Ill a Survey of Deposit Insurance Practices

Ill A Survey of Deposit Insurance Practices

recent survey of 85 different systems of deposit The survey, whose results are presented in the Sta- Aprotection found that of the 85, 67 countries of- tistical Appendix tables, will be used to throw some fered an explicit, limited deposit insurance system in light on common practices. Later, the survey will normal times (see Table A1 of the Statistical Appen- also be used to examine the extent to which good dix).46 They are the focus of the survey that follows.47 practices have been adopted, and where they have As Table 2 shows, four of the surveyed countries are been disregarded.50 It finds that countries are in- in Africa, 10 are in Asia, 32 are in Europe, four are in creasingly adopting provisions that temper incentive the Middle East, and 17 are in the Americas. problems, but certain deficiencies remain in some instances. Year of Origin of Limited Deposit Insurance Systems The Deposit Insurance Agency: Role and Responsibilities Although two of the three systems in the United States (one for commercial banks and the second for There are basically two models for the role and re- savings associations) were started in the 1930s, it was sponsibilities of a deposit insurance agency. Under a not until the 1960s that other countries began to adopt narrow construction, the deposit insurance system's the deposit insurance systems that are still in exis- 48 obligation is to pay depositors of failed banks when tence. Eight schemes were initiated in the 1960s, and instructed to do so by the appropriate authority, nine in the 1970s. As the incidence of banking crises which is frequently the bank supervisor, and to ac- escalated in the 1980s, 19 schemes were initiated quire the funds by collecting premiums and building during the decade. Thirty new limited systems com- a fund or by imposing ex post assessments. The de- menced during the 1990s, as banking problems con- posit insurance agency in 34 countries plays such a tinued to escalate on all continents. (See Figure 2). narrow role. The alternative model for the agency is Revisions to deposit insurance systems have been much more comprehensive. The agency takes charge quite common, especially since the European Union of failed banks and resolves them according to the Directive in 1994.49 country's laws. The deposit insurance agency in 33 countries carries a broad range of responsibilities

46 that often includes anticipating bank problems and Seventy-two countries had systems that were explicitly de- resolving failed banks. The narrow construction fined in law and/or regulation. Full coverage was being offered in Spring 2000 in 10 of these countries. In seven of the full-coverage dominates in Europe, but broader responsibilities are cases, comprehensive coverage replaces systems that have limited common in Asia and the Western Hemisphere. scope in normal times. Six African countries that have not fully Moreover, a number of countries have recently ratified their agreement to form a regional insurance system and broadened the role for their agencies or they are con- the system in Panama are excluded from the survey because it ap- sidering enlarging those responsibilities. None is plies the guarantee only to credit cooperatives. Russia is also excluded because of a dearth of information. known to be considering reducing that role. 47The entries in the Tables A1 through A7 of the Statistical Ap- pendix extend and update into the second quarter of 2000 the survey results presented in Garcia (1999), which were exhaustively re- Membership viewed. Every effort has been made to include new schemes in the tables and to reflect revisions that have been made to existing schemes. The author requests the reader's forgiveness if any Table A1 of the Statistical Appendix shows that, to changes have been missed, because changes are frequent at present. avoid the problem of adverse selection, 62 of the 48Some states within the United States began a deposit insurance system earlier, as did the former Czechoslovakia. 49In addition to the revisions included in the Statistical Appendix, 50Issues relating to prompt corrective action, failure resolution, the United Kingdom expected to revise its system before long. and speed of depositor compensation were not surveyed.

©International Monetary Fund. Not for Redistribution Ill A SURVEY OF DEPOSIT INSURANCE PRACTICES

Table 2. Countries with Explicit, Limited Deposit Insurance Systems

Africa Asia Europe Middle East Western Hemisphere (4) (10) (32) (4) (17)

Kenya Bangladesh Austria Latvia Bahrain Argentina Nigeria India Belgium Lithuania Lebanon Barbados Tanzania Japan Bulgaria Luxembourg Morocco Brazil Uganda Kazakhstan Croatia Macedonia Oman Canada Korea Czech Rep. Netherlands Chile Marshall Islands Denmark Norway Colombia Micronesia Estonia Poland Dominican Republic Philippines Finland Portugal Ecuador Sri Lanka France Romania El Salvador Taiwan Province of China Germany Slovak Rep. Guatemala Gibraltar Spain Honduras Greece Sweden Jamaica Hungary Switzerland Mexico Iceland Turkey Peru Ireland Ukraine Trinidad & Tobago Italy United Kingdom United States Venezuela

Source: Survey results presented in Table A7 of the Statistical Appendix.

premiums as an alternative means to combat adverse Figure 2. Decade of Origination of Explicit selection.52 Deposit Insurance Systems While objectives are not investigated by the sur- (Number of Countries) vey, they can sometimes be inferred from deposit insurance practices. For example, if the objective of the system is primarily to protect small depositors, a 30 country is likely to include all institutions that are licensed to accept deposits (particularly, small de- 25 posits) from the public. To reduce unfairness to well- 20 supervised institutions, the country makes an effort to oversee all insured institutions to the same strict 15 standards. But, where a country is more interested in maintaining financial stability, it may confine mem- 10 bership to those classes of institution that it consid- ers to have systemic importance. In this case, mem- 5 bership may be focused principally on commercial banks. There may also be subsidiary insurance 0 1930s 1940s 1950s 1960s 1970s 1980s 1990s schemes for smaller, or less systematically important groups of institutions, such as savings associations Source: IMF staff survey. and credit cooperatives. Table A2 of the Statistical Appendix examines these issues and finds that countries typically attempt to cover (in one or more insurance system) all systems surveyed are compulsory.51 Nevertheless, institutions that take deposits. Confining coverage to seven schemes are voluntary and three of the volun- licensed commercial banks tends to be the excep- tary schemes (those in the Dominican Republic, Sri tion, not the general rule (column 2, Table A3). Lanka, and Switzerland) do not impose risk-adjusted

52The voluntary system in Sri Lanka began in 1987 by charging 51The sum of the numbers of compulsory and voluntary systems its 13 members a premium of 0.04 percent of deposits. In 1992, exceeds 67 because some countries have more than one deposit the premium was raised to 0.15 percent and two banks withdrew. insurance system. Only seven members currently remain.

