Dealing with bank runs in dollarized economies is both more difficult and subject to greater risks than in other cases.16 The absence of a lender of last resort has a potential to make dollarized systems more prone to runs, and runs more difficult to stop when they occur. A strategy to address runs cannot rely to the same extent on the provision of liquidity and on deposit guarantees, the two key elements identified for other cases. The additional constraints call for more attention to financial sector soundness. In the cases where, in spite of such added vigilance, bank runs do occur, theory and country experiences suggest that the authorities may need to resort earlier and more frequently to administrative measures that restrict the availability of deposits, or else seek international support on a larger scale (Table 5).
Provision of Emergency Liquidity
The limited availability of dollar liquidity calls for a carefully designed strategy on liquidity provision. Dollar assistance can be provided as long as sufficient dollar resources are available. Sources of liquidity can include high international reserves prior to the run (e.g., Argentina in 1995); contingent facilities such as swaps and repos; or international financial support (e.g., Mexico in 1995). Drawing down such resources can, however, have costs in terms of credibility and limitations on macroeconomic policymaking. Limited availability of dollars raises the question of whether to shift assistance in part or in full to local currency (e.g., Bulgaria in 1996, Russia in 1998, and Ecuador in 1999). This option avoids the need for other administrative measures and circumvents constraints on liquidity assistance because the central bank can print local currency. However, the expansion in the money supply and the currency mismatch are likely to result in a combination of loss of international reserves, exchange rate depreciation, and inflation. Moreover, currency denomination of the liquidity is inconsistent with depositors’ preferences. Providing local currency liquidity is therefore likely to be less effective in restoring depositor confidence than liquidity provision in dollars would be.
Depositor Protection
The lack of dollar-backing sharply limits the scope for a dollar-based blanket guarantee. Depositor protection, in particular through a blanket guarantee, has often been important in addressing systemic bank runs.17 In a dollarized economy, funding constraints emerge, similar to those discussed above in the case of liquidity support. A guarantee of deposits in foreign currency can be effective as long as sufficient dollar backing is available. In some countries, most notably the transition economies, banking systems were small enough that a full government-backed blanket guarantee of foreign currency deposits may have been an option. In most others though, reserve and fiscal constraints render a blanket guarantee issued in foreign currency not credible.18
Few alternatives to blanket guarantees are available. A blanket guarantee in local currency is unlikely to instill depositor confidence, because depositors will demand their original dollars rather than the local currency counterpart. Moreover, the announcement of a local currency guarantee may aggravate the crisis if depositors fear that the banking system does not have sufficient dollar resources and, therefore, seek to buy dollars in the exchange market. Guaranteeing deposits in local currency at a fixed exchange rate (Russia in 1998) implies a loss to depositors when the exchange rate depreciates and depositors are therefore likely to withdraw their funds and convert them to foreign exchange as soon as possible. Another option would be to transfer deposits in failed banks to a public or private bank that is perceived to be sound and with significant foreign exchange holdings, backed up by corresponding holdings of central bank securities to ensure liquidity. Under the conditions of a systemic run, such banks only exist in rare cases.
Administrative Measures
Given the added constraints of dollarization for liquidity support and depositor protection, administrative measures—such as securitization of deposits, the extension of deposit maturities, or the imposition of bank holidays or other restrictions on deposit withdrawals—are more likely to be needed than when a crisis occurs in a local currency environment. These measures should be adopted under the general principles of transparency, least fiscal cost, and uniformity of treatment. However, where banks’ capacity to withstand liquidity shocks is clearly uneven, with some banks benefiting, for example, from open-ended support from their foreign parents, some countries have imposed administrative measures selectively, as in the recent Uruguayan banking crisis. The risks of asymmetric treatment of banks need to be balanced against the benefits of shielding at least part of the banking system from the adverse impact of altering financial contracts unilaterally.
Securitizing deposits is preferable to a freeze or an outright default. A securitization of deposits (e.g., Argentina in 2002, Ecuador in 1999) may be less damaging to depositors than a deposit freeze or an outright default because a secondary market in such instruments can enhance their liquidity, albeit at a loss if the bonds trade at a discount and are sold for cash before maturity. There are a variety of modalities for swapping bank deposits into bonds, and possible trade-offs need to be considered when designing the strategy. In particular, depositors could receive a government bond or a bond drawn on the commercial bank. Commercial bank securities may be the appropriate instrument if the bank is considered viable in the medium term.19 To the extent the bank holds government paper, it could also swap deposits for government bonds out of its portfolio, thereby avoiding net debt creation. If neither condition is fulfilled, the government can assist by issuing new bonds; the usefulness of this option depends on medium-term fiscal sustainability.
Securitization should be accompanied by steps to mitigate its impact on financial intermediation, the payments system, and economic activity. Specific steps could include (1) making securities transferable and stimulating a secondary market in bonds; (2) issuing securities in sufficiently small denominations; or (3) making securities for certain transactions, such as purchase of government assets, transferable. While these steps would strengthen demand for such instruments and strengthen the secondary market, the extent to which they can be used is limited by the government’s cash needs. To avoid circumvention and loss of credibility, there should be few exceptions from the general policy concerning securitization of deposits.
Extending deposit maturities may be the only choice left if deposits cannot be securitized. This option—a form of securitization with nontradable bank instruments—was implemented, for instance, in both Argentina and Uruguay in 2002. While deposits with extended maturities are frozen in the banking system and not converted into negotiable instruments, measures should be considered to improve the functioning of the payments system under such a freeze. Frozen accounts should have full mobility within the banking system, and limited weekly or monthly withdrawals should be allowed. The authorities should be prepared for the fact that mobility within the system may lead to flight-to-quality as depositors move frozen deposits to the strongest banks in the system.
Limitations on deposit withdrawals should be clearly defined, rather than being left open ended. Freezes without specified minimum weekly or monthly withdrawal amounts, or outright bank holidays, are an unstable solution to deposit runs and should be avoided. If established, they should only be in place for limited time periods to buy the authorities time to work out a permanent solution.
Nominal losses to depositors should be considered as a last resort. Such action should only be taken if all the above options cannot be implemented or have failed. Nominal losses will make reintermediation in the financial system more difficult to achieve, even in the medium run.