Hellmann–Murdock–Stiglitz Bank Capital and Deposit-Rate Regulation Model

Hellmann–Murdock–Stiglitz bank capital and deposit-rate regulation model is a dynamic microeconomic framework in business and economics theory demonstrating how financial liberalization and deposit-rate competition erode commercial bank franchise value, exacerbate risk-shifting moral hazard, and render flat minimum capital adequacy requirements Pareto-inefficient when used as an isolated regulatory instrument. Formulated by economists Thomas F. Hellmann, Kevin C. Murdock, and Joseph E. Stiglitz in their 2000 American Economic Review article "Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough?", the model resolves a central puzzle of late-twentieth-century financial instability: why banking crises across the United States, Japan, Scandinavia, and East Asia erupted shortly after authorities deregulated interest-rate ceilings while relying on Basel-style capital ratios to discipline bank risk-taking. By modeling banking as an infinite-horizon dynamic game rather than a static one-period contracting problem, Hellmann, Murdock, and Stiglitz prove that bank prudence is governed jointly by a capital-at-risk effect (current shareholder equity lost upon insolvency) and a franchise-value effect (the discounted present value of future deposit-intermediation rents forfeited when regulators revoke a failed bank's charter). Because banks competing in an unconstrained deposit market bid up deposit rates to steal market share—compressing ongoing intermediation margins and shrinking charter values—forcing banks to hold higher equity capital requirements to restore prudence inflicts a deadweight resource cost that can be avoided by combining moderate capital requirements with binding deposit-rate ceilings.

During the mid-twentieth century, banking systems in industrial economies operated under structural and price regulations instituted after the Great Depression. In the United States, Regulation Q (enacted under the Banking Act of 1933) prohibited interest payments on demand deposits and imposed binding ceilings on savings and time deposit interest rates, while branching restrictions insulated incumbent banks from intense price competition. In Japan, the Ministry of Finance maintained the convoy system of administered deposit rates. Under these regimes, commercial banks enjoyed stable net interest margins and high charter value (or franchise value)—defined as the net present value of expected future economic profits that a bank earns by remaining open and retaining its banking license.

Beginning in the late 1970s and 1980s, financial deregulation dismantled deposit-rate ceilings worldwide, including the U.S. Depository Institutions Deregulation and Monetary Control Act of 1980 and the Garn–St Germain Act of 1982. Simultaneously, deposit insurance coverage was expanded (raising U.S. federal deposit insurance limits from 40,000 USD to 100,000 USD per account), shielding depositors from bank default risk and removing their incentive to monitor bank asset portfolios, a mechanism analyzed in the Calomiris–Kahn demandable debt discipline and Diamond–Dybvig frameworks. In 1990, Michael C. Keeley documented empirically that the market-to-book asset ratios (Tobin's q) of U.S. bank holding companies declined from the 1960s through the 1980s as deregulation intensified competition, and that banks with lower charter values took on higher asset risk and operated with thinner capital buffers.

To counteract this moral hazard without reinstating price controls, international supervisors converged on the 1988 Basel Accord (Basel I), mandating minimum risk-weighted capital-to-asset ratios (initially 8 percent). The prevailing regulatory view held that forcing shareholders to commit equity "skin in the game" would make banks internalize downside losses and select prudent investments even in a liberalized deposit market. Hellmann, Murdock, and Stiglitz challenged both the static foundations of Basel I and its neglect of endogenous deposit-rate determination, proving that capital requirements interact with deposit-market competition through two opposing channels.

Infinite-Horizon Timeline and Balance Sheet Structure

Consider an economy operating over discrete periods t = 0, 1, 2, …, ∞, populated by N imperfectly competitive commercial banks and a continuum of household depositors whose aggregate supply of deposits increases with the interest rate offered. In each period t, bank i raises deposits D(r_i, r_−i), which depend positively on its own offered deposit rate r_i (∂D/∂r_i > 0) and negatively on rival deposit rates r_−i.

Let k ≥ 0 denote the ratio of shareholder equity capital to deposits. For every 1 unit of insured deposits raised at promised interest rate r_i, bank i contributes k units of shareholder equity and invests total loanable funds of A(k) = 1 + k. Shareholder equity is costly relative to deposits—owing to tax disadvantages, agency dilution costs under information asymmetry, or the liquidity premium households attach to money-like bank deposits as shown in the Diamond–Rajan fragility and liquidity creation model. Let ρ denote the opportunity cost on bank equity capital per period, where ρ > α > r_i and α is the gross return on the bank's safe, prudent investment project. Holding an extra unit of equity capital per dollar of deposits carries a positive opportunity cost of (ρ − α) per period.

Prudent versus Gambling Asset Technologies

In each period, after raising deposits at rate r_i and committing equity k, bank i allocates its total portfolio (1 + k) between two mutually exclusive lending technologies (a principal-agent problem because neither depositors nor regulators observe portfolio risk ex ante):

1. The Prudent Asset: Yields a deterministic gross return of α per unit invested with probability 1.
2. The Gambling Asset: A high-variance, negative-net-present-value project yielding a high payoff γ > α in the "success" state with probability θ ∈ (0, 1), and a low salvage payoff β < α in the "failure" state with probability (1 − θ).

