11 De-dollarizing the Hard Way

Adrián Armas, Eduardo Levy Yeyati, and Alain Ize
Published Date:
July 2006
  • ShareShare
Show Summary Details
Daniel C. Hardy and Ceyla Pazarbasioglu1 

11.1 Introduction

There is wide consensus that partials dollarization can magnify a country’s vulnerabilities (e.g.. Guide et al, 2004).2,3 These vulnerabilities relate to the country’s balance of payments, the banking sector and its borrowers and also fiscal sustain-ability. There is also broad consensus that dollarization is largely the product of macroeconomic instability or the threat thereof, which has led to a weakening of ‘faith’ in the national currency.4 Hence, dollarization may increase the vulnerabilities of a country already prone to exogenous shocks or misguided policies. How vulnerabilities are affected depends on the form of dollarization. A distinction is usefully made between ‘liability dollarization’, where banks have liabilities to savers in the form of foreign currency deposits (FCDs); ‘asset dollarization’, whereby banks have extended credit to residents (including perhaps the government) in foreign currency; ‘real dollarization’, which implies that contracted amounts even for non-financial sector transactions are denominated in foreign currency, even if transactions are settled in local currency; and ‘payments dollarization’, under which payments are settled in foreign currency. Various combinations of these types of dollarization are possible, but ‘liability dollarization’ is probably the most common, and ‘real’ and ‘payments’ dollarization are viewed as more deep-seated forms of dollarization.

Partial dollarization may not only heighten vulnerabilities, but also brings with it other advantages and disadvantages (de Nicoló, Honohan and Ize, 2005; Ize, Chapter 2 in this volume). Nonetheless, most commentators and national governments view high dollarization as undesirable. Therefore, much of the academic literature concentrates on how to minimize the bad effects of dollarization, for example, by modifying prudential rules and adjusting the conduct of macroeconomic policy, and on how to wean the economy off dollarization (e.g., Ize and Powell, 2004; Herrera, 2004). Recommended policies for de-dollarization are designed to increase the relative attractiveness of the national currency, so that economic agents will voluntarily switch back to it. However, although highly dollarized countries, especially in Latin America, were able to bring down inflation and establish independent central banks and credibility, the success with de-dollarization has been limited. It is recognized that voluntary de-dollarization is likely to be a drawn-out process because dollarization is characterized by hysteresis, that is, the process of dollarization creates new institutions and mechanisms and awareness, and it is not possible to go back to the status quo ante (e.g., Havrylyshyn and Beddies, 2003; Uribe, 1997).5 The economy may reach a locally stable equilibrium characterized by a high degree of dollarization (Ize, Chapter 2 in this volume). Successful cases of gradual, voluntary de-dollarization are rare (Box 11.1).

Given the difficulties encountered in achieving gradual, voluntary de-dollarization, consideration may be given to using a more drastic and interventionist approach, namely the compulsory conversion of foreign currency denominated claims into local currency. In this connection, one can ask such questions as

Box 11.1Instances of gradual de-dollarization

Reinhart, Rogoff and Savastano (2003) identify Israel, Mexico, Pakistan and Poland as the only cases (among 86 countries) of large and persistent reversals of deposit dollarization (measured by the share of FCDs in total deposits) that had negligible effects on financial intermediation. Galindo and Leiderman (2003) identify Chile, Israel and Poland as successful countries. Pakistan is an instance of forced de-dollarization (explained in detail below). In all these countries, the de-dollarization strategy was part of a broader stabilization programme, which included introducing new assets with alternative forms of indexation (in the case of Israel and Chile) or high ex post real interest rates on local currency assets (Mexico and Poland).1,2 Starting from levels of deposit dollarization above 50 per cent in the mid-1980s, deposit dollarization in Israel fell to less than 10 per cent by the mid-1990s after a decade of low and stable inflation and a backdrop of fiscal consolidation.

