When choosing the currency in which to denominate financial contracts, agents compare the expected volatility of the real exchange rate with that of inflation (see Ize and Levy Yeyati, 2003). The persistence of financial dollarization after inflation has stabilized may reflect continuing fears of a regime collapse and a return to high and unstable inflation.

Instead, a run against local currency bank instruments, to the extent that it is associated with a flight to foreign currency rather than local currency cash, can be stopped through an exchange rate overshoot (an asset market price adjustment) that immediately increases their expected yield against dollar assets.

The scope for “fiscalizing” dollar liquidity support, that is, for giving banks in distress domestic dollar public securities instead of foreign assets, is generally constrained by the limited market acceptance and liquidity of such instruments in times of crisis. While Argentina attempted to follow this route, through a scheme that allowed local banks to obtain liquidity abroad against public (Bonex) bonds, the scheme fizzled during the recent crisis as conditions in Argentina worsened. The size of the liquidity fund was progressively reduced by foreign banks, and access to what remained of the fund was limited as the secondary market price of Bonex bonds collapsed.

In Uruguay, the run started with Argentine depositors and later spread to Uruguayans as unfolding events in Argentina were relentlessly shown on local television stations. In Paraguay, deposit withdrawals were exacerbated by the closure of a major local bank that failed in the wake of the difficulties experienced by its Uruguayan-Argentine parent company.

In Argentina, the run extended to dollar deposits as expectations of a freeze or forced dedollarization became prevalent. (Note, however, that the abrupt decline in dollar deposits that took place in early 2002 reflects the compulsory change of these deposits into pesos—pesification—rather than deposit withdrawals.)

Moreover, dollar deposits are, on average, substantially larger than local currency deposits and fewer in number. As a result, they are more vulnerable to rumors and portfolio reallocations by a few large depositors. In Bolivia, for example, nearly half of all dollar deposits are above US$100,000 and are held by fewer than 1 percent of total depositors.

See Lane and others (1999) for examples drawn from the Asian banking crises and Garber (1996) for the case of Mexico.

As shown in Figure 4, however, the cross-country relationship between loan dollarization and deposit dollarization is generally less than proportional. A 10 percent increase in foreign currency deposits results, on average, in a 7.3 percent increase in foreign currency loans. This asymmetry was first detected by Honohan and Shi (2002).

Real dollarization, as measured by the pass-through of exchange rates on prices, is correlated with financial dollarization but is generally much lower. A regression analysis, based on data provided by Choudhri and Hakura (2001), suggests that the elasticity between financial dollarization and real dollarization is only about 0.25 (see Figure 5; similar evidence is reported by Honohan and Shi, 2002). This broad-brush empirical evidence is supported by casual evidence in highly dollarized countries, such as Bolivia and Peru, which indicates that the vast majority of wages continues to be paid in local currency, with only few exceptions (such as for some top executives). Nonetheless, dollar indexation appears to have made inroads over the years. In Bolivia, for example, it now affects most utility prices, pensions, some elements of the tax system, accounting standards, and some supplier contracts. In Peru, a number of services, including residential and commercial leases, real estate, professional services, and insurance premiums, are priced in dollars.

Figure 6 offers suggestive evidence. Levy Yeyati, Sturzenegger, and Reggio (2002) provide a formal empirical analysis.

See Tornell and Westermann (2002).

While some countries, for example, Bolivia, have well-developed dollar monetary instruments, their use for counter-cyclical policy is severely constrained. During periods of boom, the cost of sterilizing capital inflows in the presence of substantial country premiums becomes, in practice, rapidly insurmountable. During downturns, relaxing monetary policy mostly induces capital outflows and losses of international reserves.

The provision of (implicit) public insurance is justified by the government’s inability to precommit itself against bailouts, or its unwillingness to do so, due to the immediate stimulating impact on credit and economic activity of “guaranteed” access to cheap dollar financing. See Burnside, Eichenbaum, and Rebelo (2001).

See, among others, Mishkin (1997); Obstfeld (1998); Dooley (2000); and Burnside, Eichenbaum, and Rebelo (2001). Moral hazard is not the only factor underlying socially excessive levels of dollarization. Other factors discussed in the literature include market failures associated with debt defaults and inferior market equilibria derived from policy endogeneity. See Aghion, Bacchetta, and Banerjee (2001); Chamon and Hausmann (2002); and Broda and Levy Yeyati (2003). While economic agents’ limited foresight could also be invoked to justify borrowers’ preference for apparently cheaper dollar loans, the asymmetric behavior of depositors (who prefer to hold apparently lower-yielding dollar deposits) suggests that other factors are at play.

