FORUM
Deposit Insurance
Deniz Anginer and Ata Can Bertay As with other financial safety net measures, there is a natural economic trade-off associated with deposit Deposit Insurance Design and insurance. While it can enhance depositor confidence Institutional Environment and reduce the likelihood of bank runs during crisis periods, deposit insurance can also increase moral haz- ard and make financial systems more vulnerable to cri- ses during good times. From a public policy perspec- tive, it is essential to know the factors and design features that will enhance the stabilization effects of Deniz Anginer Deposit insurance is a widely used and integral part of deposit insurance while reducing the inevitable adverse The World Bank. the financial safety net provided by states across the effects. Recent literature suggests that deposit insur- globe. According to the Bank Regulation and Supervi- ance design and implementation can affect how well sion Survey (BRSS) conducted by the World Bank, over deposit insurance schemes perform in practice (see 107 countries have some form of explicit deposit insur- Anginer and Demirgüç-Kunt 2018 for a literature ance scheme in place as of 2016. This number has review). For instance, limiting coverage and scope and increased substantially from 93 in the year 2013. implementing risk-based pricing can help to alleviate During and after the global financial crisis (GFC), moral hazard problems and to internalize banks’ some countries introduced new deposit insurance risk-taking. schemes and others extended the scope and coverage The recent research also emphasizes the role that of their existing schemes to restore confidence in their the larger institutional environment plays in how effec- Ata Can Bertay banking systems. For instance, Australia and Singapore tive deposit insurance schemes are in practice as well The World Bank. introduced explicit deposit insurance to their banking as specific design features that are implemented. In systems for the first time, whereas Spain and the US particular, the research suggests that it is vital for coun- increased the limit on the amounts that are covered by tries to cultivate an environment that provides the right deposit insurance. Other countries increased the scope set of incentives for supervisors and regulators on the of securities and bank liabilities guaranteed. Most one hand, and private market participants (such as notably, Ireland extended deposit insurance to most large uninsured depositors, shareholders, and other bank liabilities, essentially offering a blanket guarantee creditors), on the other, to monitor the banks they on bonds, subordinated debt, and interbank deposits. invest in. Thus, strong institutions and the rule of law The significant expansion of explicit deposit insurance can be crucial for effective public and private monitor- during the crisis rekindled the debate about the effi- ing. In this short article, we discuss how the larger insti- cacy of deposit insurance schemes and the inevitable tutional environment affects the design, adoption, and moral hazard problems associated with providing state performance of deposit insurance schemes using the guarantees. results from the recent Bank Regulation and Supervi- A vast empirical literature established that deposit sion Survey (BRSS) conducted by the World Bank. insurance brings economic benefits by ensuring depos- In particular, we categorize economies into two itor confidence and preventing bank runs. At the same groups using a composite measure of institutional time, deposit insurance also comes with the unin- quality calculated as the average estimated index of six tended consequence of encouraging banks to take on indicators drawn from the World Governance Indica- excessive risk. This standard moral hazard problem tors. These capture various dimensions of institutional arises because deposit insurance distorts incentives quality such as accountability, political stability, gov- for bank managers, shareholders, and depositors. Bank ernment effectiveness, regulatory quality, rule of law, managers and shareholders are incentivized to take on and control of corruption. We compute the average higher risk, as they privately capture the upside returns institutional quality on a rolling basis for the years but do not internalize downside losses, which are 2005, 2010, and 2016, thus including both the pre- and socialized through the deposit insurance fund. By lim- post-GFC periods. Table 1 provides a list of countries iting downside risk, deposit insurance naturally incen- that are covered in the analyses. We classify countries tivizes greater risk-taking. Depositors also have less of as having high (above median) institutional quality if an incentive to be careful in the initial selection of their their composite institutional quality score is above the bank and monitoring its financial condition, as they are median of all countries in a given year. Likewise, coun- protected against losses when there is a bank failure. tries are classified as having low (below median) insti-