Egypt constitutes another successful case of gradual de-dollarization, where the share of FCDs in total deposits declined from 56 per cent in 1991 to 22 per cent in 1999 and, after remaining stable for a few years, further declined to 18 per cent by 2004.3 Until the early 1990s, Egypt’s banking system was characterized by interest rate controls, credit ceilings and differentiated reserve requirements (25 per cent on domestic currency liabilities and 15 per cent on FCDs). In early 1991, the authorities began a series of fiscal and monetary reforms designed to reduce inflation and enhance market forces in the allocation of credit. Bank lending and deposit rates were liberalized, credit ceilings on bank lending were removed, reserve requirements were reduced and unified across currency denominations.4 The programme led to a number of positive effects: inflation declined sharply and real interest rates in domestic currency became positive. The share of FCDs declined as confidence in the reform programme increased.

1. See Bufman and Leiderman (1995) for the case of Israel and Chopra (1994) for Poland.2. Ex ante real interest rates may have been lower insofar as probability was attached to the failure of the stabilization programme and a return to higher inflation. Local currency assets in some cases may have been depressed by regulatory restrictions prior to the stabilization programme.3. Egypt is not identified in Reinhart, Rogoff and Savastano (2003) or in Galindo and Leiderman (2003), as the former uses a different definition of de-dollarization (FCDs over M2) and the latter uses a threshold of 20 per cent.4. See Alexander, Baliño and Enoch (1995) and Handy (1998) and others. The reform programme also included introduction of prudential regulations for banks and strengthened bank supervision (see IMF, 2004a).
  • What circumstances have led to forced de-dollarization?

  • What are the costs of forced de-dollarization and how can they be minimized?

  • When will forced de-dollarization achieve permanent de-dollarization and not be reversed once restrictions are lifted?

The chapter attempts to shed light on these issues. The chapter concentrates on the case where vulnerabilities have already been incarnated in a large adverse shock and, as a consequence, the authorities have required the ‘pesoization’ of some foreign currency-denominated assets or other obligations. Gradual, voluntary de-dollarization is no longer an option, so de-dollarization ‘the hard way’ is viewed as the best available policy. However, some consideration is also given to the possibility of forced de-dollarization in more normal times, even though, to the best of our knowledge, this has never been carried out in practice.

Since instances of forced de-dollarization are heterogeneous and limited in number, the main evidence comes from analysis of country case studies. In the next section, several instances are reviewed that can be characterized as failed forced de-dollarization, because the policy was quickly abandoned or because the economy soon re-dollarized. Other instances can be viewed as relative success cases because the economy recovered rapidly from the shock that occasioned the forced de-dollarization, and dollarization was reduced permanently. They are reviewed in Sections 11.3 and 11.4.. Section 11.5 draws some lessons related to the questions posed above. Section 11.6 concludes.

11.2 Failed forced conversion experiences

By the early 1980s, chronic inflation and macroeconomic instability resulted in widespread dollarization of financial assets in both Bolivia and Peru, as private agents sought to protect their wealth from capital losses and were unwilling to save in assets with uncertain real returns. In November 1982, the Bolivian authorities de facto converted existing dollar deposits (which represented at the time 43 per cent of the broad monetary aggregate M2) into Bolivian pesos at the official exchange rate (Figure 11.1). The authorities also made it illegal to open new foreign currency bank accounts. Peru enacted similar measures to force conversion of FCDs into domestic currency in August 1985 (FCDs were about 58 per cent of total deposits a year earlier). In both cases, the imposition of foreign exchange controls was coupled with a large devaluation which effectively wiped out residents’ wealth measured in dollars.

Figure 11.1Bolivia and Peru: share of foreign currency deposits, 1975-2004

(in per cent)

Sources: Money and Banking Database (MTBS), World Economic Outlook (WEO) and Savastano (1992).

Following several years of extreme macroeconomic instability, however, both countries eventually allowed the opening of FCD accounts as domestic intermediation in local currency declined sharply and most deposits were channelled abroad.6 Bolivia lifted restrictions on domestic FCDs in 1985 following the hyperinflation episode. A similar measure was adopted by Peruvian authorities in September 1988.7 In both cases, these measures fostered a rapid re-dollarization of the economy, which has persisted until today despite the significant reduction in inflation.

