The implicit costs of government deposit insurance.

Author:Hogan, Thomas L.
  1. Introduction

    Few Americans give government-provided deposit insurance a second thought--and virtually no one calls for the abolition of the Federal Deposit Insurance Corporation (FDIC). To the extent that the FDIC is considered at all, it is politically popular. It is often credited with preventing bank runs like those of the Great Depression, and Congress has increased the maximum account balance covered by the FDIC seven times since its founding in 1933. With annual administrative and operating expenses averaging just a few cents per $100 insured, most people believe the benefits of government-provided deposit insurance exceed the costs.

    A growing literature, however, suggests that the benefits are often overstated while the costs are understated. In particular, historical studies such as Calomiris and White (1994), Curry and Shibut (2000), and Krosner and Melick (2008) question the value created by the FDIC. Hogan and Luther (2013) consider the explicit costs of the FDIC and conclude that they are significantly larger than those assumed in the standard benchmark model. We add to this literature by considering the implicit costs of FDIC insurance. Since such costs are routinely omitted, traditional cost-benefit analyses of the FDIC tend to be biased in favor of government-provided deposit insurance.

    At least two potentially significant costs are omitted in the standard benchmark model of deposit insurance. When governments provide deposit insurance, taxpayers bear an implicit cost. In times of crisis, they may be called upon to cover depositor claims in excess of the FDIC's Deposit Insurance Fund (DIF). Such a taxpayer backstop is costly. Although taxpayers are not called upon every year (and, under some systems, might never be called upon), taxpayers effectively hold a contingent claims contract. A reasonable measure of total annual expenses for FDIC insurance should include the implied cost of this contract.

    Costs also arise from the suboptimal pricing of deposit insurance. The assessment rates set by the FDIC can be described as suboptimal when the actual rate is either higher or lower than the rate predicted by actuarially fair models of deposit insurance. Since a higher-thanoptimal rate decreases intermediation and a lower-than-optimal rate encourages excessive risk-taking and increases financial fragility, borrowers, lenders, and their counterparties incur costs from suboptimal pricing. A reasonable measure of total annual expenses should include the implied costs that arise from suboptimal pricing.

    We argue that a proper cost-benefit analysis of governmentprovided deposit insurance must include the costs of implicit taxpayer guarantees and suboptimal pricing. The implicit costs of government-provided deposit insurance are real economic costs borne by taxpayers, borrowers, lenders, and counterparties. These implicit costs should be added to those traditionally recognized when engaging in cost-benefit analysis. Ignoring these costs biases the analysis in favor of government-provided deposit insurance. In what follows, we discuss Diamond and Dybvig (1983), which serves as a theoretical justification for government deposit insurance. Then, we consider the empirical evidence of the implicit costs described above but omitted from the model. We find sufficient evidence to conclude that these costs are nontrivial. (1)

  2. Theory of Deposit Insurance

    The Diamond-Dybvig (DD) model can be summarized as follows: Suppose N people each deposit 1 unit of goods into a bank. The bank invests its funds at some rate R > 1. Agents can either withdraw their funds after 1 period or 2. Those withdrawals in period 1 receive a stated rate of [r.sub.1] where 1 [r.sub.1], but when f is high, [r.sub.2]

    For simplicity, Diamond and Dybvig (1983) assume most agents are indifferent between consumption in periods 1 and 2. However, in period 1, some portion t of the agents becomes impatient and prefers to consume in that period. As long as t [r.sub.1]. However, if the number of impatient agents exceeds the tipping point (i.e., when t > f*), the payoff [r.sub.1] in period 1 becomes greater than the payoff [r.sub.2] in period 2. In this case, even the patient agents have an incentive to withdraw their funds in period 1. Since all agents--both patient and impatient--attempt to withdraw their funds in period 1, a bank run occurs.

    When a run occurs, the bank does not have sufficient capital to cover all withdrawals. All agents attempt to withdraw their funds simultaneously, and each agent withdraws [r.sub.1], > 1, making a total of N x [r.sub.1], in withdrawals. However, because the bank holds only N x 1 in deposits, there is not enough capital in the bank to pay [r.sub.1], to all agents. Some agents will be paid their return of [r.sub.1] while others will get nothing. Diamond and Dybvig (1983) assume a "sequential service constraint" such that agents arrive at the bank in random order and are paid according to their place in line until the bank is devoid of funds. Thus, the first 1 / [r.sub.1] agents receive a payment of [r.sub.1] while the remaining 1 - (1 / [r.sub.1]) agents receive a payment of 0.

    In the DD model, the ability or inability to prevent bank runs depends on what information is publicly available. If a bank has the ability to verify a customer's type as patient or impatient, then the bank can prevent runs by promising to pay only impatient agents in period 1. Similarly, if the bank knows the portion t of impatient agents in the population, then the bank can promise to pay out a maximum of t x [r.sub.1] in period 1. In this case, patient agents are guaranteed to receive [r.sub.2] > [r.sub.1] in period 2 and have no incentive to redeem their deposits in period 1, so no bank run occurs. Assured their deposits will be safe until period 2, patient depositors will never need to withdraw in period 1. The signal alone is enough to prevent runs at no cost to the bank.

    Problems arise when customer...

To continue reading