The prudent asset has a higher expected social return than the gambling asset, while the gambling asset has a higher upside return in success and triggers bank insolvency in failure:

Because (1 + k)β < r_i, if the bank chooses the gambling asset and fails (probability 1 − θ), asset liquidation value (1 + k)β cannot repay promised deposit liabilities r_i. Under limited liability and deposit insurance, depositors are paid by the insurance fund, shareholders receive zero in that period, and the banking supervisor revokes the bank's charter, terminating its access to future period profits.

Per-Period Profit Functions Under Prudent and Gambling Strategies

The single-period expected net returns per unit of deposits (net of equity opportunity cost ρk) under each strategy are:

• Prudent Per-Period Margin π_P(r_i, k): The bank pays r_i to depositors, earns α(1 + k) on assets, and incurs equity cost ρk:

which decomposes into the net deposit-intermediation spread (α − r_i) minus the excess cost of bank capital (ρ − α)k.

• Gambling Per-Period Expected Margin π_G(r_i, k): Shareholders receive (1 + k)γ − r_i with probability θ and 0 with probability (1 − θ) under limited liability, while committing equity k at cost ρk:

Subtracting π_P from π_G gives the one-period temptation to gamble:

By gambling, the bank shifts the downside loss (1 − θ)[r_i − (1 + k)β] onto the deposit insurer while capturing the upside gain θ(1 + k)(γ − α).

Dynamic Charter Value and the Intertemporal No-Gambling Condition

In the infinite-horizon game with discount factor δ ∈ (0, 1), a bank investing prudently every period survives with probability 1 and earns the prudent franchise value (charter value) V_P(r_i, k) = π_P(r_i, k)D(r_i, r_−i)/(1 − δ). If the bank gambles, it survives into the next period only with probability θ; with probability (1 − θ) the gamble fails and the bank loses its license forever. A one-stage deviation from prudence is unprofitable if and only if π_PD + δV_P ≥ π_GD + δθV_P, which yields the Hellmann–Murdock–Stiglitz No-Gambling Condition (NGC):

A bank invests prudently when the one-period static gain from gambling does not exceed the expected franchise value lost upon failure. Solving the No-Gambling Condition for the maximum sustainable deposit interest rate r̂(k) above which a bank switches from the prudent asset to the gambling asset gives:

The Franchise-Value Erosion Effect of Costly Capital

Differentiating the No-Gambling Threshold r̂(k) with respect to the capital ratio k yields two opposing forces:

1. The Positive Static Capital-at-Risk Effect [(1 − δ)(α − θγ)/(1 − θ)]: Holding more current equity k scales up invested funds (1 + k) and reduces the static one-period gain from gambling (when α > θγ).
2. The Negative Dynamic Franchise-Value Effect [−δ(ρ − α)]: Because equity is more expensive than deposits (ρ > α), forcing the bank to hold a higher capital ratio k reduces its future per-period profit margin π_P(r_i, k) = (α − r_i) − (ρ − α)k in every subsequent period. When banks place a substantial weight δ on the future, this reduction in future charter value V_P weakens their incentive to preserve their banking license. If δ is close to 1 or the cost of equity ρ is high (or if θγ > α), raising capital requirements lowers r̂(k) and pushes banks toward gambling.

Deposit-Market Stealing and Symmetric Competition

Even when ∂r̂(k)/∂k > 0, Hellmann, Murdock, and Stiglitz demonstrate a second failure of capital-only regulation in a competitive deposit market: the deposit-market-stealing effect. Suppose rival banks offer the prudent equilibrium deposit rate r*. A prudent bank pays its promised deposit rate r_i in all states of nature (probability 1), so its marginal cost of raising r_i is 1 · D. By contrast, a gambling bank pays its promised deposit rate r_i only when the gamble succeeds (probability θ < 1); when the gamble fails (probability 1 − θ), the deposit insurance fund pays depositors. Thus, the effective marginal cost to a gambling bank of bidding up the deposit rate is only θ · D < D.

Because gambling lowers a bank's marginal cost of funds, a bank planning to gamble has a stronger incentive to bid up deposit rates to steal market share D(r_i, r*) than a prudent bank. To prevent banks from bidding r_i into the gambling zone, a regulator relying on capital alone must impose a high minimum capital requirement k₁ > k₀ that deters the joint deviation of bidding up r_i to steal deposits and gambling.