ifo DICE Report I/ 2019 Volume 17 3 FORUM
Figure 1 and economically developed countries, Dewenter, Hess, and U Brogaard (2018) examine how o e me i n in titution levels of economic freedom, elo me i n in titution rule of law, and corruption in 90 2 2 a given bank’s home country 7 0 7 7 affect moral hazard. Even in a 70 set of institutionally compara- 0 ble countries, the authors find 0 that in most cases, better insti- tutions help mitigate problems 0 associated with deposit insur- 0 ance. Focusing on developing 20 countries, Cull, Senbet, and 10 Sorge (2004) show that in weak 0 institutional environments, 200 2010 201 deposit insurance reduces ote e figure pre ent t e per ent ge of ountrie it n e pli it epo it in ur n e prote tion tem for n Sour e ut or l ul tion from RSS n I 2019 ifo In titute economic growth and financial development. More importantly, empiri- tutional quality if their composite score is below the cal evidence suggests that weak institutional environ- median. Countries highlighted in bold in Table A1 are ments can prevent optimal deposit insurance design. developing countries, indicating that income groups In particular, the rule of law and other private and pub- (i.e., high-income vs. developing countries) are not fully lic contracting environment features proved important capturing the institutional quality differences. in deposit insurance adoption and design (Demirgüç- Figure 1 shows how the explicit deposit insurance Kunt and Kane 2002; Hovakimian, Kane, and Laeven coverage evolved during the last decade for these two 2003). These, in turn, impact how well deposit insur- sets of countries. Explicit deposit insurance was quite ance schemes function in a given country. Key design extensive even before the GFC: 78% of countries with features are credible limited coverage, co-insurance, high-quality institutions had it in 2005, compared to and risk-based pricing. 63% of countries with low-quality institutions. After the Co-insurance systems, in which deposit insurance GFC, explicit deposit insurance became more common covers less than 100 percent of a depositor’s account across the world, and the adoption rate in low institu- balance, are one way to incentivize depositors to mon- tional quality countries almost caught up with that of itor banks and make more prudent bank choices in the high-quality institutions group in 2016. their deposit decision. Over the past decade, co-insur- Cross-country analyses of deposit insurance ance systems have been largely removed as it is now schemes show that in settings with low institutional believed that partial payments in the event of bank fail- quality, deposit insurance can be destabilizing and can ures can increase the likelihood of bank runs. Co-insur- have adverse consequences for market discipline. ance as a design element declined in both the high and Focusing on the rule of law plus the supervision and low institutional quality countries. In particular, the strength of the legal system, Demirgüç-Kunt and Detra- percentage of countries with high-quality institutions giache (2002) examine how various measures of institu- using some form of co-insurance was 38 percent in tional quality affect how well deposit insurance works 2005, and this percentage declined to eight percent by in different countries. They find that, on average, the 2016. In low institutional quality countries, the per- existence of explicit deposit insurance increases the centage likewise declined from 39 percent in 2005 to six probability that a country will experience a banking cri- percent in 2016. sis. However, using the institutional quality measures Charging banks risk-adjusted premiums for mentioned above, they find that the probability that deposit insurance coverage is another way to alleviate deposit insurance will result in a crisis is significantly moral hazard problems. The premiums charged to lower in countries with higher levels of institutional banks can either be a flat fee, or they can be based on quality. the risk a bank poses to the deposit insurance fund. Angkinand (2009) and Angkinand and Wihlborg Under such a system, banks with higher asset or loan (2010) analyze the impact of institutional variables such risk (and thus more likely to fail) would be charged as the rule of law, corruption, and shareholder rights on higher insurance fees. Risk-based pricing can help the relationship between deposit insurance and finan- internalize the cost of risk-taking by bank managers cial stability. The authors find that institutional envi- and shareholders, which in turn would curb the exces- ronments that incentivize effective public and private sive risk-taking that results from moral hazard. monitoring can alleviate moral hazard effects associ- Although there are issues related to figuring out the ated with deposit insurance. Focusing on financially actuarially fair value of fees, the empirical evidence
