatically be pierced so that Apple is liable.

C. Transitioning to the New Regime: Compensating Existing SIFI Shareholders

Implementing the proposed reform might create a transition problem. Specifically, if a law were passed imposing extended liability, anyone holding shares in a SIFI at the time it entered into force would experience a sudden decline in share value as share prices adjusted to reflect the new, greater risk associated with owning shares in SIFIs. And while the very purpose of increasing the liability faced by shareholders is to ensure that the equity price adequately accounts for the risk posed by the corporation,¹⁶⁰ it might be problematic to impose losses on shareholders that purchased the shares when the rule of the game was limited liability. A possible solution would be compensating shareholders for the loss caused by the increased liability. In the U.S. context, shareholders could be compensated from payments from the FDIC deposit insurance fund.

While compensating shareholders for the transition from limited liability to extended liability seems desirable, it is important to carefully consider how to define the quantum of the compensation. For instance, if the compensation is determined ex ante by the regulator based on an estimation of the price drop, then it is bound to be inaccurate and arbitrary. At the same time, calculating compensation based on the observed price drop between pre-liability (say, t_pre) and post-liability (say, t_post) would also be problematic. First, it is hard to identify t_pre.Because investors will foresee that the liability rule will change, the stock prices will begin to drop even before the law is enacted. Second, basing compensation on the observed price drop would incentivize shareholders to take on excessive risk during the transition period. In fact, if the shareholders bet strong on risky projects and these projects prove to be successful, the stock price of their company will increase, and they will benefit. Instead, if these risky projects fail, the shareholders would be able to externalize the losses to the FDIC. Simply put, the shareholders would be able to play a “heads I win; tails [the FDIC] loses” game.¹⁶¹ Third, the riskiest SIFIs will face a larger price drop, and paradoxically, under this rule, their shareholders would also be the ones that receive the largest compensation.

A way out of this impasse would be structuring a mechanism akin to an auction among SIFIs. Contrary to normal auctions, we suggest a mechanism of descending prices. In our context, the auction would work as follows: first, the regulator imposes a ceiling for the compensation, say 10 percent of the market value of a SIFI before the rule was first discussed. This ceiling is likely to be inaccurate for the reasons highlighted above. Yet, because it is only needed to start the auction, its accuracy is less important. At this point, the shareholders of each SIFI will be asked to state the lowest percentage of the ceiling that they would be willing to accept. The compensation will then be set at the percentage offered by the SIFI that wins the auction (i.e., that offered the lowest percentage). To incentivize SIFIs to offer low percentages, the winner of the auction could be awarded an additional compensation, say 50 percent more than the percentage offered. To exemplify, assume that there are three SIFIs: Bank of America, JP Morgan Chase, and Goldman Sachs. Assume also that Bank of America offers to accept 50 percent of the ceiling, JP Morgan Chase 30 percent, and Goldman Sachs 10 percent. The compensation received by the three SIFIs will be 10 percent of the ceiling because it was the lowest bid. However, Goldman Sachs would receive 15 percent of the ceiling.

In addition to solving the problem of determining the price drop, this mechanism would have an important advantage. As noted above, the new liability rule will have a lower impact on the share price of safer SIFIs. For this reason, the safest SIFI would be more likely to place the winning bid and receive the added compensation.

Conclusion

The dramatic consequences of the last financial crisis are a painful reminder of why we should prevent SIFIs from taking on excessive risk. However, the complexity and the size of these financial giants make it impossible for policymakers to craft regulations that can effectively constrain SIFIs’ risk-taking. For this reason, this Article suggests that policymakers should also directly tackle the incentives of their shareholders and reduce their risk propensity.

We have argued that policymakers can achieve this goal by abandoning the current rule limiting the liability of SIFI shareholders to the amount of their investment, thus increasing SIFIs’ shareholder exposure to downside risk. In itself, this proposal is only partly new, since for three quarters of a century, roughly between the Civil War and the Great Depression, the shareholders of U.S. banks were subject to double liability. The characteristics of present-day financial markets—the presence of capital ratios for financial institutions, the dominance of institutional share ownership, and the availability of well-developed insurance and derivatives markets—create the perfect conditions for implementing a modified version of the traditional double liability rule. More precisely, we propose that SIFI shareholders face extended liability, up to twice the average share price in a twelve-month period prior to the SIFI’s default or bailout, depending on the SIFI’s systemic relevance. This would strengthen shareholders’ incentives to monitor SIFIs and prevent them from engaging in excessively risky, socially harmful conducts.

Extended Shareholder Liability for Systemically Important Financial Institutions

Share this post

Contact Us

Join Our Mailing List