Proof of the Pareto Superiority of Combining Deposit-Rate Ceilings with Capital Regulation

Let k₀ denote the minimum capital requirement needed to satisfy the No-Gambling Condition at a regulated deposit rate r̄, and let k₁ > k₀ denote the higher capital requirement required to prevent deposit-market stealing under free deposit pricing. Replacing the capital-only regime (k₁, unconstrained r) with a joint regime combining a lower capital requirement k ≥ k₀ and a binding deposit-rate ceiling r_i ≤ r̄ creates a Pareto improvement:

1. Eliminating the Gambling Deviation Bonus: A binding ceiling r_i ≤ r̄ prevents a rogue bank from bidding r_i > r̄ to steal deposits, removing the need to over-capitalize banks to k₁.
2. Restoring Franchise Value via Intermediation Rents: Capping deposit rates at r̄ < α guarantees prudent banks a positive spread (α − r̄) > 0, boosting charter value V_P = δπ_P/(1 − δ).
3. Sharing Saved Capital Costs with Depositors: Reducing required capital from k₁ to k′ ∈ (k₀, k₁) saves deadweight equity costs of (ρ − α)(k₁ − k′). By setting the ceiling r̄ slightly above the depressed competitive deposit rate under k₁, the regulator gives depositors a higher deposit interest rate and banks higher franchise value.

Regulatory Forbearance and Dynamic Consistency

Hellmann, Murdock, and Stiglitz also analyze regulatory forbearance, where supervisors cannot commit ex ante to closing an insolvent bank. The franchise-value deterrent [(1 − θ)δ/(1 − δ)]π_P works only if the bank loses its charter upon failure. If regulators forbear by leaving an insolvent bank open with probability φ > 0, the effective punishment shrinks from (1 − θ)V_P to (1 − θ)(1 − φ)V_P. As φ → 1, high franchise value ceases to discipline risk-taking because the bank retains future charter rents regardless of whether its gamble fails. Thus, franchise-value discipline requires prompt corrective action alongside margin preservation.

Imperfect Spatial Competition and Repullo's (2004) Salop Circle Formulation

In 2004, Rafael Repullo embedded the Hellmann–Murdock–Stiglitz mechanism inside a Salop circular-city model of spatial banking competition in the Journal of Financial Intermediation. Solving for the Nash equilibrium deposit rate r*(k, c) as a function of switching cost c and capital requirement k, Repullo showed that banks pass the cost of higher capital requirements onto depositors via lower equilibrium deposit rates ∂r*/∂k < 0, and proved that risk-based capital requirements (k_G > k_P) achieve the prudent allocation at lower social cost than flat capital ratios.

Why can raising bank capital requirements increase risk-taking in the Hellmann–Murdock–Stiglitz model?

Raising minimum capital requirements k has two opposing effects. The static capital-at-risk effect forces shareholders to lose more equity in the current period if a gamble fails. However, because equity is more expensive than deposits (ρ > α), holding higher capital k in every future period reduces future net profits π_P(r_i, k) and shrinks franchise value V_P = δπ_P/(1 − δ), which can outweigh the static effect when ρ and δ are high.

What is the "deposit-market-stealing effect" in banking competition?

A prudent bank repays depositors with probability 1, whereas a gambling bank repays depositors only when its gamble succeeds (probability θ < 1) and defaults onto the deposit insurance fund when it fails (probability 1 − θ). Consequently, a bank intending to gamble faces a lower marginal cost of raising deposit rates (θ < 1) and has an incentive to outbid prudent rivals to steal deposit market share.

How do deposit-rate ceilings achieve a Pareto improvement over capital requirements alone?

Preventing deposit-market stealing with capital alone requires a high capital ratio k₁, inflicting a deadweight cost (ρ − α)k₁ that banks pass on to households via lower deposit rates. A binding deposit-rate ceiling r̄ blocks deposit-market stealing directly and preserves franchise value, allowing a lower capital requirement k′ < k₁ and a rate ceiling r̄ above the depressed rate under k₁.

How does the Hellmann–Murdock–Stiglitz model relate to Keeley's (1990) charter-value hypothesis?

Michael Keeley (1990) showed empirically that U.S. deregulation increased competition, lowered bank Tobin's q (charter value), and raised default risk. Hellmann, Murdock, and Stiglitz (2000) formalized Keeley's insight in a dynamic strategic model with endogenous deposit-rate competition and proved that flat capital ratios cannot substitute for lost charter value without incurring Pareto-inefficient capital costs.

1. Hellmann, T. F., Murdock, K. C., & Stiglitz, J. E. (2000). "Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough?" American Economic Review, 90(1), 147–165. https://doi.org/10.1257/aer.90.1.147
2. Keeley, M. C. (1990). "Deposit Insurance, Risk, and Market Power in Banking." American Economic Review, 80(5), 1183–1200. https://www.jstor.org/stable/2006769
3. Repullo, R. (2004). "Capital Requirements, Market Power, and Risk-Taking in Banking." Journal of Financial Intermediation, 13(2), 156–182. https://doi.org/10.1016/j.jfi.2003.08.005
4. Matutes, C., & Vives, X. (2000). "Imperfect Competition, Risk Taking, and Regulation in Banking." European Economic Review, 44(1), 1–34. https://doi.org/10.1016/S0014-2921(98)00057-900057-9)
5. Morrison, A. D., & White, L. (2005). "Crises and Capital Requirements in Banking." American Economic Review, 95(5), 1548–1572. https://doi.org/10.1257/000282805775014254