Several lessons can be drawn from Peru’s and Bolivia’s unsuccessful experiences in combating dollarization. First, measures that attempt to reduce dollarization de facto without addressing the root causes of the private sector’s currency preferences are bound to fail (Box 11.2). FCDs held by residents in these countries reflected their attempt to save in a unit of account that had a stable and predictable purchasing power, i.e., that insured them against inflation risk. In the cases of Peru and Bolivia, however, the underlying conditions that led to dollarization in the first place – monetary instability caused by the lack of institutions that promoted monetary credibility and fiscal consolidation – were not eliminated. Thus, the banking system rapidly became re-dollarized once FCDs were re-allowed.8

Box 11.2Exchange rate changes and incentives for dollarization

An analysis of the historical distribution profile of exchange rate changes may provide some insights regarding the incentives for deposit dollarization. As shown in the figures below, in the cases of Bolivia and Peru the likelihood of large depreciations of the currency was substantially greater than that of large appreciations. This may have fuelled the perception of dollar deposits as a ‘one-sided bet’, providing strong incentives to save in foreign currency assets.

Period: pre-forced conversion

Note: The figures plot the kernel density estimates of quarterly percentage changes in the nominal exchange rate for Bolivia and Peru. The univariate kernel density estimation is performed using an Epanechnikov kernel. The sample period extends from 1970 until the last quarter before the forced conversion in each country. The return series in both countries during this period shows substantial excess kurtosis and positive skewness, indicating right fat tails and lack of symmetry.

This analysis underscores the notion that achieving de-dollarization is bound to be a protracted process; the distribution profile of exchange rate changes varies only gradually as part of a cumulative process. It follows, then, that any abrupt plan to repress the use of the dollar (like the forced conversions in Peru and Bolivia) are likely to fail, to the extent that they do not address the underlying reasons for agents’ currency preferences.

FD remained high in both Bolivia and Peru, even after success in controlling inflation, although dollarization has been declining in recent years. On the other hand, Israel and Poland were able to achieve a sustained reduction in deposit dollarization following the stabilization of their economies. Ize and Levy Yeyati (2003) use a portfolio analysis that explains this puzzle by highlighting the stochastic behaviour of prices and exchange rates as the main factors driving the hedging decisions of risk-averse households. We extend the analysis of the distribution of exchange rate variations to explain the different experiences of Bolivia, Peru, Israel and Poland (see figures below).

Patterns of exchange rate fluctuations in Israel and Poland, countries that were able to reduce dollarization, suggest a two-way exchange rate risk as well as lower probability of sharp changes in the exchange rate.

Period: 1990-2004

In contrast, in Bolivia, exchange rate fluctuations have been predominantly in the form of local currency depreciations. This evidence likely reinforces the perception that the exchange rate is ultimately flexible in only one direction, favouring dollar over peso savings.

In the case of Peru, return series suggest a high relative likelihood of a breakdown in the exchange rate system (i.e., a large-scale devaluation). This implies that agents would demand a higher risk premium on peso deposits. Agents that are averse to sudden capital losses would prefer dollar assets, providing some explanation for the persistence of high dollarization despite the decline in the level and relative volatility of inflation and the existence of inflation-linked instruments.

Period: 1990-2004

The empirical evidence suggests that greater exchange rate flexibility may act as a deterrent to FD by introducing a two-way exchange rate risk. This would alter the relative risks in favour of the domestic currency, by eliminating the ‘one-sided bet’ that pegged rates or narrow band regimes often provide.

A second lesson that can be drawn from Bolivia’s and Peru’s experiences is that dollarization is typically so ingrained, that attempts to reduce it overnight may induce massive disintermediation and capital flight. Dollarization is a form of contractual adaptation and its reversal requires a certain period of time whereby agents revise their expectations on price risk and adjust their savings portfolio accordingly.

Third, attempts to overcome dollarization by dishonouring the terms of financial contracts (as was the case in Peru and Bolivia) are likely to have long-lasting effects on domestic financial intermediation, even if FCDs are later re-authorized. As noted by Baliño, Bennett and Borensztein (1999), forced conversions entail a substantial loss of government credibility, increasing the confiscation risk perceived by domestic residents. In the case of Bolivia, for example, after FCDs were reintroduced, the spread over LIBOR on domestic dollar deposits – which reached over 900 basis points in 1987 – was still over 400 points at the end of 1996. However, it is hard to separate the impact of the forced conversion from the loss of confidence in the local currency following the hyperinflation.