©International Monetary Fund. Not for Redistribution Funding the Deposit Insurance System

Countries typically require the branches and sub- ment to back up a well-run system of deposit insur- sidiaries of foreign banks that are operating (taking ance that is met by unexpected demands on its re- small deposits) within a country to belong to the sys- sources and is in need of additional funds in order to tem (column 3, Table A2). Countries in the European carry out its responsibilities. Consequently, many Union may relax this requirement somewhat by countries make provisions for the government granting exemptions to foreign institutions that are (preferably, but not always through the ministry of covered by their home system of deposit insurance, finance) to assist a depleted fund with loans. While although they may allow them to join if coverage in 66 of the explicit, limited systems have private fund- the host country's system is more generous than the ing, 55 have access to public funding. Some have al- home scheme. Country authorities typically see their ready received financial help from official sources to responsibility to protect their citizens. Thus, they get the system started or to cope with a systemic typically do not insure the deposits that domestic banking crisis; others expect to obtain it when they banks take offshore (column 4, Table A2). The sur- need assistance (column 3, Table A3). To contain vey noted two exceptions to this general practice. moral hazard among bankers, banks must be re- The first covers countries of the European Union, quired to repay their loans, including those from the which often offer coverage to customers of their government. banks anywhere within the European Union. The The Canadian government goes further in requir- second exception is countries that are particularly ing that the Canada Deposit Insurance Corporation dependent on foreign deposits and fear the impact of (CDIC) pay a "credit enhancement fee" to the gov- their loss on the domestic financial system. ernment when it borrows funds in the private mar- kets. The rationale is that, as a Crown Corporation, the CDIC can borrow at a lower rate than it would if Funding the Deposit Insurance System it were a private corporation. The fee covers the dif- ference. Thus, when the CDIC borrows it pays a pri- Funding for the system of deposit insurance has to vate market rate. be adequate, and has to be seen by the public to be As mentioned in the discussion of good practices, sufficient, if the system is to succeed in compensat- a reticence to commit public funds would be under- ing depositors and maintaining public confidence. standable where a deposit insurance scheme is pri- To this end, there are a number of issues to be ad- vately run because of potential conflicts of interest. dressed. They include whether (1) funding should be The information in two of the tables in the Statistical mainly private, but have public backing in emergen- Appendix (Tables A4 and A8) brings to light cases cies; (2) the system should accumulate a fund or im- where such conflicts of interest may exist among ex pose ex post levies; and (3) whether to give to de- post systems. The two tables show that seven of the positors or the insurance system legal priority over privately administered ex post systems do have ac- the assets of the failed bank. cess to back-up funding from the government. In two other instances, however, the authorities explicitly deny that they offer backup funding. The situa- Private Funding tion is unspecified and unclear in the other three in- A deposit insurance system that is privately stances. Only one ex post system (in the funded encourages bankers to keep their institutions Netherlands) is operated by the government. sound. All but one of the 67 of the explicit, limited systems in the survey are privately funded by their member institutions. Only Chile offers an exception; Ex Post Schemes and Funded Deposit its system is fully funded by the government. Insurance Schemes A country has a choice between funding its system of deposit insurance ex ante by regularly charging Official Backing premiums to member banks and accumulating them As discussed in Section II, an underfunded in a fund, or imposing a levy on surviving member scheme will prove to be an obstacle to closing failed institutions after a member bank fails. As shown in banks and so may lead to costly forbearance.53 column 4 of Table A3, most (58) countries have Countries usually decide that they want the govern- opted for a funded system. However, nine countries—Austria, Bahrain, Ger- many (for its private system), Gibraltar, Italy, Lux- embourg, the Netherlands, Switzerland, and the 53The best known example of an insolvent insurance scheme is perhaps the Federal Savings and Loan Insurance Corporation in United Kingdom—fund their systems solely or the United States, which practiced forbearance for a number of mainly by imposing a levy on members after a bank years with costly consequences for U.S. taxpayers. fails and its depositors need to be compensated. Ex

©International Monetary Fund. Not for Redistribution Ill A SURVEY OF DEPOSIT INSURANCE PRACTICES post funding is more popular in Western Europe than recommend transparency and sharing information, it elsewhere. That popularity is waning, however; Ger- is perhaps not surprising that most recently created many instituted a government-run funded scheme in systems have opted to build a government-spon- 1998, France changed from an ex post to a funded sored fund. system in 1999, Bahrain has draft legislation to fund One of the notable ambiguities in ex post systems its system, and Italy is reported to be considering concerns the base on which the insurance obligation switching to a funded system. is to be calculated. While four ex post countries base the insurance obligation on insured deposits and another uses total deposits, the base is less specific in Contrasting Ex Post and Funded Deposit Germany's private schemes, Italy, the Netherlands, Insurance Systems and Switzerland (Table A4). Differences need not be inherent in the design of One of the principal differences between ex post the two different forms of deposit insurance sys- and funded systems is in the coverage they offer. Ex tems; but in practice, they exist. There are, in fact, a post schemes typically offer low coverage. For number of differences—some important—between example, Austria, Luxembourg, the Netherlands, funded and most ex post systems.54 First, five of the Switzerland, and Bahrain all offer coverage at less ex post schemes began in Europe in the late 1970s than per capita GDP (see Figure 1). Only Italy (at 5.5 and early 1980s, which is earlier than many of the times per capita GDP) and Germany's private sys- funded schemes.55 Second, ex post schemes were tem offer coverage above the commonly used rule- often initiated by groups of bankers seeking mutual of-thumb of twice per capita GDP. In addition, Aus- protection, whereas funded insurance systems have tria (for business deposits), Gibraltar, Luxembourg, more typically been sponsored by the government. and the United Kingdom also impose a haircut on Third, seven of the ex post systems have remained deposits under their systems of coinsurance. both privately funded and administered. (Bahrain's There is likely to be a difficulty for a government- and Gibraltar's privately funded systems are jointly run central bank or supervisory agency to pass confi- administered.) Four of the ex post, privately admin- dential information on financial condition of a mem- istered schemes have government backing, how- ber bank to a privately run bankers' club that is ever—a situation that presents potential conflicts of operating an insurance system. Deposit insurance interest. staff in Argentina and France have explicitly men- Funded schemes appear to be more rule-based and tioned this difficulty. Germany's private system re- offer less discretion for the administrators and less quires its members to be audited and classified by uncertainty for those insured than ex post systems. the Auditing Association of German Banks, which The reason may be that ex post systems are privately can impose disciplinary measures. run by their member institutions and they lack the authority of a government agency to promulgate and enforce rules. So, typically, private systems have not Guarding Fund Resources transparently specified members' responsibilities re- Investing fund assets wisely will guard fund regarding sharing the costs of compensating deposi- sources. The survey noted—in column 5 of Table tors. They also often lack backstop funding from the A3—that many systems place their resources in government; are limited in their roles and responsi- domestic government securities. Some encourage in- bilities; and, because they are privately run, have vestment in safe assets abroad. Unfortunately, a num- difficulty in obtaining information from the supervi- ber of deposit insurance systems invest their funds in sor and the central bank. Given that good practices domestic banks, which places them in jeopardy.