The dollarization of domestic public debt largely reflects factors similar to those driving financial intermediation. Public administrations view the probability of a large devaluation as remote under their current tenure, discount the cost to subsequent administrations of dealing with a currency and debt crisis, or choose to borrow in foreign currency to signal their commitment to a stable exchange rate. Incentives to borrow in dollars are exacerbated in heavily dollarized economies by the high cost of issuing large volumes of domestic currency debt (e.g., of not accommodating the public’s preference for dollar-denominated debt).

While this section deals primarily with liquidity issues, solvency issues in the aftermath of a banking crisis and a large depreciation in a highly dollarized country may also constitute major obstacles in view of the widespread losses to which banks are likely to be exposed. However, differences in relation to managing banking crises in nondollarized countries are more quantitative than qualitative. Thus, no fundamental modifications in policy response seem to be necessary. A broader discussion of such issues is contained in Hoelscher and Quintyn (2003).

Notwithstanding the crucial role played by blanket guarantees in stopping bank runs, a full assessment of their costs and benefits needs to be mindful of their moral hazard implications.

Where a funded deposit insurance system is in place, partial coverage in foreign currency may be available from accumulated funds. However, deposit insurance funds will not generally provide sufficient backing for blanket guarantees.

Under a crisis situation, however, banks’ financial conditions are often unknown. In the context of a forced securitization, therefore, bank securities might be offered as a voluntary option to government bonds, possibly along with a voluntary option to convert them into bank equity.

Causality may also apply in the opposite direction, however, as tighter prudential norms on foreign exchange lending in these countries may have contributed to maintaining a lower level of dollarization.

When a bank with X percent of its liabilities denominated in dollars is allowed to have a long structural dollar position equivalent to X percent of its capital, this is equivalent to separating the bank into a “dollar bank” (accounting for X percent of the total) and a “local currency bank” (accounting for 1 – X percent of the total), each with its proportional share of the bank’s total capital. Under such an arrangement, the bank’s capital-to-asset ratio becomes fully immune to exchange rate fluctuations.

Not all currency boards have followed this approach, however. For example, Hong Kong maintains strict limits on all foreign exchange exposures, including against the U.S. dollar, to which the currency board is pegged.

In Argentina, during the Tequila Crisis, reserve requirements were lowered from 43 percent to 20 percent. In Peru in 1998, following a significant recall of external credit lines to banks, the central bank reduced foreign currency marginal reserve requirements from 45 percent to 20 percent.

In Bolivia, time deposits with longer than two years’ maturity are exempt from the 12 percent reserve requirement and early withdrawals are prohibited. These deposits are actively negotiated in the stock market, at discounts that can be significant when liquidity market conditions are tight.

In 1985, as part of a stabilization package, Israel introduced a minimum holding period of one year on new dollar-linked PATAM deposits, reducing their attractiveness relative to other indexed financial instruments (see Bufman and Leiderman, 1995). PATAM deposits were subsequently phased out and replaced by alternative, dollar instruments (PATZAM and PAMAH), which are subject to decreasing reserve requirements as their maturity increases but are otherwise unrestricted.

These normally require the definition of a liquidity management strategy, based on liquidity gaps and stress scenarios (whose parameters may be provided by the supervisor).

In addition to holding large international reserves, preapproved contingency lines from foreign banks have been used to augment possible liquidity support. For example, Argentina’s preapproved credit line from foreign private banks reached around US$6.1 billion in 1996. However, as explained above, the line proved to be mostly unavailable when it was most needed.

The moral hazard failure described above is compounded by fear of floating, as the monetary authorities, held hostage by high financial dollarization, may maintain a rigid exchange rate against all odds, thereby increasing, in the end, the risk of a catastrophic collapse.

Indeed, the recent Ecuadoran experience, where sharp limits on liquidity support introduced in the wake of full dollarization led banks to accumulate large liquid asset holdings abroad, provides an interesting illustration of the impact of removing moral hazard. Along similar lines, Gonzalez-Eiras (2000) found that the introduction in Argentina of a credit contingency line, to expand liquidity support to banks, led domestic banks to reduce their domestic liquidity relative to foreign banks (which were less dependent on such support because of their increased reliance on support from parent banks).

For example, in Mexico (where dollar deposits are restricted), banks lend in “synthetic” dollars, based on peso funding and forward market hedging.

Unlike capital requirements, general provisions smooth out the credit cycle by providing a “speed bump” in times of rapid credit growth and a “capital adequacy buffer” in the downward phase of the credit cycle. General provisions are under discussion or have already been introduced by a number of countries (including Austria, Australia, Croatia, France, Greece, Portugal, and Spain), notwithstanding a lack, so far, of international agreement on their use. Objections have come from tax authorities (because of revenue losses if tax deducibility is allowed) and some accounting bodies (on transparency grounds).