Fourth, attempts to reduce dollarization by banning completely the use of the dollar will not actually increase the resilience of the economy, especially in small open countries. Foreign currency assets become a natural hedge to inflation risk in the case of pass-through from exchange rate movements to prices. Thus, a certain level of dollarization is not only unavoidable, but also desirable to mitigate the currency risk in the economy (especially for those agents whose income is denominated or indexed to the dollar).

Finally, de-dollarization strategies often usefully include measures supportive of a reintermediation in local currency such as the introduction of savings instruments whose return is indexed to inflation in the domestic currency (Holland and Mulder, Chapter 10 of this volume). Measures to nurture the domestic currency as a store of value were lacking in the forced de-dollarization episodes of Bolivia and Peru during the 1980s.9

11.3 Pakistan: policy-induced dollarization and reversal

Pakistan’s experience with dollarization over the past two decades illustrates the role that both macro- and microeconomic policy can play in inducing dollarization and then reversing it. It also provides an example of the forced de-dollarization of deposits, where lending to the private sector was not heavily dollarized, and of how de-dollarization can become part of a virtuous circle of stabilization and increased confidence.

Policy-induced dollarization

Pakistan’s economic development during the 1970s and 1980s was widely perceived to be impeded by the country’s balance of payment’s constraint, even while the country was able to accumulate a substantial stock of foreign debt. At the same time, the domestic financial system was centrally directed subsequent to the nationalization of domestic banks in 1972; interest rates were set administratively, and credit was allocated according to a complex development plan. The country also suffered several bouts of political and economic instability. Faced with these uncertainties and poor financial returns in Pakistan, capital flight became substantial. Moreover, large numbers of Pakistanis emigrated, notably to the Gulf states following the 1973 oil boom; the emigrants remitted funds, but also accumulated assets abroad.

The authorities sought to tap the emigrant community for funding by allowing the opening of non-resident FCDs, first in 1973 (Husain, 2003). No questions were asked about the source of funds, and returns were tax-free. Funds could be freely withdrawn to make foreign payments. Inflows started slowly, but built up over the course of the decade (Mirakhor and Zaidi, 2004). However, domestic lending in foreign currency was not permitted; the inflow of FCDs was effectively used as a contribution to official international reserves.

Starting in late 1988, the authorities launched a general reform programme for the financial system: entry by domestic private banks was permitted; foreign banks were allowed to compete in more lines of business; some state-owned banks were privatized; prudential regulations were tightened and better enforced; interest rates and credit allocation were liberalized; and market-based government bills and bonds were introduced (see State Bank of Pakistan, Financial Sector Assessment, various issues). With some interruptions, the process has been maintained until the present, when the Pakistani financial system is considered to be relatively sophisticated, profitable and efficient (IMF, 2004b; see also Bonacorsi di Patti and Hardy [2005] on the history of reforms and the effects on the banking system).

Concurrent with the financial liberalization process, residents were allowed to open FCDs in 1991 under the same favourable conditions as applied to non-resident FCDs. The authorities seem to have been motivated mainly by their continued concern to build up international reserves or service foreign debt. Although foreign debt and debt servicing were rising, there was little concern at the time about longer-term sustainability.

FCDs rapidly gained in popularity, and the non-resident FCDs also began to grow. FCDs contributed a large proportion of the monetary growth witnessed in the mid-1990s (Figures 11.2 and 11.3). One main source of the inflows was remittances, and therefore the increase in FCDs was closely linked to earnings of expatriate Pakistanis working in the Gulf, which fluctuated with demand for their services (Hyder, 2003). The availability of remunerated, relatively safe FCDs and improved banking services led to a shift away from the use of informal remittance systems (known as Hindi systems), but in many cases transactions may have gone through the active kerb market. Although data are not available, placement in FCDs may also have been viewed as a substitute for capital flight and the retention of savings in foreign currency cash. Partly as a result, recorded remittance inflows remained broadly stable through the 1990s despite the decline in oil prices and consequent decline in earnings by Pakistanis in the Gulf, and greatly eased the balance of payments constraint facing the country (Figure 11.4; see Mirakhor and Zaidi, 2004).

Figure 11.2Pakistan: stock of FCDs

(in $billions)

Source: State Bank of Pakistan, IFS and staff estimates.

