Comment on "Crises in Competitive versus Monopolistic Banking Systems"

This paper (Boyd, De Nicoló, and Smith 2004, this issue of JMCB) is intended as a contribution to the now-burgeoning literature on the causes and effects of banking crises. This literature has proven to be rich in theoretical and empirical content and also has potential policy relevance (see, for example, Allen and Gale 2000). Although the term "banking crisis" has a distinctly different connotation from "bank failure"-in that the former refers to a systemic panic and the latter to an individual bank failure-these terms have often been used interchangeably in the literature in the context of "representative bank" models.¹ This paper too belongs to the genre of representative bank models.

The paper poses two broad questions: (1) Is the probability of a banking crisis higher in a competitive or monopolistic banking system and how does the inflation rate affect these probabilities? (2) What are the probabilities of real output losses in competitive and monopolistic banking systems?

To address these questions, the paper develops a standard intertemporal risk-sharing model with risk-averse depositors, who face a relocation/preference shock, to rationalize banks as insurance providers in a monetary, general equilibrium setting. Depositors have no alternatives to bank deposits in terms of their investment choices. Banks receive deposits that they invest in cash reserves and a "storage asset" with a positive return. A banking crisis occurs when deposit withdrawals are so high that banks exhaust their reserve assets. An interesting twist in this result is that a banking crisis need not be associated with liquidation of the storage asset, and liquidation, which is associated with real resource losses, will occur only for sufficiently high withdrawals. That is, optimizing banks do not begin to liquidate their storage assets immediately when confronted with an exhaustion of their reserves.

On the issue of how bank industry structure affects the probability of a banking crisis, the paper finds that if the nominal interest rate (rate of inflation) is below some threshold, a monopolistic banking system will be associated with a higher probability of a crisis than a competitive banking system. The basic intuition is that a monopoly bank earns higher expected profits with the storage technology than by liquidating it. A competitive bank, which competes with other banks to maximize depositor utility, earns zero expected profit, so it faces a lower opportunity cost by holding reserves as opposed to the storage asset, relative to a monopoly bank. Consequently, a competitive bank holds larger reserves to enhance depositor insurance than does a monopoly bank, and faces a lower crisis probability.

For nominal interest rates above the (endogenously-determined) threshold, a monopolistic banking system has a lower probability of a crisis than does a competitive banking system. The intuition is as follows. A monopoly bank holds smaller reserves and pays a lower interest rate than a competitive bank. This implies that the probability that reserves will be exhausted depends on the interplay between these two forces-the level of reserves and the level of the interest rate. The interest rate effect dominates when the inflation rate is high, so that in this case the monopoly bank has a lower crisis probability.

Moreover, an increase in the inflation rate always increases the crisis probability. The reason is that as inflation increases, so does the opportunity cost of holding reserves relative to investing in the storage technology and hence there is a concomitant decline in the reserves banks keep.

On the question of the output losses associated with banking crises, the paper finds that, conditional on an output loss occurring, expected losses under competition are higher than those under monopoly. The intuition is that as long as the reserve ratio under competition is higher than under monopoly, output losses under competition occur only at higher withdrawal levels, and hence the conditional expected loss is higher under competition.

My overall assessment is that this is an interesting and elegant paper. It is also quite well exposited, particularly the Introduction where...