Blanket Guarantee Survey

What were the specific motivations for introducing the BG program?

Despite each BG program’s involving a similar desire to protect creditors from losses on covered liabilities, usually with the stated objective of promoting liquidity for institutions or the financial system broadly, the programs studied can be categorized as reactions to actual runs, preemptive attempts to bolster confidence and prevent the loss of funding, or some blend of the two:

• Actual Run: Widespread loss of funding causes policymakers to adopt a BG program to halt or reverse that loss.
• Concerns about Confidence: The fear that creditors may be losing confidence causes policymakers to adopt a BG program in a preemptive fashion to prevent a loss of funding.

In our earlier studies of account guarantee (AG) and bank debt guarantee (DG) programs, we noted two other common motivations: competition from other jurisdictions that guaranteed creditors of their financial institutions, which pressured policymakers to adopt similar programs to prevent competitive disadvantage; and supranational requirements adopted by bodies such as the European Union (EU). Interestingly, we did not see such forces at work in the BG cases. In the case of Ireland, European authorities criticized the country’s unilateral blanket guarantee in the Credit Institutions Financial Support Scheme (CIFS); however, they responded not with similar blanket guarantees but by expanding existing AG programs (ECB 2008). (For a list of acronyms used in this survey, see the Glossary.)

In several BG cases—Jamaica, Sweden, and Thailand—the authorities first guaranteed all liabilities for specific troubled financial institutions and later expanded the blanket guarantee to a broader set of institutions. One consideration in deciding to guarantee a particular institution is whether doing so will ultimately create pressure to expand the guarantees.

Foreign exchange pressures were an important motivation in several BG cases. For financial systems dependent on foreign creditors, the threat of currency devaluation can cause those creditors to withdraw funding. As will be discussed, blanket guarantees may be less effective when foreign exchange pressures are present.

Source: Authors’ analysis.

For those programs introduced in response to actual runs, the question of who was running and to where becomes important. In Jamaica, creditors ran from local institutions to the branches of foreign banks. Ecuador saw shifts to US dollar deposits and the withdrawal of trade credit lines provided by foreign banks. In Korea, the run first took place on foreign debt. Thailand experienced runs on finance companies and small banks. In Sweden, there were runs on one of the largest banks in the country and on Marknadsbevis (a form of short-term borrowing, like commercial paper (CP), which were then a significant source of funding for finance companies).   

In some instances, the specific nature of an actual or feared run resulted in a BG program designed to cover the liabilities or institutions at risk. Korea’s BG program began by guaranteeing foreign debt, which was the focus of creditor runs, and later expanded to include deposits as well. In Ecuador, the threat posed by withdrawn trade credit lines caused policymakers to target such liabilities with its BG program.

Was the BG program intended to work on its own or was it introduced alongside, or otherwise seen as operating in tandem with, other policy interventions?

One of the earliest decisions confronted in designing a BG program is whether the program will operate alone or in combination with other policy interventions. As discussed in Key Design Decision No. 1, Purpose, in implementing BG programs policymakers are typically responding to, or anticipating, creditor runs. Emergency liquidity facilities intended to restore lost funding were for this reason a common complement to BG programs. Indeed, researchers such as Laeven and Valencia (2008) and Kane and Klingebiel (2004) have used a decline in emergency liquidity facility usage as a proxy for the success of BG programs. Their assumption is that if such usage goes down following the introduction of a BG program, the BG program has succeeded in restoring liquidity because banks’ need for the emergency liquidity facility has receded.

Given that blanket guarantees are a far broader version of a tool—deposit insurance—that existed in several of the studied jurisdictions even before the crisis, several of the BG programs worked alongside deposit insurance systems. In Ireland, the CIFS acted as a supplement to the existing Deposit Guarantee Scheme (DGS). The DGS covered deposits up to EUR 100,000 (USD 140,000),^FPer Yahoo Finance, EUR 1.00 = USD 1.45 on September 30, 2008. while the CIFS covered deposits above this amount and other liabilities not guaranteed by the DGS. Denmark’s General Guarantee Scheme (GGS) supplemented an existing, privately funded deposit insurance system called the Guarantee Fund for Depositors and Investors, which covered deposits up to EUR 50,000.   

