Diamond–Dybvig Bank Run Model
The Diamond–Dybvig model is a foundational economic model showing how commercial banks provide liquidity transformation while remaining vulnerable to self-fulfilling depositor runs.
The Diamond–Dybvig bank run model, developed by economists Douglas W. Diamond and Philip H. Dybvig in their landmark 1983 paper "Bank Runs, Deposit Insurance, and Liquidity," is the foundational economic theory of financial intermediation and banking panics. The model demonstrates how commercial banks perform essential maturity and liquidity transformation by pooling illiquid, high-return long-term investments while offering liquid, short-term demand deposits to risk-averse depositors.
Diamond and Dybvig proved that while this arrangement improves social welfare compared to autarkic financial markets, it inherently creates multiple Nash equilibria: an efficient equilibrium where only individuals with genuine consumption needs withdraw early, and a systemic panic equilibrium where fear of insolvency drives all depositors to run on the bank. Awarded the Nobel Memorial Prize in Economic Sciences in 2022 alongside Ben Bernanke, the model provides the primary theoretical rationale for government deposit insurance and central bank lender-of-last-resort facilities.
Theoretical Foundations and Model Setup
Analyzing the vulnerability of commercial banking requires modeling the fundamental friction between illiquid long-term production and uncertain short-term consumer liquidity needs. In uncoordinated financial markets, individuals who face sudden liquidity shocks must liquidate productive capital prematurely at heavy losses; financial intermediaries emerge specifically to pool idiosyncratic liquidity risks across depositors, transforming illiquid physical assets into liquid claims.
Agents, Endowments, and Liquidity Types
The model operates across three chronological periods: T = 0, 1, 2. The economy is populated by a continuum of identical, risk-averse depositors, each possessing an initial endowment of 1 unit of a consumption good at T = 0. Depositors have identical utility functions u(c), where u is twice continuously differentiable, strictly increasing, and strictly concave, satisfying relative risk aversion greater than 1: -c u''(c) / u'(c) > 1.
At T = 1, each depositor privately learns their consumer type:
- Type 1 (Early Consumers / Impatient): Occur with probability t (where 0 < t < 1). They derive utility only from consumption at T = 1: U = u(c1).
- Type 2 (Late Consumers / Patient): Occur with probability 1 - t. They derive utility from consumption at either period, but prefer T = 2: U = u(c1 + c2).
Because consumer types are private information, agents cannot trade insurance contracts against becoming early consumers directly in competitive markets.
Production Technology and Liquidity Transformation
The economy features a single productive investment technology. For each unit invested at T = 0:
- Liquidating prematurely at T = 1 yields a salvage recovery return of 1 (zero net return).
- Allowing the investment to mature fully at T = 2 yields a return of R > 1.
Under financial autarky (without banks), impatient agents liquidate their asset at T = 1 and consume c1 = 1, while patient agents hold until T = 2 and consume c2 = R.
A bank improves upon autarky by offering a demand deposit contract: depositors deposit their 1 unit at T = 0. The bank ensures a withdrawal value of r1 > 1 at T = 1 to anyone who requests it, financed by liquidating a fraction t r1 of the bank's long-term portfolio. The remaining assets mature at T = 2, yielding:
r2 = (1 - t r1) R / (1 - t)
Because depositors are risk-averse, the socially optimal contract sets 1 < r1 < r2 < R, providing liquidity insurance by smoothing consumption across states of nature.
Multiple Equilibria and Bank Runs
The presence of the demand deposit contract inherently creates coordination vulnerability through sequential service constraints.
The Sequential Service Constraint
Commercial banks service withdrawal requests sequentially on a first-come, first-served basis, as depositors queue at bank branches or digital portals. The bank cannot verify whether an individual withdrawing at T = 1 is genuinely an impatient consumer or a patient consumer claiming to be early.
If a patient agent observes or expects that a proportion f of total depositors will withdraw at T = 1 such that f r1 ≥ 1, the bank will be forced to liquidate its entire investment portfolio prematurely at T = 1. In that scenario, nothing remains for depositors at T = 2: r2 = 0.
The Run Equilibrium as a Coordination Failure
This structure generates two distinct Nash equilibria:
- Good (Efficient) Equilibrium: Patient depositors believe that only impatient depositors will withdraw at T = 1 (f = t). Under this belief, r2 > r1, meaning it is strictly optimal for patient depositors to leave their funds deposited until T = 2. The optimal allocation is sustained.
- Bad (Bank Run) Equilibrium: Patient depositors believe that other patient depositors will withdraw early (f > t). If a patient depositor waits, the bank will run out of assets and pay them 0 at T = 2. Consequently, their best response is to join the queue and withdraw promptly at T = 1 to secure r1 before assets are exhausted.
The bank run is largely self-fulfilling: panic alone transforms a fundamentally solvent institution into an insolvent failure.
Policy Interventions and Institutional Solutions
The Diamond-Dybvig framework provides a rigorous microeconomic foundation for institutional mechanisms designed to eliminate self-fulfilling panic equilibria. Financial regulators and central banks utilize several complementary policy tools to stabilize maturity-transforming intermediaries while preserving efficient intertemporal liquidity provision.
Suspension of Convertibility
The bank can adopt a policy of suspending cash convertibility: it ensures to pay r1 to early withdrawers only until total withdrawals reach the expected fraction t. Once t withdrawals occur, the bank closes its windows until T = 2.
Because patient depositors know that early withdrawals cannot exhaust the bank's portfolio, they have no incentive to panic, ruling out the run equilibrium. However, if the true aggregate fraction of impatient consumers t is uncertain and fluctuates stochastically, rigid suspension locks out genuine early consumers, imposing substantial welfare losses.
Government Deposit Insurance
The definitive solution demonstrated by Diamond and Dybvig is government-backed deposit insurance, funded through taxation.
A public deposit guarantee (such as the Federal Deposit Insurance Corporation in the United States) promises that all depositors will receive their full contractual entitlement regardless of when they withdraw. Because depositors are assured of their payouts, the incentive to run is eliminated. Furthermore, because the existence of insurance fully prevents panic, bank runs do not materialize in equilibrium, meaning the deposit insurer rarely needs to disburse funds.
Summary of Banking Regimes
| System Framework | Maturity Transformation | Liquidity Insurance | Vulnerability to Runs | Equilibrium Stability |
|---|---|---|---|---|
| Financial Autarky | None; individual storage | None (c1 = 1, c2 = R) | Zero risk of run | Unique, suboptimal equilibrium |
| Uninsured Bank | High (illiquid to liquid) | High (r1 > 1, r2 > r1) | Highly vulnerable | Multiple Nash equilibria |
| Suspension Policy | High | Partial (inflexible if t varies) | Low (prevents panic) | Unique, but inefficient under shocks |
| Insured Banking | High | Optimal and complete | Zero (runs eliminated) | Unique, Pareto-optimal equilibrium |
References
- Diamond, D. W., & Dybvig, P. H. (1983). Bank runs, deposit insurance, and liquidity. Journal of Political Economy, 91(3), 401–419.
- Goldstein, I., & Pauzner, A. (2005). Demand-deposit contracts and the probability of bank runs. The Journal of Finance, 60(3), 1293–1327.
- Postlewaite, A., & Vives, X. (1987). Bank runs as an equilibrium phenomenon. The Journal of Political Economy, 95(3), 485–491.