Granting Depositors or the Deposit Insurance 54 The author is grateful to Charles Siegman for this insight. System/Deposit Insurance Agency Legal Priority 55Initially, a number of the ex post schemes were voluntary, but by the end of the twentieth century all were compulsory. All but From their actions, half of the countries surveyed one (Bahrain) of the nine ex post systems are located in Europe (where information was available) perceive the fis- and all but two of the European ex post insurance systems (Gibral- tar and Switzerland) belong to countries that are members of the cal advantages of giving depositors priority over a European Union. The EU Directive on Deposit Guarantee failed bank's assets to be more important than the Schemes requires member countries to offer compulsory deposit incentive risks. The other half assess the balance insurance. Consequently, two member countries (Germany and differently. Table A3, column 6, shows that 31 of Italy) with ex post schemes that were previously voluntary now the countries surveyed gave legal priority to deposi- offer compulsory systems of deposit insurance. In addition, France switched from a voluntary, ex post system to a compulsory tors or the deposit insurer, but 30 countries did not. funded scheme in mid-1999. As a result, today, both funded and In other countries, such as Hong Kong SAR and ex post schemes are now typically mandatory. Malaysia, priority is/was used as a way to protect

©International Monetary Fund. Not for Redistribution Funding the Deposit Insurance System depositors without establishing a formal system. Controlling Administrative Costs (Malaysia found legal priority insufficient on its own to maintain depositor confidence during the Limiting administrative costs is an additional way Asian crisis and also introduced an explicit full to protect system resources. The staff may be kept guarantee. Hong Kong SAR is considering intro- small, but may be supplemented in emergencies by 56 borrowing skilled employees from other agencies. ducing a limited system of deposit insurance.) The scheme (particularly a narrow system of deposit insurance) may also be managed by another agency, Choosing When to Begin where it would remain largely dormant until needed. A country must make an important decision re- While the survey did not inquire systematically into garding when to introduce a deposit insurance such practices, column 6 of Table A8 of the Statisti- scheme. Beginning one too soon before the banking cal Appendix provides some information on country system has been strengthened can lead to risky be- practices with regard to running the deposit insur- havior at weak banks that will lead to major expen- ance scheme. ditures in resolving failed banks and can cause the system to become insolvent. Yet, countries are Setting a Target for the Fund often tempted to begin a limited, explicit system when a crisis is imminent or in progress in the mis- Many countries find it useful for the deposit insur- taken belief that it will avoid or cure the crisis. Lim- ance agency to set a target level for the fund (usually ited coverage will not prevent uninsured depositors expressed as a percentage of total or insured de- from running to safer havens. Even without a sys- posits) that would allow it to attain and retain finan- temic crisis but fearing runs, the authorities may cial viability and avoid the financial deficiencies that consider setting the coverage rate high—perhaps lead to forbearance for troubled banks and/or insol- too high. vency of the fund. Private funding needs to be suffi- If the public perceives that all banks are weak, cient to meet all demands that can be expected to be there is a risk of a "flight from the system and from placed upon it in normal times and in moderately ad- the currency" to banks abroad. Otherwise, there will verse circumstances. When the insurance system is be a flight "to quality" from weak banks to safer in- new, the target will be initially set after forecasting stitutions within the country. Only full coverage can the income and expenses (including outlays to com- (but not necessarily will) counter flights to quality pensate depositors of failed banks) of the fund. The and from the currency. Thus, to initiate a limited de- target then provides an indication of the premiums posit insurance system when there is a risk of de- that need to be set, and subsequently whether they posit runs, is to invite such runs. Setting high, but should be reduced when the fund exceeds its target limited, coverage does not resolve the dilemma. Not level, or raised to replenish a depleted fund. Setting only may there be runs by those depositors who hold an appropriate target demands a realistic assessment deposits above the limit; but politically, later reduc- of the condition of the banking industry, the size and ing the coverage level in order to reduce moral timing of the financial demands that are likely to be 57 placed on the fund, the system's ability to borrow hazard will prove very difficult. Faced with a sys- when necessary, and the industry's ability to pay the temic crisis, a country has two main courses of ac- necessary premiums without prejudicing its prof- tion to avoid runs: retain its existing implicit guaran- 58 tee; or institute a full, explicit, temporary guarantee. itability, solvency, and liquidity. Questions concerning the placement of a full guar- Canada has adopted a different approach. It re- antee and its removal are addressed in more detail in quires its system to estimate its future losses and to Section IV. make provision for them. Such provisions form a part of the accumulated fund, which may not be sufficient to compensate depositors in the interval be-

56 fore the failed bank's assets are sold. Should the Among countries granting full guarantee (Costa Rica—for CDIC need liquidity or underestimate the demands state-owned banks, Ecuador, Honduras, Indonesia, Japan, Korea, Malaysia, Mexico, Thailand, and Turkey), Ecuador, Malaysia, that are placed upon it, it can borrow from the mar- Mexico, and Turkey had previously granted depositors priority. kets or the government to supplement its small fund. 57Countries, such as the United States, set a high level when it This process involves partial reliance on ex post raised coverage in 1980 and had to wait for inflation and the funding to repay the borrowed funds. growth of GDP to reduce the coverage ratio to more incentive- compatible levels. The $100,000 limit set by the United States in Column 2 of Table A4 shows that 29 countries 1980 was virtually nine times per capita GDP at that time. It has (most of which have a funded system) maintain a taken 19 years to reduce the coverage ratio to the current more incentive-compatible level of three times 1999 GDP. The exces- sively high coverage contributed to the S&L crisis in the 1980s. 58If the banking industry is very weak, a deposit insurance (See Garcia and Plautz, 1988, pages 257-279). scheme may not be feasible until it has been restructured.

