Deposit insurance and banking crises in the short and long run.

• Author: Chu, Kam Hon

There has been a rising global trend for countries, especially in the emerging economies, to institute explicit deposit insurance schemes in the last two decades. During this period, financial markets in the world have been frequently plagued by instabilities and banking crises, notably the Mexican crisis in 1995 and the Asian financial crisis in 1997, not to mention the recent meltdown of Argentina. Among the ninny arguments in favor of deposit insurance, protection of small depositors and prevention of systemic banking crises are the ones most often put forward by regulators to rationalize deposit insurance from a public-interest perspective. (1) Nonetheless, the public-interest argument based on protection of small depositors cannot adequately justify deposit insurance because there are alternatives such as short-term treasury securities (Benston and Kaufman 1988: 65), checkable money market mutual funds (Cowen and Krozner 1990), and government savings bonds (Chu 2000) that can achieve the same goals at lower cost.

The justification of deposit insurance therefore rests to a large extent on its effectiveness in averting systemic banking crises and contagious bank runs due to asymmetric information (Diamond and Dybvig 1983). That public fear is widespread, even though several studies have clearly demonstrated that the contagion argument is exaggerated (Benston and Kaufman 1995; Calomiris and Mason 1997; Kaufman 1994, 2000), and that asymmetric information does not necessarily lead to bank runs because banks have incentives to signal their quality (Chu 1999). As Kaufman (2000) summarizes, the evidence for the United States strongly suggests that contagious bank runs are neither widespread nor long lasting, and there is no evidence that a bank run drives a solvent bank into insolvency. (2) Nonetheless, many countries have established deposit insurance during or after banking crises or financial instabilities, hoping to restore stability and prevent future crises. A well-known example is the Federal Deposit Insurance Corporation, which was set up after the United States experienced massive bank failures during the Great Depression. More recent examples include those East Asian countries, such as Malayasia and Indonesia, hit hard by the Asian financial crisis. Indeed, financial crises can be extremely costly. Although the cost of restructuring the banking industry varies from country to country, ranging from 4.3 percent to 45 percent of GDP (Dziobek and Pazarbasioglu 1997), its distribution appears to skew toward the high-cost end. (3) These high-cost figures tend to justify, at least on the surface, the existence of a financial safety net such as deposit insurance.

However, are countries entirely immune from banking crises after instituting deposit insurance schemes? The answer is definitely no, as evidenced by the notorious U.S. saving-and-loans debacle in the 1980s (Kane 1989), not to mention similar incidents in other countries like Canada (Carr, Mathewson, and Quigley 1995). In evaluating feasibility of deposit insurance, therefore, a relevant question is, does deposit insurance reduce the likelihood of a banking crisis? If so, deposit insurance is justifiable because the expected benefits from avoiding substantial welfare or output losses due to a severe banking crisis are likely to outweigh the total cost of deposit insurance. Against that backdrop, this paper compares the banking stabilities of 174 countries during the 1980-2000 period to examine whether banking crises are less likely to occur in countries with deposit insurance than in those without. The empirical approach and findings of this study shed light on the relation between deposit insurance and banking crises over time.

Some empirical studies have addressed the issue of whether deposit insurance undermines or promotes banking stability. These include individual country studies such as those by Keeley (1990), Grossman (1992), Cebula and Belton (1997) for the United States, and by Carr, Mathewson, and Quigley (1995) for Canada. In brief, these studies provide evidence that deposit insurance tends to cause banking instability because of the moral hazard problem that induces depository institutions toward excessive risk taking. On a global scale, Demirguc-Kunt and Detragiache (2002) have recently used a panel of 61 countries for 1980-97 to examine post-deposit insurance banking stabilities. They find that deposit insurance does tend to induce banking crises, particularly among countries with weak institutional environments.

The main finding of these empirical studies is reinforced by my own findings in this study, albeit different approaches and methods of investigation are adopted. Although the shared objective is to determine whether deposit insurance promotes or reduces banking stability, I allow for the possibility that the likelihood of a banking crisis after the introduction of deposit insurance may change over time. To address that possibility, I employ contingency table analysis to analyse the data. The findings suggest that deposit insurance promotes shortrun banking stability but induces long-run instability. Those results have significant policy implications for banking regulation and reform.

Exploring the Historical Record

Caprio and Klingebiel (1996, 1999, 2000) have recently compiled the major banking crises--systemic and borderline--in the world during the last three decades. Meanwhile, Garcia (2000), Demirguc-Kunt and Sobaci (2001), and Lee and Kwok (2000) have provided detailed information about deposit insurance around the world, including the years when they were set up. From the information provided by these studies, the experiences of 174 economies regarding deposit insurance and banking crises during the period 1980-2000 are compared using contingency table analysis to examine the effectiveness of deposit insurance in averting banking crises. (4) The intuition behind the empirical method is straightforward and similar to a typical statistical study in epidemiology. Countries are classified into two groups depending on whether they have established explicit deposit insurance schemes or not as of a particular year. The proportions of countries having systemic banking crises and borderline crises in subsequent years in each group are then compared to see whether they are statistically different. (5) Analogous to a statistical study in epidemiology, the absence of deposit insurance is the exposure status or risk factor and the existence of banking crisis is the disease status in this study.

Unlike the epidemiologist, however, the macroeconomist usually does not enjoy the privilege of conducting controlled experiments. Nevertheless, a prospective study, also known as a cohort study, is applied here: Imagine back in 1980 a "natural experiment" was conducted, in which sample countries were classified on the basis of the presence or absence of deposit insurance as of 1980, followed forward in time, and in each group the proportions of countries subsequently having major banking crises were recorded. (6) As the statistical analysis compares the proportions of countries having banking crises in each group, a country would be counted only once even though in reality banking crises recurred in some countries. To avoid double counting, a country is classified as having a systemic crisis if it had experienced both systemic and borderline crises during the period under study. It is noted that a total of 57 countries set up explicit deposit insurance schemes after 1980. The prospective study briefly described earlier considers only their pre-deposit insurance experiences with banking crisis. However, these countries' pre- and post-deposit insurance experiences with banking stability are also compared in another statistical analysis below.

For the prospective study, only 19 countries, including the United States and Canada, had set up deposit insurance schemes by 1980. In spite of deposit insurance, about two-thirds of them encountered either systemic or borderline banking crises in subsequent years. Similarly, out of the 155 countries without deposit insurance schemes at that time, 110 later experienced banking crises. In the former group, the proportions of countries with systemic crises and borderline crises are 0.42 and 0.26, respectively. Both figures are not statistically different from their respective counterparts of 0.50 and 0.21 in the latter group. Results of the 2 x 3 contingency table analysis are summarized in Table 1. In other words, countries with deposit insurance are equally likely to suffer crises in subsequent years when compared with countries without deposit insurance. (7) There is no evidence indicating that deposit insurance promotes banking stability.

To be sure...