Liquidity-focused interventions will be insufficient, however, when financial institutions’ capital has eroded. For these institutions, injections of capital become necessary. Recapitalization programs are for this reason another recurrent component of packages involving BG programs.

The need to combine blanket guarantees with other interventions often resulted in BG programs that policymakers embedded within a larger umbrella. In Finland, the blanket guarantee was part of the Government Guarantee Fund (GGF), which also included capital injections and funding for asset management companies. Policymakers paired Jamaica’s blanket guarantee with bank rehabilitation efforts as part of the Financial Sector Adjustment Company (FINSAC). Indonesia’s IBRA similarly combined the blanket guarantee with bank rehabilitation.

In several countries, BG programs existed alongside support packages from the International Monetary Fund (IMF) and other multilateral bodies. This included Ecuador (IMF, World Bank, Inter-American Development Bank, Andean Development Corporation), Indonesia (IMF), Ireland (IMF, EU), Korea (IMF, World Bank, Asian Development Bank), Mexico (IMF), and Thailand (IMF). As Laeven and Valencia (2008) note, even where BG programs appear to have been successful in stopping creditor runs, the existence of such multilateral support packages may actually account for the restoration of confidence.

Figure 3 shows which types of interventions policymakers combined with BG programs. Many of these combinations were not as obvious as those contained under the GGF, FINSAC, or IBRA umbrellas. The determinations reflected in Figure 3 are based on some evidence indicating that policymakers thought of the BG program in question as operating in tandem with these other interventions.

Figure 3: Other Key Interventions Combined with BG Programs

Source: Authors’ analysis.

The existence of a blanket guarantee can also prompt policymakers to undertake additional interventions. With guarantees leaving the state responsible for creditor losses stemming from failed institutions, policymakers may be inclined to take the steps necessary to prevent such failures. This might include interventions that policymakers would not normally consider. In Sweden, for example, the Riksbank extended liquidity to troubled banks on an unsecured basis—that is, without imposing the central bank’s usual collateral requirements—because the government had already assumed responsibility for the banks’ obligations through the blanket guarantee (Borio, Vale, and von Peter 2010; Lundgren 2009).^FAlthough there is no indication that the Riksbank’s unsecured lending resulted in problems in this case, we have highlighted this practice as requiring caution because departing from typical collateral requirements may have consequences in certain contexts.

As will be discussed further in the Conclusion, the failure to combine BG programs with more comprehensive measures aimed at correcting underlying problems in the financial system can undermine the effectiveness of blanket guarantees. This appears to have been the case in Mexico, where the Fondo Bancario de Protección al Ahorro (FOBAPROA) initially restored a degree of confidence to the banking system before concerns returned stemming from a failure to take adequate steps to put the system on a sustainable path to full recovery (World Bank 1998). BG programs can also be combined with other policies or programs that will be counterproductive. In Ecuador, the imposition of a financial transaction tax and the introduction of a bank holiday and partial deposit freeze contributed to continued creditor runs even after policymakers established a blanket guarantee (IMF 2000; Jácome H. 2004).

What legal basis did policymakers rely on in establishing the BG program? Was there a need for new legislation or could they rely on existing authority?

A key consideration for policymakers is the legal authority that they will rely upon in establishing a BG program. In some jurisdictions, established frameworks may enable policymakers to use existing authority to implement blanket guarantees without any additional act by a legislative body. In other jurisdictions, such an additional act may be required before a BG program can be adopted. The legal basis for a BG program can have important implications for the program’s speed and design.

In several cases, policymakers announced BG programs without specific legal authority and only created that authority weeks or months later. Guarantees based on a government’s announced intention to stand behind the liabilities of its financial institutions are sometimes referred to as “political guarantees” and present a unique situation from a legal perspective. When such intentions are not transposed into law, they may have unclear legal status. In Mexico, for example, the law required FOBAPROA to make an annual announcement concerning the amount of bank obligations it would protect in the coming year. In 1993 and 1994, FOBAPROA announced that it would try to stand behind all liabilities of the banking system other than subordinated debt. A 1995 report from the World Bank cites Mexican authorities and private lawyers as viewing these announcements as creating no actual legal obligation on the part of FOBAPROA (World Bank 1995). Whether or not it created a legal obligation, the government honored the BG commitment by using FOBAPROA for bank recapitalization and the purchase of non-performing loans from troubled banks, thereby protecting creditors.