©International Monetary Fund. Not for Redistribution Ill A SURVEY OF DEPOSIT INSURANCE PRACTICES

Figure 3. Insurance Fund Targets and Actual Levels Attained

Kenya Romania Argentina Macedonia Croatia Actual Level Taiwan Province of China Fund Target Germany Tanzania Finland India Slovak Republic Hungary United States Italy Belgium Poland 0 2 4 6 8 10 12 14 16 18 20 Percentage of insured deposits

Source: IMF staff survey.

target level for the fund, which is often expressed as sate depositors under ex post assessments. Charging a desirable percentage of insured deposits. Eight premiums on all of the deposits that a bank holds is have explicitly not set a target. Gibraltar, Italy, and easier to administer than to charge selectively. the United Kingdom have small targets for covering Twenty-seven countries do so, but many consider administrative expenses in their ex post schemes, but charging premiums on categories of deposits that are small size does not reflect on the adequacy of the not eligible for insurance inequitable. Thirty-six capital resources of the system in these countries. countries, therefore, levy charges against insured de- The target in funded systems ranges from a low of posits (Table A4, column 4). 0.4 percent of all deposits in Poland to the very high Some systems (six) impose charges on the total levels of 20 percent of insured deposits in Kenya, value of deposits in those categories that are eligible and an unrealistically high 50 percent in Ecuador. for insurance. These are referred to as "insurable de- The level of accumulation actually achieved by most posits" in Table A4. Seven others, such as Austria, countries falls below the targeted level. Funding de- Belgium, Canada, Guatemala, Sweden, Peru, and ficiencies are not universal, however. Ukraine re- Taiwan Province of China, go farther and charge ports a healthy balance of 10 percent of insured de- only for the amount of deposits that would be com- posits in its fund, Tanzania approximates the target pensated if the bank were to fail. That is, premiums for its fund, and the United States' balance in its are paid on the sum of deposits that lie below the funds exceeds their targets. (Italy exceeded its low limit of coverage. That amount is referred to as target for meeting administrative expenses until the "amount covered" in Table A4. fund recently became depleted by bank failures in The latter procedure is more equitable in that it the southern region). Figure 3 shows the varying tar- avoids this cross-subsidization of insured deposits gets to which countries aspire and the levels (ex- by noninsured deposits, but it can be much more dif- pressed as a percentage of insured deposits) that they ficult to administer. In many cases, the information actually maintain. available (for example, the English translation of a country's deposit insurance law) was not sufficiently precise to determine whether premiums were The Premium Base charged on insured categories of deposits or only on As deposits are the entity that is insured, most sys- the amount actually covered. These cases are listed tems use deposits as the base on which to charge as "insured deposits" in the table. Clearly, calculat- premiums or to calculate the levy needed to compen- ing correctly the amount of deposits actually insured

©International Monetary Fund. Not for Redistribution Funding the Deposit Insurance System

involves knowing the size distribution of the de- there are at least two practical problems to risk-ad- posits of each depositor in each insured bank. Not all justing the premiums banks pay for the deposit guar- countries collect the information necessary to make antee. Solving these problems may be expected to this calculation, especially on a regular basis. elicit country-specific responses. A few countries use a base other than deposits. The first problem is accurately forecasting the de- For example, Norway bases its charges on risk-ad- gree of risk that a bank places on the fund—it is a justed assets. Poland charges premiums on deposits skill that is currently undeveloped. Moreover, the but sets an upper bound to those premiums that is risk premiums imposed by the deposit insurance based on risk-adjusted assets. agency need to be based on objective criteria so that they can be justified to the bank and the courts, Premiums Levied should the bank challenge the ruling. Two popular candidates for inclusion in the calculation of a A scheme that relies on an accumulated fund will bank's risk to the deposit insurance system are capi- need to charge adequate premiums. Table A4 shows tal adequacy and supervisory rating. Some countries that 58 systems charge premiums at regular inter- use one, some the other, while others, including the vals. The size of the premium needed to maintain a United States, combines capital adequacy and healthy fund will depend on the current condition of CAMELS rating into a composite measure. There the banking system and its future prospects. Premi- are disadvantages to these measures, however. Capi- ums charged in 1999 ranged from a temporary zero tal adequacy, even when accurately measured, tends percent of deposits for strong banks in the United 59 to be a lagging indicator of bank condition, and is States, and a regular low of 0.005 percent in also subject to manipulation through a bank's system Bangladesh, and a high of 2 percent in Venezuela, of loan classification and provisioning. Although su- which has experienced severe banking problems in pervisory ratings are kept confidential in most coun- the mid-1990s (see Table A4 of the Statistical Ap- tries, they will be revealed if the bank's annual ac- pendix). Figure 4 shows a distribution of premiums counts report the premium the bank is paying.61 An by size, and deposit base. The mode of the distribu- alternative, more direct approach to risk-adjustment tion is 0.15 percent and the medium lies in the band (used by Norway, Poland, and Germany's system for between 0.2 an 0.3 percent. savings and cooperative banks) is to charge flat-rate Without detailed knowledge of the condition of premiums on risk-adjusted assets instead of deposits, each country's banking system and deposit insur- so that banks with less risky assets pay less for their ance system, judging whether the premiums being insurance.62 This approach saves the banks some ef- charged are adequate to cover immediate outlays or fort because they have already calculated the risk- to accumulate a fund sufficient to survive the next adjusted assets in order to assess their capital ade- banking crisis is difficult. However, it is noticeable quacy. However, it may place undue emphasis on that the actual level of the accumulated fund falls imprecisely measured risk-adjusted assets. well below its target level in one-third of the 29 A number of other countries use a complex for- countries that maintain a target, suggesting that pre- mula to assess risk (Argentina, Canada, Italy, Kaza- miums in these countries are not currently adequate khstan, Romania, and Taiwan Province of China). To to meet the needs they face. Moreover, lax account- retain confidentiality and track risk accurately, the ing permits systems in some countries to report mis- 60 calculation of the risk premium can be designed to be leadingly healthy levels of accumulated resources. complex; yet there is a valid argument for simplicity, In addition, premiums in other countries that do not transparency, and accountability in premium setting. set a target may also be insufficient to cover the risks The just-mentioned characteristics may be desirable that the system faces. when the financial system is sound, but are unattain- able when it is weak. Thus, countries may want to an- Risk-Adjusting Premiums nounce their intention to risk-adjust premiums and Adjusting the premiums that banks pay for risk is then set a timetable for successive stages of widening conceptually a challenging process. In addition, the premium band so that the banks have time to make complementary adjustments as they wish.