4 ifo DICE Report I/ 2019 Spring Volume 17 FORUM
shows that risk-adjusted pre- Figure 2 miums perform better than U R flat-rate premiums in reducing bank risk (Demirgüç-Kunt and o e me i n in titution elo me i n in titution Detragiache 2002; Hovakimian, Kane, and Laeven 2003). 90 Risk-based pricing was ini- 0 tially pioneered in the US in the 70 early 1990s and quickly spread 0 to other countries. In 1997, only 0 four countries (Finland, Peru, 0 Sweden, and the US) used risk- 0 based pricing for deposit insur- 20 ance fees (Demirgüç-Kunt and 10 Huizinga, 1999). As of 2016, this 0 number has increased to 55. 200 2010 201 ote e figure pre ent ountrie in i epo it in ur n e fee /premium rge to n r e on ome Figure 2 shows that use of risk- e ment of ri based premiums for deposit Sour e ut or l ul tion from RSS n I 2019 ifo In titute insurance in high institutional quality countries has increased Figure 3 substantially in recent years. F As of 2016, 83 percent of coun- tries in this group reported umul te fun o er tot l in ure epo it 10 charging premiums based on 9 risk. Although there has been an increase in the low institu- 7 tional quality group, it is still well below the high institu- tional quality countries: only 38 percent in 2016. Implementing credible 2 limited coverage ex ante is 1 another crucial design fac- 0 tor for deposit insurance to 1 1 0 0 0 1 1 2 work effectively. In theoretical In titution l u lit models of deposit insurance, ote e figure pre ent o t e r tio of umul te fun to tot l in ure epo it i orrel te it t e in titution l u lit in e in 201 bank runs happen as a result Sour e ut or l ul tion from RSS n I 2019 ifo In titute of self-fulfilling phenomena (see, for instance, Diamond and Dybvig 1983 and extensions). Lack of confidence The recent experiences with deposit insurance in in the banks causes investors to rush to be the first in Cyprus and Iceland illustrate the importance of ade- line to withdraw their funds. If depositors believe that quate funding for deposit insurance for it to be credi- other investors will not run, then only investors with ble. In a sense, all insurance schemes are underfunded, real liquidity needs withdraw their funds. The bank as it is impossible to have funds in place to fully cover can meet these demands without costly liquidation of all potential losses of depositors. Yet depositors expect assets. Nevertheless, if everyone believes that a run the government to step in during a crisis and provide a will occur, then it becomes a self-fulfilling prophecy as full backstop. However, this type of intervention depositors run to avoid being last in line. The bank is requires the government to have the political will—and then forced to liquidate its long-term assets in a costly more importantly, the economic resources—to do so. way. This results in unnecessary economic losses as an In countries where the institutions have deteriorating otherwise solvent bank is forced to liquidate. In these and poorly governed finances, intervention is not models, the effectiveness of deposit insurance relies always a viable option, and underfunding can be a real heavily on depositors’ confidence that the insurance is possibility. These countries tend to also suffer from credible. Even if there is a small chance that the deposit political instability, and it may be challenging to bring insurance scheme will run out of funds, then it is different stakeholders together to agree on providing rational for depositors to run to the bank and withdraw funds to a dispersed group of depositors. their funds. Thus, deposit insurance schemes must be In theoretical models, the economic cost of credible ex ante in order to stop contagious runs (Bon- deposit insurance is zero, since deposit insurance elim- fim and Santos 2017; Calomiris and Powell 2001). inates an equilibrium in which everyone runs. If deposit
ifo DICE Report I/ 2019 Volume 17 5 FORUM