Figure 4 divides the BG programs into those requiring a new act by a legislative body and those that relied on existing legal authority. Included in this latter category are programs created pursuant to an executive or other emergency decree. In the AG and DG contexts, the need for a legislative act often (but not always) resulted in significant delays between the announcement of a program and its implementation (McNamara, Feldberg, et al. 2020; McNamara, Kulam, et al. 2022). Interestingly, for BG programs this does not appear to have been as big an issue. In most BG programs, there was little to no delay between announcement and implementation, even where legislative approval was required. Two exceptions were Finland and Sweden. In Finland, there was a six-month gap between a government statement committing to support banks under any circumstances and legislation enacting its BG program. It is not clear that observers viewed the government’s initial statement as the announcement of a blanket guarantee. In Sweden, approximately three months elapsed between announcement and implementation. It could be that BG programs were a more aggressive intervention introduced only after crises had become more acute, ensuring adequate political will for rapid implementation.

Figure 4: BG Program Legal Authority

Source: Authors’ analysis.

European Union member states had to consider not only their own national legal requirements, but also the guidelines and directives of the EU. Despite this, Ireland acted unilaterally in announcing the CIFS. Only after this announcement did it engage with EU authorities to secure the necessary approvals. This resulted in criticism from those authorities (ECB 2008).

The absence of an adequate legal framework can create problems for BG programs. This is especially true with respect to the framework for resolving failed institutions. An inadequate resolution framework can hamper the ability to deal with failed institutions and protect creditors pursuant to the blanket guarantee. Under Thai law, for example, authorities were initially only able to close firms and make cash payouts in satisfaction of the guarantee. Thai lawmakers subsequently passed legislation to give authorities the ability to take over and resolve insolvent financial institutions.

Who was responsible for the day-to-day activities of the BG program?

A key question in designing a BG program is who will administer it. The first decision is whether to create a separate entity or to give the responsibility to the central bank, bank regulator, treasury, an existing deposit insurance agency, or a newly created insurer (see Figure 5). Particularly for newly created entities, one question is whether they will have the resources and expertise necessary to effectively administer a BG program. An additional consideration is the other financial stability roles of the responsible entity. For example, a BG administrator that is also responsible for bank supervision may have access to information that helps it administer the guarantee.

Figure 5: BG Program Administration

Source: Authors’ analysis.

Although private entities have often been involved in AG programs (see McNamara, Kulam, et al. 2022), this appears to be much less common in the BG context. Denmark was unique among our BG programs in employing a private-public model, in which a private industry association (the Private Contingency Association, or PCA) partnered with a newly created public body (the Financial Stability Company, or FSC) to administer the blanket guarantee. Otherwise, the BG programs studied were publicly administered.

Changes in administration over the course of a BG program can signal underlying problems or even be the source of such problems. In Finland, the newly created GGF lacked staff of its own and relied on employees assigned from the Finnish central bank and bank regulatory agency. This was both inadequate given the number of troubled banks and potentially problematic from a conflict-of-interest standpoint, given the role of these bodies in bank supervision. Finland ultimately adopted legislation reforming the administration of the GGF and adding permanent staff (Nyberg and Vihriälä 1994). In Indonesia, shifts in administration between IBRA and the Indonesian central bank delayed payment of interbank claims under the blanket guarantee, undermining its credibility (Enoch et al. 2001).

Who was responsible for the oversight of the BG program? Were there any disclosure requirements?

After determining who will be responsible for the day-to-day activities of a BG program, policymakers must then decide how those activities will be overseen. Because many of the programs in our survey were separate government agencies or corporations, they had boards providing oversight.

The composition of boards is an important consideration. They typically included government appointees from the finance ministry, central bank, and bank regulator. Many had independent members who had no connection to the government.

Required disclosures can be another important avenue for effective oversight. Many of the BG programs were mandated to issue regular reports on their activities and submit to periodic audits. The absence of these measures or the failure to make the results public could trigger concerns about a lack of transparency. In Mexico, FOBAPROA did not initially make its financial statements publicly available (World Bank 1995). Following a change in control of the national legislature, legislators launched a nine-month investigation into FOBAPROA (Haber 2005).