59By law, the Federal Deposit Insurance Corporation in the United States does not impose premiums on the highest quality 61 The current proposal by the Basel Committee on Banking Su- banks when its fund is above its statutory target level of 1.25 per- pervision to allow supervisors to set institution-specific capital ad- cent of insured deposits, as it is in the year 2000. equacy ratios would make it more difficult for the public to dis- 60For example, a deposit insurance scheme may have made a fi- cover its supervisory rating. 62 nancial assistance loan to a very weak bank that may not be repaid Poland also uses risk-based assets to provide an upper bound but allows the bank to keep operating. The system may not have on premiums that are, in practice, determined as a percentage of made provision for the losses expected on this loan. total deposits.

©International Monetary Fund. Not for Redistribution Ill A SURVEY OF DEPOSIT INSURANCE PRACTICES

Figure 4. Premiums on Total Deposits, Insurable Deposits, and Covered Deposits

Total Deposits

El Salvador Nigeria Ecuador Ukraine Estonia Poland Brazil Maximum United States Honduras Minimum Trinidad & Tobago Morocco Philippines Uganda Dominican Rep Norway Sri Lanka Kenya Greece Tanzania Korea India Oman Bangladesh 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Percentage premiums

Insurable Deposits Macedonia Venezuela Turkey Lithuania Argentina Mexico Romania Czech Republic Maximum Bulgaria Minimum Kazakhstan Colombia Slovak Republic Latvia Finland Ireland Croatia Hungary Iceland Portugal Jamaica Spain Bahamas Japan 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Percentage premiums Covered Deposits Peru Guatemala Sweden Maximum Canada Minimum Taiwan Province of China Belgium Germany 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Percentage premiums

Source: GDP per capita for 1999, World Economic Outlook. 1Weighted by total deposits, IFS.

©International Monetary Fund. Not for Redistribution Deposit Coverage

The second problem is that a degree of subsidy is monitor the condition of his/her bank. Protecting the inherent in insurance. If premiums were to precisely stability of the financial system by promoting confi- represent a bank's risk to the fund, they would be- dence and avoiding bank runs is a second high-prior- come prohibitively expensive for already weak insti- ity objective. Resolution of a number of conceptual tutions. This observation reinforces the argument and practical issues follow from a careful explica- that the gradations in risk-adjustment should be in- tion of the system's priorities. troduced slowly, so that institutions can adapt their behavior over time by improving their management Limited and Full Coverage control practices in addition to reducing their risk exposure and thus the subsidy they are to receive Currently, almost all countries place limits on the from their stronger peers. explicit coverage they offer. Six countries (Ecuador, Given the difficulty in executing an equitable sys- Honduras, Japan, Korea, Mexico, and Turkey) that tem of risk-adjusting premiums, a surprisingly large normally have explicit but limited coverage, have number of countries attempt to do so. The systems temporarily, but explicitly, extended full coverage in 24 countries (Argentina, Canada, Colombia, during times of acute financial distress (see Table A5 Ecuador, El Salvador, Finland, France, Germany's of the Statistical Appendix).64 These countries plan private system, Hungary, Italy, Kazakhstan, Mace- to return to limited coverage when they can. An ad- donia, Mexico, Norway, Peru, Poland, Portugal, Ro- ditional three countries without a system of deposit mania, Sweden, Taiwan Province of China, Turkey, insurance (Indonesia, Malaysia, and Thailand) have the United States, and its two Asian island protec- explicitly extended full coverage during their finan- torates—the Marshall Islands and Micronesia) cur- cial emergencies, and all but Malaysia have already rently set risk-adjusted insurance premiums (column announced their intention to convert to limited cov- 6, Table A4).63 This is a marked increase in number erage when their crises are over. from earlier in the decade of the 1990s. Setting the Limits on Coverage Deposit Coverage The coverage limit should be low enough to encourage large depositors and sophisticated creditors Limiting the coverage offered by the system of to monitor and discipline their bank. Sophisticated deposit insurance is the most common way to con- depositors exert this discipline by demanding higher tain the moral hazard that deposit protection offers deposit rates from weaker banks in compensation for both to banks and their depositors. The actual objec- the higher risk of loss they are accepting; in other tives chosen for the system will influence a number circumstances, depositors may withhold funds en- of decisions that have to be made. These decisions tirely from a particularly troubled bank. include: (1) what types of institutions should be eli- There is a wide range to the limits that a country gible to join the system; (2) which financial instru- sets for its deposit insurance system, but there is ments should be covered; (3) which types of deposi- greater uniformity in the European Union, where a tors should be covered and which excluded; (4) the minimum coverage (€20,000 in the year 2000) is amount that is covered; (5) whether the basic prescribed. Translating the limits countries offer into amount should be covered in full or whether a hair- either 1998 dollars or euros, coverage ranges from cut should be imposed on the covered amount under the dollar equivalent of a low of $120 in Ukraine to a a system of coinsurance; and (6) deciding whether to high of $253,520 in Norway (excluding countries of- switch to full coverage in a systemic emergency. fering a full guarantee). While Germany's official This survey examines country practices with regard scheme offers limited coverage, Beck (2000) argues to all of the seven questions. that the private system offers virtually unlimited While country objectives were not explicitly sur- coverage to customers of member commercial veyed, clearly they will influence membership and banks. Corrective discipline is exercised by fellow coverage. It is, for example, possible to discern from members, which will be assessed ex post to meet de- countries' behavior with regard to coverage that in ficiencies, rather than by depositors. most countries, consumer protection is one of the top Any given limit expressed in dollar values (as in priorities for a system. Deposit insurance is designed column 2 of Table A5) will be more generous in a to conserve the time and money of the small deposi- country that has a low level of per capita income tor for whom it is not feasible or not cost-effective to than where incomes are higher. While IMF staff typically uses the world average of per capita GDP as a

63Each of these systems is compulsory despite the risk-adjusted premiums. 64Jamaica recently ended its full coverage.