Figure 11.3Pakistan: contributions to broad money growth

(annual change in per cent of beginning of period broad money)

Source: State Bank of Pakistan, IFS and staff estimates.

Figure 11.4Pakistan: balance of payments and international reserves

(in $billions)

Source: State Bank of Pakistan, IFS and staff estimates.

Depositors were attracted to the FCDs (primarily denominated in US dollars) because they offered a ‘safe haven’ against domestic inflation and other shocks, while also yielding a relatively high return and facilitating the making of foreign payments.10 More specifically, the return on FCDs was linked to LIBOR, while returns on Pakistani rupee (PR) deposits were at or below the State Bank of Pakistan (SBP) discount rate and moved broadly in line with it. The SBP discount rate was often 10 percentage points or more above LIBOR, but the PR often suffered large step depreciations. Therefore, the total, after-tax and risk-adjusted return on FCDs was very favourable, especially for those whose consumption basket contained many imported items (Figure 11.5).

Figure 11.5Pakistan: cumulative returns on local and foreign currency assets

in local currency, December 1992 = 1)

Source: State Bank of Pakistan, IFS and staff estimates.

Banks were eager to acquire FCDs, which offered them high returns for little or no exchange rate risk: except for a limited amount of trade financing, banks were not allowed to lend domestically in foreign currency, and had to sell foreign currency acquired through FCDs to the SBP. Banks could use the rupees acquired this way to acquire high-yielding local-currency assets, such as the newly introduced Treasury bills.11 Banks could keep a closed foreign currency position on account of the forward foreign currency cover provided by the SBP: the SBP committed itself to providing forward foreign exchange cover at an administered price. While the forward premium varied and gradually increased in the course of the 1990s, it was consistently less than the ex post rate of depreciation of the PR. Hence, the Central Bank essentially subsidized the mobilization of FCDs. Furthermore, in 1986, the SBP started offering banks a dollar deposit facility yielding 17 per cent, which was far above what they could obtain on correspondent accounts abroad. The foreign banks, which lacked large branch networks with which to mobilize retail PR deposits, were especially active in attracting FCDs.

In addition, several forms of foreign currency bearer certificates were introduced over time. The bearer certificates could be bought for cash with no questions asked about the source of the funds and were tax-free. They were intended to attract flight capital and funds kept in foreign currency cash as a store of value and to evade taxes. However, the volume of foreign currency bearer certificates was small relative to the stock of FCDs or of government debt generally.12

FCD freeze and de-dollarization

The situation changed abruptly when, in late May 1998, Pakistan tested a nuclear bomb, which provoked a shut-down in access to international debt markets and funding from international financial institutions. Pakistan was also made subject to sanctions. This political shock led to a rapid outflow of foreign exchange reserves, including through withdrawal of FCDs (which reached a peak shortly before the onset of the crisis), the drying-up of capital inflows and remittances and pressure on the exchange rate.13

In response, the authorities took a number of emergency measures, including the freezing of FCDs and the reimposition of various exchange controls.14 However, after a short period, the authorities allowed frozen FCDs to be withdrawn in rupees, at the official exchange rate, which was much less favourable than the market rate.15 In due course, the authorities also offered to convert FCDs into medium-term dollar-denominated bonds, with a coupon set a few percentage points above LIBOR.16 Nonetheless, newly deposited foreign currency remained free of restrictions on withdrawals and also on interest yield, but was no longer tax-exempt. The forward cover scheme for banks was discontinued and instead the SBP imposed a special reserve requirement on FCDs. The SBP also disallowed the use of FCDs as collateral.

Since then, conditions have normalized. The SBP has been dismantling exchange controls and restrictions on the foreign currency operations of banks. Meanwhile, access to international capital markets was regained and relations with international financial institutions restored (IMF, 2004c; see also Lorie and Iqbal, 2005). Macroeconomic performance improved: real growth rates have picked up to about 6 per cent; inflation has fallen to the 3-4 per cent range (from around 10 per cent during much of the 1990s); the current account has swung into surplus; the ratio of international reserves to short-term external debt has risen from a low of 25 per cent to over 250 per cent (see also Figure 11.4); and the budget deficit has fallen from over 4 per cent of GDP to under 2 per cent. Domestic financial markets, such as the Karachi stock exchange and the market for real estate, have enjoyed something of a boom.