How did policymakers describe the need for and objectives of the BG program? Were there any special communication issues?

Communication is of particular importance for BG programs. As with AG programs, broad public awareness and understanding of BG programs are essential to their success. BG programs cannot instill confidence if creditors do not know they exist or understand how they are intended to work. At the same time, because of the stigma that can be associated with crisis interventions, BG administrators need to communicate in a manner that does not erode the very confidence that the programs are intended to safeguard. One potential approach is to highlight that broad political support exists for addressing the problems that are the focus of the interventions. In Sweden, authorities used a joint announcement with the political opposition to unveil the country’s BG program. Some have credited this broad political support as having contributed to the BG program’s strong credibility (Englund 2015).^FOn the flip side, the lack of political unity can create additional obstacles to a successful BG program. Political turmoil in Ecuador, Indonesia, and Thailand hindered those countries’ BG programs.

As noted in Key Design Decision No. 1, Purpose, the BG programs studied could have different underlying motivations. A key consideration from a communication perspective was how policymakers described those motivations to the public.

Despite the importance of effective communication, some programs seem to have experienced difficulties on this front. In Thailand, a lack of clarity about the exact liabilities to be covered by the blanket guarantee and the way it was funded undermined confidence (Nabi and Shivakumar 2001; Sharma 2013). In Korea, the finance minister failed to specifically mention trust funds in his statement on the blanket guarantee, resulting in KRW 200 billion (USD 175 million)^FPer OFX, USD 1.00 = KRW 1,040.00 on November 19, 1997. in withdrawals from 23 trust funds across Korea over the course of three hours. This panic was arrested only after an association of trust funds issued a statement clarifying that the funds were indeed protected (Agence France Presse 1997). Indonesia’s president initially prohibited IBRA from disclosing its operations publicly. As a result, depositors and creditors viewed IBRA as non-operational and did not believe that their funds were guaranteed (Enoch et al. 2001).

Both Korea and Indonesia appear to have learned from these initial communication errors. In Korea, as the expiration of the blanket guarantee neared, the Korea Deposit Insurance Corporation (KDIC) embarked on a large-scale public awareness campaign to communicate the main concepts of the transition to a limited deposit insurance scheme. This included large public events, seminars, print advertisements, advertisements by insured institutions, Internet-based advertisements, and media channel advertisements. To gauge public awareness of the transition, the KDIC also collected surveys (IADI 2012; KDIC 2001). In Indonesia, researchers have credited effective communication about a series of bank closures in April 1998 with bolstering confidence in the country’s BG program (Enoch 2000).

In the event that payouts needed to be made under a BG program, where did the money come from and how much was available? What would happen if the primary source of funding for a BG program proved to be insufficient?

A particularly important design decision is the source and size of the funding that will be used to protect creditors in the event of an institution’s failure. A BG program that creditors perceive as backed by an insufficient funding source will not instill confidence in the safety of guaranteed liabilities. While many of the BG programs charged fees on participating institutions as discussed in Key Design Decision No. 10, Fees, authorities typically did not expect such fees to fully cover creditor losses. This possibility raises the question of who would be responsible for losses after fee revenues have been exhausted. In general, the BG programs relied on the state’s fiscal resources to support the guarantees. Authorities in Denmark employed a unique approach to this question, using the private-public nature of the country’s BG program to make the banking sector responsible for a first-loss position before the state covered any losses beyond this amount.

The role of the state as the ultimate source of support for BG programs meant that the state’s fiscal position could have considerable influence on the programs’ credibility. Researchers have credited strong government finances with contributing to the credibility of Sweden’s BG program (Edmonds 2015). Conversely, concerns about the state’s finances undermined BG programs’ credibility in Thailand, Ecuador, and Ireland. In Thailand, authorities struggled to restore public confidence because of the size of liabilities the program covered and a lack of clarity about the mechanism for funding (Nabi and Shivakumar 2001; Santiprabhob 2003). In Ecuador, creditors viewed the guarantee as credible for small institutions, but doubted the ability of the state to honor the guarantee in full in the event of the failure of a large institution (Beckerman and Solimano 2002).