©International Monetary Fund. Not for Redistribution Ill A SURVEY OF DEPOSIT INSURANCE PRACTICES

limit. Consequently, Figure 1 shows coverage per Figure 5. Deposit Coverage by Continent capita country by country and Figure 5 shows that (unweighted) the unweighted average coverage ratio worldwide is 2.4 times per capita GDP; with the highest average in the Middle East and the lowest in Europe. The weighted average is lower at 2.1 times per capita Middle East GDP. This number falls to 1.8 times per capita GDP if the largest country, the United States, which has Western Hemisphere relative high coverage, is excluded. The individual

Africa country offering the highest per capita coverage is Oman, which guaranteed up to 8.8 times 1999 per Asia capita GDP. The lowest ratio for coverage that ap- pears in the survey is that of Ukraine, which covers World only a small fraction of per capita GDP.

Europe As Figure 6 shows, countries typically cover a high percentage of the number of deposit accounts. 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Excluding countries offering a comprehensive guarantee, the percentage of accounts covered in full is Source: IMF staff survey. typically over 90 percent, although it is lower in Kenya, Nigeria, Sri Lanka, and Tanzania (see Table A6 of the Statistical Appendix). As is appropriate, the guarantee covers a typically much smaller percentage of the value of deposits, ranging from rough rule of thumb for appropriately limiting cov- negligible in Estonia and Sri Lanka, to 12 percent in erage, coverage observed in the survey is sometimes Tanzania, to a high of 76 percent in Norway (see high and can considerably exceed the rule-of-thumb Figure 7).

Figure 6. Percentage of the Number of Depositors Covered

Norway United States Colombia India Hungary Trinidad & Tobago Romania Finland Bangladesh Brazil Argentina Denmark Croatia Uganda Latvia Chile Spain Taiwan Province of China Guatemala Jamaica Canada France Kenya Nigeria United Kingdom Tanzania Sri Lanka 0 10 20 30 40 50 60 70 80 90 100 Percentage

Source: IMF staff survey.

Figure 7. Percentage of the Value of Deposits Covered

Norway India Italy Croatia United States Spain Denmark Hungary Slovak Republic Taiwan Province of China Lithuania Argentina Finland Canada Bulgaria Trinidad & Tobago Colombia Jamaica Bangladesh Uganda Nigeria Ukraine Latvia Kenya Tanzania Brazil Chile Estonia Sri Lanka

0 10 20 30 40 50 60 70 80 90 100 Percentage

Source: IMF staff survey.

Adjusting the Limits on Coverage (Kyei 1995; Lindgren and Garcia 1996). In fact, per- deposit coverage is offered in only one country (the Coverage can be adjusted upwards over time to Dominican Republic) today. reflect higher GDP and faster rates of inflation. If the Most countries apply their limit to the sum of all coverage ratio was initially set very low because the the deposits that a customer holds at any particular system fund needed time to build its resources, the bank. This arrangement allows a depositor to obtain level can be raised as the fund matures. The adjust- coverage above the limit by splitting his/her funds ments can be made by indexing coverage or making, across a number of banks. This relaxes the limit preferably rare, adjustments. There is both an advan- somewhat, but does not allow the largest depositors tage and a disadvantage to indexing coverage levels. to obtain full coverage. Some countries might want The advantage is that it avoids setting unduly high to go further and seek to limit coverage at any point limits initially; the disadvantage is that it will be in time to the sum of any individual depositor's ac- hard for the public to keep abreast of repeated counts across all banks, regardless of the number of changes in the coverage level. Also, indexing cover- accounts held in any or all banks. In fact, Chile at- age typically results in unrounded numbers, whereas tempts to do this by imposing a limit on coverage round figures are easier to remember and use. available in any year on deposits held by any single depositor anywhere in the financial system. Applying the Limit Per Deposit or Per Depositor Coinsurance Conceptually, a country could apply its coverage limit to each and every deposit that a depositor held Some countries attempt to strike a balance be- anywhere in the country. This would allow a deposi- tween discouraging moral hazard and avoiding sys- tor to split his/her funds into a number of accounts at temic runs by adopting a system of coinsurance. In the same bank and gain virtually unlimited coverage. 20 deposit insurance systems (those in Austria, Far fewer countries do this today than did a few Bahrain, Bulgaria, Chile, Colombia, the Czech Re- years ago. There has been a shift from per-deposit to public, the Dominican Republic, Estonia, Germany, per-depositor coverage since the earlier surveys Gibraltar, Ireland, Italy, Kazakhstan, Lithuania, Lux-

©International Monetary Fund. Not for Redistribution III A SURVEY OF DEPOSIT INSURANCE PRACTICES embourg, Macedonia, Oman, Poland, Portugal, and the United Kingdom) the depositor loses a small per- Figure 8. In Practice: Exclusions in the centage of the covered deposit but is reimbursed for Year 2000 the majority by the system. (See column 5 of Table (Number of Schemes) A5). The haircut commonly ranges from 10 percent to 25 percent, but some countries, such as Bulgaria and Kazakhstan, have multiple tranches on which 60 they impose successively higher haircuts that range up to 50 percent of the initial deposit. To protect 50 "widows and orphans," it is preferable to cover a 40 very small deposit in full and coinsure above that level. This dual arrangement will reduce the incen- 30 tive for retail runs while maintaining market discipline. In fact, 15 countries impose a haircut on all of 20 the insured deposit, while only five use coinsurance above the basic coverage limit. 10 0 Interbank Insider Government Foreign Illegal High rate Which Deposits to Cover and Which exchange to Exclude? Source: IMF staff survey. Most countries aim to protect small depositors while requiring larger depositors to monitor the condition of their bank and contain moral hazard. Of the two contrasting approaches used to achieve these two objectives, one makes the objective of deposit raising the costs of intermediation for their stronger protection politically and conceptually clear, while competitors. Eighteen countries guarantee only, or the other is easier to administer and effect speedy mainly, household deposits. Evidently, as shown in compensation.65 Figure 8, a large number of countries find it worth- The survey revealed that eight systems cover de- while to undertake the administrative burden of giv- posits of all types and 21 cover most kinds (see col- ing preference to less sophisticated depositors— umn 2 of Table A6). However, 17 systems exclude probably for political rather than financial reasons. all foreign currency deposits and nine schemes in countries that are in, or aspire to be in, the European Full Coverage in a Crisis Union exclude some non-EU currency deposits from coverage. Some countries that cover foreign deposits Ecuador, Honduras, Japan, Indonesia, Korea, (for example, France, Honduras, Jamaica, Latvia, Malaysia, Mexico, Thailand, and Turkey currently and Ukraine, among others), pay out in domestic offer coverage in full to depositors and creditors of currency to help protect the system from exposure to all types and also most other liabilities. Chile covers foreign exchange risk. Fifty-four systems do not demand deposits in full to protect the payment sys- cover interbank deposits. Thirty-three systems ex- tem, but offers limited coverage on other types of ac- clude government deposits and 34 countries explic- counts. The comprehensive coverage in six of these itly do not guarantee the deposits of insiders who countries (Ecuador, Honduras, Japan, Korea, Mex- could use privileged information to take advantage ico, and Turkey) overrides the regular deposit insur- of the guarantee. Twenty-three countries explicitly ance coverage in these countries. The other three exclude illegal deposits in their deposit insurance countries previously had no explicit system. laws. Nine countries exclude deposits that pay ex- Most of these countries began offering full cover- ceptionally high rates. This exclusion serves to dis- age when they perceived a financial emergency, with courage weak institutions from bidding for deposits the intention of replacing full emergency coverage and gambling for recovery with the proceeds and with a limited system after the banking system had been restructured to soundness. Sweden, Finland, and Jamaica also offered full coverage during their finan- 65The U.S. Federal Deposit Insurance Corporation (FDIC) and cial crises but have already retracted it and replaced it the Canada Deposit Insurance Corporation (CDIC) each have with a system of limited coverage. As discussed fur- computer programs into which they feed a failed institution's data ther in Section IV, Indonesia, Malaysia, Thailand, and to identify insured depositors and the amounts owed to them. The Turkey plan to replace their comprehensive guaran- fewer the exclusions from coverage, the more effective are these programs. Lacking exclusions, the FDIC typically compensates tees when their crises are over, but have not specified depositors within three days. a time for doing so. Ecuador plans to limit coverage

