Why High Leverage is Optimal for Banks

The following post comes to us from Harry DeAngelo, Professor of Finance at the University of Southern California, and René Stulz, Professor of Finance at Ohio State University.

In our paper, Why High Leverage is Optimal for Banks, which was recently made publicly available on SSRN, we focus on banks’ role as producers of liquid financial claims. Our model assumes uncertainty and excludes agency problems, deposit insurance, taxes, and other distortions that would lead banks to adopt levered capital structures. We show that, under these idealized conditions, high bank leverage is optimal when there is a market premium for the production of (socially valuable) liquid claims. The analysis thus implies that high bank leverage – not Modigliani and Miller’s (1958) leverage irrelevance principle – is the appropriate idealized-world baseline for analyzing bank capital structure in the presence of a demand for liquid financial claims per se.

Many economists believe that bank leverage is excessive, and should be curtailed by regulations requiring much larger equity cushions. For example, twenty prominent economists conclude that, with much more equity funding, banks could perform all their socially useful functions and support growth without endangering the financial system. Admati and Hellwig (2013), building on Miller (1995), Pfleiderer (2010), and Admati, DeMarzo, Hellwig, and Pfleiderer (2011), detail the case for strong regulatory limits on bank leverage. As Myerson (2013, p. 3) discusses, the case rests on Modigliani and Miller’s (1958, MM) leverage irrelevance theorem, which shows that, in perfect markets, equity is no more expensive than debt as a source of capital. MM’s debt-equity neutrality principle leads Admati and Hellwig (2013, p. 191) to conclude: “increasing equity requirements from 3 percent to 25 percent of banks’ total assets would involve only a reshuffling of financial claims in the economy to create a better and safer financial system. There would be no cost to society whatsoever.” In fact, with the MM theorem at work, increasing bank equity requirements well above 25 percent would have no social costs.

This argument treats banks as firms that make loans and ignores banks’ role as producers of liquid financial claims. The idea that liquidity production is intrinsic to financial intermediation is discussed extensively in the literature by, among others, Diamond and Dybvig (1983), Diamond and Rajan (2001), Gorton (2010), Gorton and Pennacchi (1990), and Holmstrom and Tirole (1998, 2011). If banks’ credit-screening technology enables them to make better loans than other parties could and all other MM assumptions hold, banks could adopt all-equity capital structures with no loss in value. However, if banks generate value by producing liquid claims for financially constrained firms and households, those with high-equity capital structures are not competitive with banks or bank-substitutes that have less equity.

We develop a simple segmented-markets model that deviates from MM by inclusion of some financially constrained firms (non-banks) and households that willingly pay a premium for liquid financial claims to provide assured access to capital. The model also allows, but does not require, the existence of some financially constrained parties that willingly pay a premium to obtain bank loans because their access to capital markets is costly or otherwise impaired. Our model indicates that:

  • High leverage is an essential, uniquely optimal feature of bank capital structures when liquidity is priced at a premium due to demand for assured access to capital.
  • Banks choose high leverage despite the absence of agency costs, deposit insurance, tax motives to borrow, reaching for yield, ROE-based compensation, or any other distortion.
  • Greater competition that squeezes bank liquidity and loan spreads diminishes equity value and thereby raises optimal bank leverage ratios.
  •  If conventional banks face regulatory limits on leverage while shadow banks do not, the former will be at a competitive disadvantage to the latter. Liquidity production will migrate from regulated banks into the unregulated shadow-banking sector.
  • When liquidity is priced at a premium, banks can and will choose safe asset structures to support capital structures that maximize the production of safe financial claims to satisfy the demand for liquid claims.
  • Because the MM capital-market conditions enable banks to construct perfectly safe asset and liability structures, there is no chance of bank default and no chance of systemic meltdown when these conditions hold.

These conclusions reflect the fact that a market price premium to induce banks to produce (socially valuable) liquid financial claims is incompatible with MM’s leverage irrelevance result. A liquidity premium is a price signal sent to banks and other liquidity producers to supply liquid financial claims that meet the demand for such claims from financially constrained firms and other parties that desire assured access to capital. The liquid financial claims (cash balances and similar claims) held as assets by financially constrained parties are drawn from the supply of safe claims produced by the liability structure decisions of banks and other liquidity suppliers.

When market prices embed a liquidity premium, bank capital structure decisions matter and are jointly optimized with asset structure to capture the maximum value from liquid-claim production. Debt and equity are not equally attractive sources of bank capital. Debt has a strict advantage because it has the informational insensitivity property – immediacy, safety, and ease of valuation – desired by those seeking liquidity. High bank leverage is accordingly optimal when the MM model is modified to include a price premium to induce (socially valuable) liquidity production.

Because MM’s debt-equity neutrality principle assigns zero weight to the social value of liquidity, it is an inappropriately equity-biased baseline for assessing whether the high leverage ratios of real-world banks are excessive or socially destructive. Nor does the MM theorem imply that regulatory limits on bank leverage are desirable because they would reduce systemic risk without increasing banks’ funding costs. In our idealized setting, such regulations would raise banks’ funding costs without changing systemic risk (which is nil because banks’ asset structures are optimized under conditions of uncertainty to produce safe financial claims).

The analysis thus cautions against accepting the view that high bank leverage must be the result of moral hazard, other agency problems, or tax motives to borrow. In our modified MM model, there are no agency costs or taxes, yet high bank leverage is optimal. High leverage is simply the result of intermediation focused on producing (privately and socially beneficial) liquid financial claims. The analysis also cautions against concluding that bank leverage must be too high because operating firms maintain much lower leverage. Banks create value through their capital structure choices. In our idealized setting, banks construct safe asset structures (from risky loans and securities) to support greater production of liquid financial claims, which manifests in high debt capital structures. Operating firms create value through real project choices, which commonly entail significant cash flow uncertainty.

Capital structure is a sideshow for value creation at operating firms, but it is the star (or at least a star) of the show at banks. The risky asset structures of most operating firms are poorly suited to support high leverage or large-scale production of liquidity. Banks exist because specialization and the associated cost efficiencies give them a comparative advantage over operating firms in arranging their asset structures to support capital structures that produce liquid financial claims to meet demand.

The general point is that, given a material market demand for liquidity, intermediaries will emerge to meet that demand with high leverage capital structures, which are made possible by (diversified) asset structures optimized to support the greatest possible quantity of liquid financial claims. This is the fundamental reason why MM’s debt-equity neutrality principle is an inappropriate equity-biased baseline for concluding that bank leverage should be curtailed significantly.

The paper thus shows that, if we take an idealized model (as MM do) and include a demand for liquid claims, the right baseline for banks is high leverage, not indifference to leverage. An important caveat is that real-world banks fall far short of the perfect asset-side diversification of risks they adopt under the idealized conditions we study. Regulation limiting leverage accordingly makes sense for real-world banks, but our analysis suggests a cautious approach to such regulation. Radical leverage reductions seem obviously desirable when moral hazard and other such problems that increase systemic risk are benchmarked against MM’s leverage irrelevance result. But since high leverage is the idealized-world norm for banks, a better approach weighs the possibility that severe capital standards might significantly impair production of socially valuable liquidity, and perhaps exacerbate systemic risk by inducing a substitution of liquid-claim production into the unregulated shadow-banking sector.

The full paper is available for download here.

Both comments and trackbacks are currently closed.