The CIFS in Ireland was the focus of widespread uncertainty from the outset. In its first response to the CIFS released days after the program’s announcement, the European Central Bank (ECB) highlighted its concern that the size of the government’s exposure was “very large” at a time when the Irish budget was already deteriorating (ECB 2008). Such concerns undermined the credibility of the CIFS. The Irish example also illustrates that a blanket guarantee can end up imposing significant fiscal costs on the state. To avoid the failure of institutions covered by the guarantee, Irish authorities undertook costly rescue operations. Ultimately, the overall cost of the Irish response to its banking crisis totaled around 40% of GDP, the second costliest in advanced economies since the Great Depression (IMF 2012).

What types of institutions were eligible to have their liabilities guaranteed under the BG program? Was participation mandatory or voluntary?

In determining what institutions will participate in a BG program, policymakers must define eligibility and decide whether participation will be mandatory or voluntary. While in the AG and DG program contexts authorities often restricted eligibility to credit institutions, BG programs frequently encompassed a broader range of institutions (see Figure 6). And while BG programs typically defined eligibility in terms of institution type, they could also target specific entities. Ireland’s CIFS initially covered six specific institutions, resulting in concerns about preferential treatment (ECB 2008). Ireland ultimately expanded coverage to include certain Irish subsidiaries of foreign banks. Even with this expansion, coverage remained limited to 11 institutions and excluded many banks operating in Ireland (Woll 2014). In Sweden, in addition to the entities eligible as a result of their banking charters, authorities specifically included eight other credit institutions in the country’s BG program because of their vital role in the payments system. (Drees and Pazarbasioglu 1995). Programs could also include a broader range of institutions, while in practice being primarily focused on a specific type. In Korea, merchant banks were the main institutions in need of support and ultimately the largest recipients of aid; commercial banks, securities companies, insurance companies, mutual savings and finance companies, and credit unions were also eligible.

The treatment of foreign banks is also a consideration. In Denmark, EU guidelines required that foreign branches lacking equivalent coverage in their home countries be eligible for inclusion in the deposit side of Denmark’s BG program. Such branches were not entitled to have their other liabilities covered. In Ireland, the designation of six specific institutions for eligibility initially excluded even the Irish subsidiaries of foreign banks. This caused a run on National Irish Bank, a subsidiary of Danske Bank. While Irish authorities subsequently expanded coverage to include certain Irish subsidiaries of foreign banks, many banks operating in Ireland remained ineligible. In Sweden and Finland, by contrast, foreign branches were eligible for the BG programs.

Figure 6: Eligible Institutions

Mandatory participation was common for BG programs. For those programs that were voluntary, participation could either (a) be automatic with an ability to opt out or (b) require an affirmative opting in. Denmark’s BG program was an example of the former. Membership in the private association that co-administered the BG program resulted in automatic enrollment. Institutions could leave the association to opt out. Indonesia had an opt-in program. Eligible institutions had to sign an agreement to participate.

What types of liabilities were eligible to be guaranteed under the BG program?

In addition to eligible institutions, policymakers must also determine what liabilities can be included in BG programs. Equity, as a non-liability, was universally excluded from BG programs. Most BG programs similarly excluded subordinated debt because it is also intended to absorb losses. Ireland faced significant criticism for covering dated subordinated debt, although this appears to have been required by Irish law (Baudino, Murphy, and Svoronos 2020).

Another interesting element is the treatment of existing (as opposed to newly issued) liabilities. As Baudino, Murphy, and Svoronos (2020) point out, medium- and long-term debt that is already in place cannot run: as a legal matter, creditors cannot demand their money before the debt is due. Thus, the typical motivations underlying a guarantee (the desire to stop or prevent a run) do not apply. Yet all of the BG programs studied included existing liabilities within their coverage. Potential explanations include a legal inability to treat existing debt differently from newly issued debt (as appears to have been the case in Ireland), together with more generalized concerns about the risk of imposing losses on even non-runnable liabilities. Such losses on non-runnable liabilities may (a) undermine confidence among the holders of other, runnable forms of liabilities and (b) result in asset sales among holders of non-runnable liabilities, producing contagion.