©International Monetary Fund. Not for Redistribution Governance in the year 2001, Japan and Honduras in 2002, Korea but is sometimes the bank supervisor. The deposit by the end of 2000, and Mexico plans to complete insurer was found to be subordinate to the central phasing out its blanket coverage by 2005. bank in one-third of the countries, and to either the ministry of finance or the supervisory agency in an- Timing the Repayment other third. Even so, the deposit insurance agency does not normally have the power to grant or with- The speed with which a depositor regains access draw bank licenses, to supervise, or to provide to his funds affects the value of the coverage of- lender-of-last-resort credit to failing banks, because fered. Delay reduces any stated coverage value ac- that would detract from the stature of the supervisor cording to the time-value of money and the incon- and possibly diminish its effectiveness. venience imposed on the depositor by the delay. Whatever the degree of coverage, small depositors at failed banks typically need access to their insured The Board of Directors funds rapidly. In effect, delaying payment reduces The survey shows that privately run systems are the value of coverage and increases systemic uncer- operated by bankers, who also typically are included tainty. Thus, it behooves the deposit insurer to com- on the boards of jointly run systems, while less fre- pensate insured depositors immediately, but cer- quently on government-run systems (see column 6, tainly within 30 days; otherwise, the credibility of Table A8). Government-run schemes typically in- the system can be undermined, depositors may run clude representatives of the supervisory agency, the from weak banks, and the retail payment systems ministry of finance, and the central bank represented may be disrupted. Depositors, finding themselves ex officio on the board of the deposit insurance without their transactions and savings balances, agency. In relatively few instances, the chairman and may curtail their expenditure, which can cause or the majority of the board are worthy, experienced, exacerbate a recession. The FDIC's practice of pay- but independent members of the public with no cur- ing compensation within three days is a good exam- rent ties to the banking industry. ple of prompt payment. In many countries, how- Good practices suggest that the government ever, the survey found that repayment is remarkably should provide back-up funding. Consequently, slow. The EU Directive allows countries three leaving financial decisions to a board of bankers is months to make payment, but allows them to extend likely to result in an underfunded scheme. Never- the time period in unusual circumstances. (Column theless, the privately run schemes in Argentina 4, Table A7.) and Germany have been successful to date. How- ever, bankers can form a consultative committee to advise the board of a publicly funded deposit insur- Governance ance scheme. There is a wide dispersion in arrange- ments regarding the running of the system. Thirteen Correct alignment of the deposit protection schemes are privately administered, 39 are run by the scheme with respect to three topics facilitates con- government, and 16 are jointly operated (columns trol over agency problems.66 The first necessity is to 3-5, Table A8). The authorities are able to exert ensure adequate funding. The second is to specify some influence over some privately run schemes, clearly the system's role and responsibilities. They such as those in Argentina and Brazil. were discussed earlier. The third is to design an ap- The danger of banks providing insufficient pri- propriate organizational structure. vate resources to maintain the solvency of the fund, as they hope to be subsidized by a government, ap- pears to be a reality in more than half of the systems Organizational Structure surveyed. Nine of the privately run systems and 12 In the quest for potential independence and appro- of the 16 jointly run systems have financial backing priate accountability, in 29 countries the deposit in- from the government. The remaining privately run surer constitutes a separate, independent legal entity schemes and 4 jointly operated systems, however, (see column 6 of Table A8). Nevertheless, in a num- have attempted to avoid this particular agency prob- ber of instances, it was, either in law or in practice, lem by refraining from providing government finan- under the control of a government agency, which is cial support. In some cases, the law is silent on the usually the central bank or the ministry of finance, subject of funding; in others, government financial backing is explicitly foresworn. However, whether those governments that have made a precommit- ment not to fund the system can sustain this com- 66As discussed in footnote 1, agency problems involve a lack of coincidence between the principal in a transaction and the party mitment in face of an underfunded system remains acting as the agent for its execution. to be seen.

Table 3. Characteristics of Deposit Insurance Systems by Continent in 1995 and 2000

Number of Deposit Coverage Insurance Systems Is Has Mainly Risk-Ad justs Administration No % Compulsory a Fund Premiums Private Joint Government Household interbank No forex 1995 2000 Growth 1995 2000 1995 2000 1995 2000 1995 2000 1995 2000 1995 2000 1995 2000 1995 2000 1995 2000

Africa 4 4 0 4 4 4 4 0 0 0 1 0 0 4 3 1 0 3 2 3 3 Asia 7 10 43 4 7 7 10 2 4 1 0 2 1 4 9 0 2 5 7 3 5 Europe 23 32 39 II 31 13 25 0 12 7 10 5 11 7 12 5 12 II 31 4 12 Middle East 2 4 100 1 4 1 3 0 0 1 0 0 2 0 2 0 0 0 2 0 1 Americas II 17 55 6 16 9 16 2 8 0 2 4 2 6 13 0 4 2 12 1 5 Total 47 67 43 26 62 34 58 4 24 9 13 II 16 21 39 6 18 21 54 11 26

Source: IMF staff survey.