insurance is credible and depositors do not run, then ance funding ratio in the univariate analysis. We also taxes do not have to be imposed ex ante to fund the find that deposit insurance coverage indexation (with deposit insurance scheme. However, as credibility can respect to, for example, prices or per capita GDP) is be an issue in low institutional quality countries, much more common in low institutional quality coun- deposit insurance schemes have to be sufficiently tries. In 2016, 44 percent of the countries in the low funded to assure depositors that there will be resources institutional quality group had some form of indexa- available to cover the losses should their bank fail. tion, up from 11 percent in 2010. The percentage of Accumulating funds to assure this confidence can be countries in the high institutional quality group that highly costly, but it is necessary in low institutional had indexation was only 16 percent in 2016, up from ten quality countries. Consistent with this notion, the percent in 2010. This observation also supports the empirical evidence from the BRSS survey shows that idea that low institutional quality countries are trying the size of accumulated funds with respect to total to keep their deposit insurance coverage credible by insured deposits is negatively related to institutional automatically adjusting the coverage in response to quality. Figure 3 shows the relationship between the higher inflation or per capita income. insurance funding ratio (accumulated funds divided by Although adequate funding of insurance schemes total insured deposits) and institutional quality. We see is important for deposit insurance to be credible, dur- that low institutional quality countries tend to accumu- ing the GFC, many countries substantially expanded late more funds ex ante, possibly to build credibility. In both the scope and the coverage of deposit insurance particular, a one standard deviation increase in institu- in order to restore stability in their banking sectors. tional quality (0.81 points increase in the index) is Setting clear and limited commitments ex ante is just related to a 1.3 percent reduction in the deposit insur- as crucial as credibility for deposit insurance to work effectively. Expanding cover-
Figure 4 age beyond what was prom- ised to depositors during the U F crisis had the effect of reinforc- 201 2010 ing market expectations that the government will step in to bail out banks and depositors should the need arise. These o e me i n in titution types of expansions reduce 21 market discipline and can lead to greater risk-taking by banks. Consistent with this notion, a number of papers have shown elo me i n in titution that more generous deposit 27 insurance coverage and scope result in greater moral hazard 0 10 20 0 0 0 0 70 (Honohan and Klingebiel 2000; ote e figure pre ent ountrie in i t e epo it in ur n e fun i u e for purpo e ot er t n epo itor Demirgüç-Kunt and Detragi- prote tion ache 2002). Sour e ut or l ul tion from RSS n I 2019 ifo In titute Moreover, limited ex ante Figure 5 commitment by governments
also reduces the costs that arise from providing insurance o e me i n in titution during times of distress. As elo me i n in titution bank runs often coincide with 0 7 deteriorating economic condi- 70 tions and declining asset val- 0 ues, ex post expansion of guar- antees can be very costly for 0 taxpayers (Allen, Babus, and 0 Carletti 2009). Since fiscal 0 costs are limited, ex ante com- 1 20 mitment not to expand insur- 10 ance can improve the reliabil- 0 ity and credibility of deposit E p n ion of o er ge In re e in mount insurance schemes. Limited ote e figure pre ent ountrie intro u ing nge to t eir epo it prote tion tem re ult of t e 2007 2009 glo l fin n i l ri i commitment also ensures that Sour e ut or l ul tion from RSS n I 2019 ifo In titute deposit insurance schemes are
6 ifo DICE Report I/ 2019 Spring Volume 17 FORUM
Table 1 harmonized across countries. This approach levels the List of Countries (Median Institutional Quality playing field across different countries and helps to Is between Bulgaria and South Africa) reduce regulatory arbitrage whereby investors move funds to countries where they expect the local author- Above median institutions Below median institutions ities to increase coverage during times of stress. Australia Angola Despite the benefits of limited commitment, dur- Austria Argentina ing the GFC, there was a significant expansion of Belgium Armenia deposit insurance in both scope and coverage. As of Bhutan Bahrain 2016, around one-fourth of high institutional quality Botswana Bangladesh countries and one-third of low institutional quality Canada Belarus countries reported compensating deposits that were Cayman Islands Belize not explicitly covered at the time of a bank failure. Moreover, deposit insurance funds have also been used Chile Bosnia and Herzegovina for purposes other than covering specific depositor Costa Rica Brazil losses. Figure 4 shows the percentage of countries in Croatia Bulgaria each institutional quality group in which depositor Cyprus Burundi funds were used for other purposes, such as liquidity Denmark Colombia support, bank resolution, or recapitalization of weak Estonia Dominican Republic banks. In 2010, 27 percent of countries in the low-qual- Finland El Salvador ity institutions group used deposit insurance funds for France Fiji other purposes, compared to just 21 percent of coun- Germany Ghana tries in the high-quality institutions group. However, Hong Kong SAR, China Greece after the crisis, a higher percentage of countries in the
