As the financial system’s capital was being depleted in the 2007-2009 financial crisis, some banks curtailed their dividends but others, especially securities firms, continued to pay dividends well into the depth of the crisis. Indeed, some firms — including those that entered financial distress — actually increased their dividends during the crisis.
In our paper, available here, we provide a framework that can accommodate such divergence in the reactions of financial intermediaries in their capital decisions. Using this framework, we ask how divergent interests of the banks’ stakeholders are likely to play out during times of heightened financial distress.
Dividend payouts shift the relative value of claims between a firm’s stakeholders, but in the banking sector, interlocking balance sheets across banks introduce another dimension to the distributional implications of dividend policy. Dividend payouts can shift the relative value of stakeholders’ claims across firms as well as within each firm. Through this channel, one bank’s capital policy affects the equity value and risk of default of other banks through the classic risk-shifting incentives described in Jensen and Meckling (1976). In a model where such externalities are strong, the private equilibrium has excessive dividends and inefficient recapitalization relative to the efficient policy that maximizes banking sector equity value. In this way, banking sector capital takes on the attributes of a public good even among bank shareholders. This aspect of bank capital as a public good is in addition to the broader macro arguments concerning undercapitalized banks being a drag on economic activity, thereby perpetuating weak banks and slow growth (Shin (2016a)).
We highlight the role played by banks’ franchise values as the determinant of equilibrium behavior. Lower franchise values exacerbate the incentive problem and lead to worse outcomes. To the extent that banks’ market-to-book multiples provide a proxy for franchise values, our arguments take on added significance in the current environment of depressed market-to-book ratios in the banking sector.
In the simplified setting of our model with a fixed terminal date, the optimal policy that maximizes the value of the bank (total equity plus debt value) is to pay no dividends. This is because paying dividends will decrease the total value of the bank by increasing the probability that its franchise value will be lost. However, we show that in practice when the dividend policy of a bank is set to maximize only its equity holders’ value, the dividend policy reflects a tradeoff between (i) paying out to equity holders the available cash today rather than transferring it to creditors in default states in the future; and (ii) saving the equity’s option on the franchise value since dividend payout raises the likelihood of default and thus foregoing this value. When debt is risky (i.e, when bank leverage is sufficiently high), the optimal dividend policy depends on the bank’s franchise value. If the franchise value exceeds a critical threshold, the effect in (ii) dominates and it becomes optimal for the bank not to pay any dividends. However, if the franchise value is below the critical threshold such that risk-shifting benefits in (i) become dominant, then the bank would pay out all available cash as dividends.
Equilibrium in the dividend game reflects the spillovers across banks, and the benchmark model with common knowledge of payoffs gives rise to multiple equilibria. We refine these equilibria using global game methods, and derive the unique switching equilibrium. In this equilibrium, each bank’s payout policy depends on its franchise value. There is a critical threshold for the franchise value such that a bank pays no dividends above the threshold and pays maximum dividends below the threshold.[1]
The equilibrium is inefficient relative to the capital policy that maximizes the equity value of the banking sector. This inefficiency is in addition to the classic debt overhang problem described by Myers (1977). In our model, banks do not have an incentive to curb excessive dividends, as the benefits accrue only to their creditors who in turn are their interconnected counterparties. The innovative features of our two-bank model are that (i) interconnectedness can give rise to strategic complementarities in dividend decisions; and (ii) the socially optimal outcome can be obtained through coordination among shareholders of interconnected banks. Our model applies generally to all firms in financial distress, but is most relevant in the context of bank holding companies, especially broker-dealer firms, as they have high levels of leverage and interconnectedness.
The main innovation in our paper is in modeling this interaction of two agency problems — an agency friction between interconnected banks superimposed over an agency friction between the banks and their outside creditors. While risk-shifting in the one-bank model can be efficiently priced or contracted away (using covenants, e.g.) by individual bank creditors, restricting dividends is desirable in the case of interconnected banks over and above the shareholder-creditor conflict. In particular, cutting dividends can make even equity claims of both banks more valuable, but this is an externality that they do not internalize. While outside creditors are also hurt by this agency friction between interconnected banks, they can limit such risk-shifting by only using exclusive contracts (e.g., covenants that are tied not just to the behavior of banks they lend to, but also of other banks).[2]
We extend our model to incorporate negative dividends, which we interpret as equity issuance. We show that the result on excessive dividends in our benchmark model generalizes to under-capitalization of banks in this extension. By issuing equity, a bank increases the value of claims of its creditors, among which are shareholders of its interconnected banks. Banks pay the cost of equity issuance but do not internalize this positive externality. One key question is why banks do not find ex ante mechanisms to mitigate such ex post agency problems, e.g, by including dividend cutoff or earnings retention covenants in bank debt contracts. We review the classic arguments in the context of our model and discuss the possible impediments to addressing the inefficiencies through the classic contracting approach.
ENDNOTES
[1] The solution method does not rely on the iterative dominance argument of Morris and Shin (1998, 2003), but focuses on showing uniqueness of equilibrium, as first examined by Goldstein and Pauzner (2005).
[2] Note, however, that the presence of explicit or implicit government guarantees can imply that even in the one-bank case, creditors may not have incentives to limit shareholder dividends. In effect, the shareholder-creditor conflict in the presence of a government guarantee turns into a conflict between the taxpayer and the bank as a whole.
This post comes to us from Viral V. Acharya, Professor of Finance at New York University’s Leonard N. Stern School of Business, a Fellow at the Centre for Economic Policy Research and the National Bureau of Economic Research and the C. V. Starr Professor of Economics at New York University; Professor Hanh Le of the University of Illinois at Chicago; and Hyun Song Shin, a Professor at Princeton University and an Economic Adviser and Head of Research at the Bank for International Settlements. It is based on their recent paper, “Bank Capital and Dividend Externalities,” available here.