Trends and Convergence to Second, more countries—more than one-third of Good Practice the total—now risk-adjust their deposit insurance premiums. Only two countries (other than the United A number of major changes in deposit insurance States' protectorates in Asia) were identified in Kyei practices can be observed in comparison to the ear- (1995) as adjusting their insurance premiums for lier surveys of Kyei (1995) and Lindgren and risk. The increase in risk-adjusting countries has oc- Garcia (1996). A summary comparison between the curred in Africa, Europe, and the Americas. Assum- present survey and that of Kyei (1995) is presented ing that the risk-adjustment is being well executed, in Table 3. It shows that there are many more (67) this change constitutes a substantial shift toward explicit systems in 2000 than there were in 1995, good practices. since a number of the countries listed in Kyei Third, there is a shift away from voluntary systems (1995) as having implicit schemes have replaced to compulsory schemes. Today, nine of every ten sys- them with formal, explicit schemes in a major shift tems seek to avoid adverse selection in this way, toward following good practices. In addition, other whereas only just over half did so in the mid-1990s. countries that were not included in the survey The switch has occurred not only in Europe, as a re- by Kyei have also recently put in place explicit sult of the 1994 European Union Directive on De- systems. posit Guarantee Schemes, but the trend has also been A number of other trends have developed since noticeable in the Middle East and in the Americas.67 Kyei's survey. First, the increase in formal systems Fourth, there has also been a trend toward funded has been marked in Europe, where the number has systems. While only a few countries have switched risen from 23 in 1995 to 32 in 1999. There has been from ex post levies to funded systems, newly created no increase in the number (4) of systems in Africa, systems have almost universally been funded. although six Central African countries agreed to (Gibraltar's new scheme is the exception in that it form a regional system of deposit insurance in 1999. follows the ex post practice of the United Kingdom.) A year later, however, only two of the signatories Funded schemes are universal in Africa and Asia and have ratified the agreement, which will not go into have increased elsewhere to reach dominance world- effect until all do so. There has also been some wide. Schemes that maintain a fund have been ob- growth in the number of systems in the Middle East served above to be more rule-based and less ambigu- and the Americas. In the Western Hemisphere, the ous in practice than the ex post systems favored by number of explicit systems has risen to 17 in 2000 bankers' clubs. In this respect, the recent emphasis from 11 in 1995. In addition, a number of additional on funded systems is another example of conver- countries in the Americas are planning to introduce gence toward good practices. formal systems to replace their implicit ones. The in- troduction of new systems may be expected in Asia as countries recover from their financial crises and 67While the regional scheme agreed to in Africa in 1999 would replace the full guarantees they have put in place be voluntary, the signatories attempted to combat adverse selec- with limited coverage. tion by proposing to risk adjust premiums.

Fifth, while most systems continue to be funded moral hazard is not a problem there. Nevertheless, it primarily by their member institutions, an increasing could be a problem for countries outside the Euro- proportion of systems have access to back-up funding pean Union that aspire to join the Union and so emu- from the government, as recommended in Section II. late the European Union's deposit insurance system In 2000, over three-quarters of systems have received coverage even when it is many times their per capita or can expect to receive government assistance when GDP. necessary. There has been a small commensurate Worldwide, there is a small but statistically signif- shift—from just under to slightly more than half—in icant negative relationship between per capita GDP favor of public administration of systems. This shift is and the deposit insurance coverage ratio. That is, to be expected as government funding is likely to be poorer countries, on average, offer higher coverage accompanied by government control. The remaining in relation to GDP than do richer countries. Perhaps systems of deposit insurance divide themselves be- they do so to enable their banks to complete interna- tween privately run and jointly run schemes. tionally and to discourage deposits from migrating Sixth, virtually all countries now provide coverage abroad. The inverse relationship is evident in all re- per depositor, rather than per deposit, which tends to gions except Europe, but is particularly strong in lower the effective coverage ratio. The reduction in Africa. This result indicates that moral hazard is pre- the number of per-deposit coverage noted in 1996 sent worldwide, but is stronger in developing coun- survey has continued so that by 2000, the number tries than in Europe. had been reduced further to only one country. Despite shifts in favor of good practice, areas re- There is also an increasing standardization of main for improvement. For example, some countries practices with regard to system coverage as a result have not resolved the potential conflict of having a of a EU directive, particularly among those countries privately run system of deposit insurance with gov- that are, or aspire to be, members of the European ernment financial support. In addition, deposit com- Union. In the interests of competitive equity among pensation practices are often surprisingly slow. This banks from different countries, the EU directive di- problem may be exacerbated by an increasing trend verges from good practice in one respect, however. toward excluding categories of depositors from cov- By requiring the same minimum coverage limit erage. For example, a large increase in the restriction (€20,000 by year 2000) in all member countries, the of system protection to individuals, households, and directive provides low per capita coverage in rich nonprofit organizations is observable, mostly be- countries but a higher ratio in poorer countries cause former Soviet countries made explicit their old within the European Union, which might, as a result, practice of guaranteeing household deposits. Such be more exposed to moral hazard.68 However, the restrictions have been adopted by over one-quarter mandatory coverage is so low in rich countries that it of countries, whereas only one-eighth did so in remains relatively low even in less affluent EU 1995. There has also been an increase (from 45 per- countries, so that moral hazard from this source is cent to 80 percent) in the number of systems that ex- unlikely to be a serious problem. Because there is so clude interbank deposits from coverage. The change little variation in coverage, the correlation coeffi- occurred in both Europe, where the EU Directive on cient between per capita GDP and the coverage ratio Deposit Guarantee Schemes lists interbank deposits is not significantly different from zero among mem- as a candidate for exclusion, and also in the Ameri- ber countries of the European Union, suggesting that cas. There has also been, perhaps surprisingly in light of the recent currency crises, a trend—apparent on all continents—toward excluding foreign currency deposits from coverage. In 2000, 39 percent of 68€20,000 is 0.4 times per capita GDP in Luxembourg, which had the highest per capita GDP in the European Union in 1998, countries excluded all or some deposits denominated but more than twice per capita GDP in Portugal. in foreign currency.