Hungary Guatemala high-quality institutions group used deposit insurance Iceland Guyana funds for other purposes—65 percent compared to Ireland Honduras 35 percent of countries in the low-quality institutions group. Israel India Most of these changes came during the finan- Italy Indonesia cial crisis. Figure 5 shows the percentage of coun- Jersey Jordan tries in each institutional quality group that have Korea, Rep. Kenya made changes to their deposit insurance schemes in Latvia Kyrgyz Republic response to the GFC. Most countries, especially those Liechtenstein Lebanon in the high-quality institutions group, significantly Lithuania Lesotho increased both the limit and the type of accounts cov- Luxembourg Malawi ered under deposit insurance. Specifically, 73 percent Macao SAR, China Maldives of the countries in the high-quality institutions group Malaysia Mexico increased the coverage amount. In the US, for exam- Malta Moldova ple, the guaranteed limit (per depositor, per bank) was Mauritius Morocco increased from USD 100,000 to USD 250,000 in 2008 to restore confidence in the banking system at the height Netherlands Mozambique of the financial crisis. Of the countries in this group, 43 New Zealand Nicaragua percent also increased the type of liabilities covered Norway Nigeria by deposit insurance. In Ireland, deposit insurance was Oman Pakistan expanded to cover all bank liabilities. There was also Poland Panama significant expansion in low institutional quality coun- Portugal Peru tries: 18 percent expanded the scope, and 55 percent Romania Philippines increased the amount covered by deposit insurance. Seychelles Russian Federation Although it is difficult to quantify the long-term effects Slovak Republic Sri Lanka of these expansions, they will nonetheless have an Slovenia Suriname adverse impact on market discipline in the future. South Africa Tajikistan In this article, we have shared some empirical Spain Tanzania snippets from the latest BRSS survey. Overall, the results in the survey reinforce the importance of the Switzerland Thailand larger institutional environment in how well deposit Taiwan, China Trinidad and Tobago insurance schemes are designed and function. It is United Kingdom Uganda important to emphasize that poorly designed schemes United States Vanuatu in lower-quality institutional environments can Uruguay Zimbabwe increase the likelihood of a banking crisis. Thus, it is Source: WGI (2019). possible for explicit deposit insurance to do more
ifo DICE Report I/ 2019 Volume 17 7 FORUM
harm than good for financial stability in countries with such environments.
REFERENCES
Anginer, D. and A. Demirgüç-Kunt (2018), “Bank runs and moral hazard: A review of deposit insurance”, The World Bank. Angkinand, A. P. (2009), “Banking regulation and the output cost of banking crises”, Journal of International Financial Markets, Institutions and Money, 19(2), 240–257. Angkinand, A. and C. Wihlborg (2010), “Deposit insurance coverage, ownership, and banks’ risk-taking in emerging markets”, Journal of International Money and Finance, 29(2), 252–274. Bank Regulation and Supervision Survey (2019). World Bank. https:// www.worldbank.org/en/research/brief/BRSS. Bonfim, D. and J. A. C. Santos (2017), “The importance of deposit insur- ance credibility”, mimeo. Calomiris, C. W. and A. Powell (2001), “Can emerging market bank regu- lators establish credible discipline? The case of Argentina”, 1992–99. In Prudential supervision: What works and what doesn’t (147–196). Uni- versity of Chicago Press. Cull, R., M. Sorge and L. W. Senbet (2004), “Deposit insurance and bank intermediation in the long run”, BIS Working Paper No. 156. Demirgüç-Kunt, A. and E. Detragiache (2002), “Does deposit insurance increase banking system stability? An empirical investigation”, Journal of Monetary Economics, 49(7), 1373–1406. Demirgüç-Kunt, A. and E. J. Kane (2002), “Deposit insurance around the globe: Where does it work?”, Journal of Economic Perspectives, 16(2), 175–195. Dewenter, K. L., A. C. Hess and J. Brogaard (2018), “Institutions and deposit insurance: Empirical evidence”, Journal of Financial Services Research, 54(3), 269–292. Diamond, D. W. and P.H. Dybvig (1983), “Bank runs, deposit insurance, and liquidity”, Journal of Political Economy, 91(3), 401–419. Honohan, P. and D. Klingebiel (2000), “Controlling the fiscal costs of banking crises”, World Bank Policy Research Working Paper Series No. 2441. Hovakimian, A., E.J. Kane and L. Laeven (2003), “How country and safe- ty-net characteristics affect bank risk-shifting”, Journal of Financial Services Research, 23(3), 177–204. World Governance Indicators (2019), World Bank. http://info.world- bank.org/governance/wgi/#home.
8 ifo DICE Report I/ 2019 Spring Volume 17