The currency in which a firm denominates its products does not provide, as such, any guarantee that its dollar-based income will be robust to exchange rate movements.

This approach would implicitly recognize that banks are not expected to be prepared to withstand real depreciations of exceedingly large size and small likelihood. The definition of an appropriate value-at-risk cutoff would need to be based on a macroeconomic and statistical analysis of shocks and needed exchange rate responses.

Even when capital charges are set on foreign exchange positions, it may still be necessary to impose limits on open foreign exchange positions as a percentage of capital. The latter are equivalent to high marginal capital requirements.

Introducing a minimum holding period on dollar deposits can be viewed as an extreme variation of differentiated liquidity requirements.

Box 4 presents such a calculation for Peru.

Similar benefits might be given under similar conditions to the fully owned subsidiaries of large international banks if the subsidiary meets minimum commitment-enhancing criteria, such as bearing the name of the parent, in addition to benefiting from legally binding assurances of support issued by the parent bank.

The impact of increased foreign bank participation on the resilience and depth of banking systems is currently at the center of an active ongoing debate. Note also that dollarization itself may already give foreign banks a competitive edge. See Swoboda (1968) for an early discussion of this issue.

See Broda and Levy Yeyati (2003).

To further limit the moral hazard associated with the support of dollar deposits, where deposit insurance is in place, higher premiums may be required on dollar deposits than on local currency deposits.

Admittedly, the scope for developing secondary markets for private financial instruments is more limited in less-developed financial systems. However, as Bolivia’s experience suggests (where long-term bank deposits are actively traded in the stock exchange), secondary market trading of private instruments can develop even in relatively simple financial systems.

The scope for (and benefits of) derivative transactions such as foreign exchange forwards is limited in nascent markets by the lack of counterparties on the supply side of the market. Typically, banks that supply the forwards cover their position by buying foreign exchange spot, thereby pressuring the market in the same way as a straightforward purchase of foreign exchange. When forwards are supplied by unhedged domestic market participants, counterparty risk increases as the capacity of counterparties to meet the terms of the contract under a large devaluation becomes uncertain.

Mishkin and Schmidt-Hebbel (2001) provide broad cross-country evidence in support of inflation targeting as a way to consolidate confidence in the currency. The fact that countries that have recently adopted an inflation-targeting framework, such as Mexico, Poland, Hungary, and the Czech Republic, have made progress in containing or reducing dollarization also suggests that a strong anti-inflation commitment may have benefits for dedollarization. However, it is too early to derive strong conclusions from these experiences. Peru, which formally adopted an inflation targeting framework in 2001, has seen some encouraging signs pointing to increasing confidence in the local currency but has yet to experience a significant decline in financial dollarization. Moreover, some countries, such as Bosnia and Herzegovina, Lithuania, and Slovenia, also experienced declines or halts in dollarization after having adopted strong stabilization policies, although these were not backed by formal inflation targeting.

See Baliño and others (1999).

Price indexation can also provide a superior alternative to indexation through floating interest rates. While floating rates limit inflation risk, they exacerbate interest rate risk. The use of floating-rate mortgages during the late 1990s in Colombia, where dollarization is prohibited, was largely responsible for the financial deterioration of the mortgage industry, as the very high interest rates associated with the defense of the currency during 1998–99 were passed on to mortgage holders, leading them to default en masse on their obligations. Brazil during the 1980s and early 1990s provides another interesting example of the pitfalls of limiting dollarization by regulation. The widespread use of the overnight interest rate to index most financial instruments gave rise to “indexed money,” undermining monetary policy and worsening inflationary inertia.

Asymmetric conversions into price-indexed instruments can otherwise give rise to a mismatch in banks’ balance sheets that exacerbates their vulnerability to interest rate risk. In Colombia, for example, floating-rate mortgages were converted into price-indexed instruments following the 1998–99 banking crisis. While the switch did reduce borrowers’ exposure to interest rate risk, it magnified that of banks as most banks’ liabilities continued to be expressed in pesos at very short maturities or with floating interest rates.

In Bolivia, for example, the recently introduced Unidad de Fomento de la Vivienda (UVF), which is indexed to the consumer price index, reflects a consumption basket that is heavily weighted toward basic staple goods. However, the ownership structure of bank deposits, which reflects the country’s income distribution, is biased toward the higher income groups (nearly 50 percent of deposits are above US$100,000), whose consumption baskets are intensive in imported goods.

As long as proper arrangements are in place to ensure safe settlements and limit the central bank’s exposure to settlement risk, settling in dollars on the central bank’s books is basically equivalent to settling in local currency from a risk management perspective.