In the AG context, the question of which accounts are to be included or excluded from a guarantee takes on even greater significance given the risk that the introduction of a guarantee will itself trigger a run from non-guaranteed accounts to guaranteed accounts (McNamara, Kulam, et al. 2022). This appears not to have been as big a concern in the BG context as a result of the generally expansive coverage such programs provide. An exception was Jamaica’s BG program, where there is some evidence that the announcement of the program caused a shift to liabilities covered by the guarantee (although documents do not specify the non-guaranteed liabilities from which the shift occurred) (Bonnick 1998).

As discussed in Key Design Decision No. 1, Purpose, certain BG programs responded to runs on specific liabilities and thus had coverage targeting those liabilities. Given that Korea’s run focused on foreign debt, its BG program began by guaranteeing such debt specifically, before expanding to include deposits and collateralized commercial paper as well. In Ecuador, the threat posed by withdrawn trade credit lines caused policymakers to target such liabilities with its BG program.

The treatment of interest on liabilities is also a consideration. In Thailand, for example, the guarantee covered both principal and accrued interest. In Korea, the guarantee initially covered both principal and interest. However, after the emergence of moral hazard concerns in the form of weak banks offering high interest rates to attract deposits, authorities excluded interest for large depositors and covered interest for small depositors only up to the market rate (Baliño and Ubide 1999). Ecuador excluded deposits that paid too high a rate of interest outright as a means of preventing moral hazard.

The liabilities covered by a BG program could change over time. In Ireland, authorities reduced the scope of the non-deposit liabilities covered by its BG program as it gradually phased out the program. In Ecuador, authorities ended coverage for the trade credit lines initially targeted by the country’s BG program while leaving the deposit coverage in place. Indonesia’s BG program initially covered derivative transactions broadly, before authorities excluded all derivatives other than currency swaps from coverage.

What fees, if any, did institutions have to pay as part of the BG program? Were such fees flat or risk based?

As noted in Key Design Decision No. 7, Source(s) and Size of Funding, many of the BG programs funded themselves at least partially through fees paid by covered financial institutions. In determining the fees for participation in a BG program, there is a tension between the appropriate fee given the risk to the guarantor and the desire to avoid charging institutions too much at a time when they are already strained. Figure 7 shows how each BG program approached the fee question. Unlike AG and DG programs, few BG programs used risk-based fees (see McNamara, Feldberg, et al. 2020; McNamara, Kulam, et al. 2022). This may stem in part from the speed with which many of the BG programs were implemented, leaving little time to properly calibrate a risk-based fee structure. Ireland’s CIFS, one of the few BG programs using a risk-based fee, targeted a small number of specific institutions rather than the banking system or financial system as a whole. BG programs that used risk-based fees sometimes relied on general determinations of risk. In Finland, for example, the risk-based fee was determined by the GGF’s board based on its assessment of the riskiness of an institution’s activities. In contrast, AG and DG programs more consistently tied fees to specific measures like credit default swap spreads or credit ratings. Only Korea’s program explicitly linked fees with an institution’s credit rating.

Figure 7: BG Program Fees

Source: Authors’ analysis.

How and how quickly would creditors be protected in the event of a failure?

For guarantees of any type to be effective, creditors must be confident that they will have full and prompt access to their guaranteed funds (McNamara, Feldberg, et al. 2020; McNamara, Kulam, et al. 2022). Delayed payments undermine confidence in guarantees, particularly if that delay also imposes losses on creditors. This was the case in Ecuador, where delays in guarantee payouts amid rising inflation and a depreciating local currency contributed to continued runs on the banking system (Jácome H. 2004). To avoid this outcome, some BG programs specified the period of time during which any guarantee payouts had to be made. In Thailand, for example, the Financial Institution Development Fund (FDIF) established a 30-day period for the payment of funds in the event of an institution’s insolvency.

Cash payouts were not the only mechanism by which creditors could be protected as part of a BG program, however. Purchase and assumption transactions—in which a healthy institution would take on the liabilities of a failed bank—were a common approach in the event of insolvency. As discussed in Key Design Decision No. 3, Legal Authority, the lack of an adequate framework for pursuing alternatives to cash payouts in the event of insolvency can hurt the efficacy of BG programs.

