Andrew Berg

Dollarization manifests itself in two forms: one voluntary and the other largely involuntary. The former type, called full or de jure dollarization, occurs when a country adopts a foreign currency, often the U.S. dollar, as its sole legal tender. The involuntary form, called partial or de facto dollarization, is less under the direct control of the authorities and occurs when a foreign currency circulates alongside a national currency, with bank deposits and loans possibly also denominated in a foreign currency. Most researchers place full dollarization within the realm of exchange rate regime choices in the economic literature, while partial dollarization is typically analyzed as an outcome of disorderly macroeconomic conditions. These two branches of analysis are nonetheless closely linked, as full dollarization is a more attractive choice for countries that find themselves largely dollarized de facto. This article surveys recent research at the IMF on the theoretical and empirical aspects of both partial and full dollarization.

Much of the research on partial dollarization has focused on its underlying causes. Savastano (1996) emphasizes that, while macroeconomic instability is the cause of dollarization, its manifestation depends on the institutional framework. Countries that did not allow dollar deposits, for example, observed more capital flight and increased use of foreign currency, whereas countries with sufficiently flexible financial systems, such as Brazil, saw a proliferation of indexed domestic currency instruments, instead of dollarization.¹ A central feature of de facto dollarization is that it tends not to abate, even after the macroeconomic instabilities that caused it have been brought under control. Mongardini and Mueller (2000) find some econometric evidence of such a “ratchet effect” in foreign currency deposits in the Kyrgyz Republic, but not in cash dollars in circulation (for which they have some survey data).²

The literature on partial dollarization makes a distinction between currency substitution—dollarization of money as a means of payment—and asset substitution, which refers to dollarization of stores of value. Earlier research, such as Agénor and Khan (1996), focused on currency substitution, examining the rate of return on foreign currency in money demand functions.³ As Savastano (1996) notes, however, most of the available data is on interest-bearing, foreign-currency time deposits, for which a focus on asset substitution makes more sense. Moreover, the Asian crisis motivated increased analysis of dollarized loans as well as other financial deposits; large quantities of loans were dollar-denominated and this had important implications for the economic effects of exchange rate fluctuations. More recent research on dollarization has thus concentrated on asset substitution.

Ize and Levy-Yeyati (1998) show that the degree to which the financial system is dollarized depends on relative volatilities of the inflation rate (more volatility makes foreign currency deposits less risky) and the real exchange rate (more volatility makes these deposits more risky in terms of domestic prices). Ize and Levy-Yeyati also explain the persistence of dollarization after stabilization in terms of the fact that real exchange rate volatility, in many cases, decreases faster than inflation volatility.⁴ Catão and Terrones (2000) present a model of the banking system which emphasizes how banking and credit market imperfections determine deposit and loan dollarization.⁵ Mourmouras and Russell (2000) show how, in the absence of interest-bearing bank deposits, foreign currency surrender requirements may lead to smuggling as a way for residents to accumulate cash dollars.⁶

The implications of partial dollarization for monetary and exchange rate policies and financial supervision is critical to the operational work of the IMF. Berg and Borensztein (2000) conclude that a high degree of currency substitution strengthens the case for fixing the exchange rate, while dollarization as asset substitution has various offsetting effects.⁷ Ize (2001) argues that inflation targeting may be feasible even in the presence of substantial currency and asset substitution.⁸ Baliño and others (1999) emphasize that dollarization—understood as asset substitution—has implications similar to those of capital account openness in general.⁹ Poirson (2001) concludes that countries with a high degree of partial dollarization are, in fact, more likely to choose a more rigid exchange rate regime.¹⁰

IMF researchers have analyzed a variety of country experiences with partial dollarization. Unteroberdoerster (2002) discusses the case of Vietnam; Fritz-Krockow and others (2001) look at Haiti; and Rumbaugh and others (2000) analyze dollarization in Cambodia. Zamaróczy and Sa (forthcoming) estimate the amount of cash dollars in circulation in Cambodia and conclude that the country is almost completely dollarized. Lizondo and others (2001) analyze monetary policy and the advisability of inflation targeting in Peru.¹¹

Interest in full dollarization has grown sharply in recent years, both out of dissatisfaction with the alternatives and because several countries, including El Salvador and Ecuador, joined Panama among the ranks of sizable, fully dollarized countries. Berg and Borensztein (2000) analyze the costs and benefits of full dollarization as compared with its closest alternative, a currency board. They attempt to quantify the tradeoffs in the case of Argentina, in particular the cost of foregone seignorage and possible benefits from lower borrowing costs. They observe, however, that the potentially most important considerations—notably, the loss of the “exit option” to devalue in the face of major shocks, and the advantages of deeper integration—are harder to evaluate. They conclude that two groups of countries are most likely to find dollarization attractive: those already highly integrated with the United States (or other country whose currency is to be adopted) and those already highly dollarized de facto.¹² Calvo and Reinhart (2000) also point out that countries that are already substantially dollarized de facto will lose little in going to full dollarization.¹³

Several papers have looked more closely at the various costs and benefits of dollarization. Bogetić (2000) argues that growing partial dollarization and financial development imply that seignorage losses from dollarization would be small in many cases.¹⁴ Parsley and Wei (2001) find that currency boards, and especially dollarization, strongly promote goods market integration, far beyond what is directly associated with exchange rate stability alone.¹⁵

Dollarization is sometimes seen as an irreversible decision but Liberia, one of only two countries with a long history of dollarization (the other is Panama) reintroduced its own currency in the 1980s. Indeed, Abrams and Cortés-Douglas (1993) provide what amounts to a manual for introducing a new currency, based on the experiences of countries that “deruble-ized” after the breakup of the Soviet Union.¹⁶

As Mishkin and Savastano (2000) emphasize, actual experiences with dollarization are limited, and the long history of dollarization in the most important case, Panama, is mixed.¹⁷ Bogetić (2000) reviews the experience of Panama and concludes that it has been broadly positive. De la Torre and others (forthcoming) describe the deep banking and exchange rate crisis that led to Ecuador’s January 2000 move to full dollarization and emphasize dollarization’s stabilizing role; Offerdal and others (2000) provide an early, cautiously positive, assessment of the outcome.¹⁸ The addition of countries like Ecuador and El Salvador to the ranks of fully dollarized countries will provide important material for future research.