The provision and design of deposit insurance systems presents governments with an unprecedented set of challenges. Deposit insurance systems are typically motivated by a desire to decrease the risk of systemic bank runs (e.g., Diamond and Dybvig (1983)) and to protect small, uninformed depositors (e.g., Dewatripont and Tirole (1994)). Often, however, they are blamed for increasing the incentives of banks to take excessive risk by reducing, or even completely eliminating, the incentives of depositors to monitor and discipline their banks (e.g., Kane (1989) and Calomiris (1999)).
Market discipline by depositors is commonly understood as a situation in which depositors penalize riskier banks by requiring higher interest rates or by withdrawing their deposits (i.e., a leftward shift in the supply of deposits when bank risk increases). The main challenge policymakers are facing is how to design a deposit insurance scheme that protects the financial system from systemic bank runs without unduly reducing market discipline. To date, however, there is very little empirical evidence on the effects of implicit or explicit deposit insurance on market discipline, and there is even less evidence on how various design features of a deposit insurance scheme might affect market discipline. This paper investigates these questions in a natural experiment setting.
The analysis takes advantages of the introduction of an explicit deposit insurance system in Bolivia and compares the behavior of depositors before and after the introduction of this system. The comparison is between implicit, blanket guarantees and explicit, partial deposit insurance. The country specific circumstances during the sample period and the characteristics of the deposit insurance system are ideal for our purposes. First, the sample period, 1998 to 2003, is characterized by a recession that significantly weakened the banking sector, providing depositors with reasons to worry about the safety of their deposits a natural prerequisite for market discipline (Flannery and Sorescu (1996)). Second, apart from the introduction of a deposit insurance scheme in December 2001, there were no other major regulatory reforms during the sample period. This makes it possible to construct consistent time-series and to compare the behavior of depositors before and after the regime-change. Finally, a unique feature of the Bolivian deposit insurance system allows us to quantify the effect of the deposit insurance coverage rate on market discipline.
Previous studies provide more indirect tests. Demirgüç-Kunt and Huizinga (2004), for example, examine the effect of deposit insurance on market discipline by exploiting the cross-sectional variation of countries with or without explicit deposit insurance. Their results, however, could be attributed to other characteristics of the financial system that are correlated with the existence of a deposit insurance system. This problem becomes more serious when they examine how different design features of a deposit insurance system (e.g., coverage rate, co-insurance) affect market discipline. In particular, the format of their dataset forces them to examine each feature separately, making it difficult to assess the effect of each characteristic. Using within country variation, Martinez Peria and Schmukler (2001) examine the effect of deposit insurance on market discipline by comparing the behavior of small depositors that are fully insured with large depositors that are only partially insured. Finding that small depositors do not discipline their banks could be attributed to the effect of deposit insurance, but also to alternative hypotheses (e.g., small depositors are less sophisticated, have higher switching costs).
Both studies, however, have additional identification problems. To establish the existence of market discipline, one should show that bank risk is positively correlated with deposit interest rates and negatively correlated with the volume of deposits. Finding both effects is crucial in order to eliminate alternative, demand-driven, hypotheses (Park (1995)). Both studies, however, provide evidence either on interest rates or on deposits, but not both at the same time. In particular, Demirgüç-Kunt and Huizinga (2004) provide evidence only on interest rates. Instead, Martinez Peria and Schmukler (2001), who compare small and large depositors, have problems with both dimensions of their test: the interest rate series are not available by size of account and the results from deposits could be biased by the automatic migration of accounts across size categories.
Download
The Impact of Explicit Deposit Insurance on Market Discipline