A key consideration for BG programs introduced in the context of foreign exchange pressures is how foreign creditors will be protected. Payment in domestic currency (an approach adopted in both Thailand and Indonesia, for example) involves the risk of loss to foreign creditors. But payment in foreign currencies may undermine the credibility of the guarantee based on the government’s foreign reserves position. Given this difficulty in protecting foreign creditors, it is perhaps not surprising that BG programs appear to be less effective when foreign exchange pressures are present, as will be discussed in the Conclusion.

Were there other requirements for eligible institutions or creditors?

Crisis interventions often require the institutions receiving assistance to agree to certain restrictions as a condition for participation. Typically, these restrictions seek to address moral hazard concerns or to otherwise make the interventions more politically palatable. DG programs often had restrictions on balance-sheet growth, dividend payments, and executive compensation (McNamara, Feldberg, et al. 2020). However, these types of restrictions were not a significant feature of AG programs (McNamara, Kulam, et al. 2022).

The BG programs studied are more similar to AG programs in this respect. It could be that policymakers deemed specific restrictions overly burdensome because BG programs, like AG programs, were intended to be more broad-based and were often mandatory. In lieu of specific restrictions, policymakers often cited heightened supervision as a development that would accompany BG programs to address concerns such as moral hazard. This approach to addressing moral hazard in the BG context requires that a jurisdiction have adequate capacity to engage in such supervision, which was not always the case.

Those BG programs that did have other restrictions typically focused on addressing behaviors that a guarantee could promote. Ecuador, Indonesia, and Thailand all imposed policies that sought to address the incentive of insured banks to compete for deposits with high interest rates. Such competition could encourage depositors to shift funds to banks that paid higher interest rates, resulting in further contagion in a banking sector already facing heightened liquidity demands. Also, guaranteeing those interest payments could ultimately be expensive for the government, if banks offering high interest rates later required government bailouts. By imposing caps on the rates of interest that could be paid on deposits, policymakers sought to limit this form of moral hazard.

Indonesia also limited loan growth and Thailand capped foreign borrowing. In Denmark, there was a ban on dividends, stock buybacks, aggressive marketing, and unspecified risky behavior. It is perhaps not surprising that Indonesia and Denmark imposed among the most significant restrictions on participating institutions given that their BG programs were the only ones that were voluntary.

In at least one case, the absence of other restrictions seems to have resulted in increased risk-taking by institutions. In Korea, policymakers did not set an interest rate cap on deposits and weak banks began offering higher rates to attract deposits. As discussed in Key Design Decision No. 9, Eligible Liabilities, this ultimately caused authorities to exclude excess interest from coverage under the guarantee (Baliño and Ubide 1999).

How long was the BG program intended to last? Did it ultimately end as originally scheduled?

Most authorities intended their BG programs to function only as long as crisis conditions persisted. This raises the question of when, and under what conditions, the programs would end. Scaling back a BG program prematurely may reaggravate the liquidity pressures policymakers intended the program to address. Allowing it to continue too long into the chronic phase of the crisis increases the risk of moral hazard. Decisions about when to terminate a BG program must balance these considerations.

Figure 8 indicates both the originally intended and actual durations of the BG programs. Given the need to determine when conditions are right to end a BG program, several of the programs either did not specify an end date in advance or provided a more general statement of intended duration rather than something firm. The potential risk with an open-ended approach is that policymakers will ultimately allow the program to remain in place beyond the point where it is no longer necessary. This may have been the case in Finland, where banks had stabilized by 1995 but the guarantee remained in place until the end of 1998 (Borio, Vale, and von Peter 2010).

Figure 8: Intended and Actual Durations

All BG programs ultimately gave way to permanent, but more limited, deposit insurance schemes. As shown in Figure 8, while some of these deposit insurance schemes pre-dated the BG program in question, in most instances authorities specifically adopted such schemes as a means of replacing the BG programs. Similarly, the AG programs we studied in our earlier series typically evolved into permanent deposit insurance schemes. Kane and Klingebiel (2004) noted that one of the biggest challenges in introducing a blanket guarantee is how to then cut it back. A sudden shift from the full guarantee provided by a BG program to a far less expansive guarantee could be destabilizing. For that reason, several of the BG programs employed a phased approach to move away from a blanket guarantee. This typically involved a series of pre-defined points at which the amount of coverage provided or the liabilities that were eligible for coverage would be scaled back until the desired non-crisis level of coverage was reached.