The stock of old FCDs eroded as depositors took out funds in deposits or bonds (Figures 11.2 and 11.3). A significant stock of new, unrestricted FCDs was built up, but never represented a large share of total deposits; demand for rupee deposits has been growing rapidly. Today, FCDs constitute a relatively minor part of investors’ assets. The financial system as a whole is on a sounder footing and offers a wide range of products.

The sustained decline in deposit dollarization can be attributable to a number of factors:

• Investors’ demand for dollar assets was discouraged. By introducing a freeze on the withdrawal of deposits in dollars, but allowing withdrawals in PRs, FCDs lost their ‘safe haven’ status. This might have discouraged some remittances from passing through the formal financial system (remittances dropped sharply in 1999-2001), but the overall flow of remittances recovered subsequent to the crisis, largely because of increased demand for Pakistani expatriate labour in the Gulf region as the oil price recovered. Local currency investments (for example, in the stock and housing markets and the new, relatively liquid government securities, but also regular bank accounts) started offering better overall returns than FCDs.

• Banks’ incentives to mobilize FCDs were reduced. Schemes favourable to FCDs were terminated and various regulatory means, such as higher reserve requirements, were introduced to promote de-dollarization. Financial sector reforms also opened up alternative, lucrative local currency markets for banks. However, the SBP absorbed much of the exchange rate risk. Neither during the period of dollarization nor thereafter did banks take large foreign currency positions. Nor did corporates or households have significant liabilities denominated in dollars. Hence, de-dollarization was not complicated by a deterioration in bank soundness.

• There was little hysteresis in dollarization because dollar lending and ‘real’ dollarization of contracts and transactions in the domestic economy were not widespread. First, the FCDs did not give rise to a multiplicative creation of foreign currency money because foreign currency lending was not permitted. Second, the institutional arrangements for the dollarization of transactions and contracts were not already in place. De-dollarization did not involve the abandonment of a payment system based on the use of dollars or the wholesale rewriting of contracts denominated in dollars. Third, the disruption in the relationship between foreign currency depositors and banks and the subsequent decline in FCDs affected the SBP because of the forward cover scheme, but did not affect borrowers. Indeed, the direct effect on borrowers of the freeze on FCDs and the concurrent depreciation was slight; they endured higher interest rates and restrictions on trade financing, but they did not suffer a large capital loss.

• The foreign exchange crisis was caused by a political security crisis, rather than a purely economic shock or a failure of economic policy. The crisis was predominantly one of liquidity rather than solvency. Hence, confidence in the national currency and the national economy was not fundamentally undermined. The fact that banks, depositors and bearer bond holders suffered no large losses on a cash basis in the process helped underpin general confidence.17

• The achievement of macroeconomic stabilization was a precondition for de-dollarization. While there was feedback from de-dollarization and the concomitant shift in demand back to PR deposits to macroeconomic stabilization, causation ran primarily from the control of monetary expansion, fiscal consolidation, better debt management and structural reforms to lower demand for foreign currency assets. Furthermore, the improvement in the balance of payments (and the development of cheaper domestic sources of government financing) meant that the authorities had less temptation to reintroduce incentives to attract FCDs from Pakistanis abroad.

11.4 Argentina: crisis-induced de-dollarization

After a history of recurring financial crises throughout the 1970s and 1980s, Argentina went through its deepest crisis in early 2000 culminating in the abolishment of the decade-long currency board and debt default.18 The deposits of the banking sector were rescheduled and bank balance sheets were de-dollarized at asymmetric rates (Arg$l per dollar on the asset side and Arg$1.4 per dollar on the liability side). Calvo (2002) stresses the sudden reversal of capital flows to Latin America in the late 1990s and distinguishes the ability of different Latin American countries to cope with the reversal depending on the degree of openness of the country and the extent of liability dollarization. He argues that being a closed economy with a very high degree of liability dollarization, Argentina had to experience a very large change in the real exchange rate to eliminate the current account deficit.

Following implementation of the Convertibility Law and the restoration of macroeconomic stability in 1991, Argentina’s financial sector underwent a significant transformation. Banking system efficiency improved markedly through substantial consolidation, privatization and increased entry of foreign institutions. By the end of the decade, the banking sector was composed of half as many institutions as in 1995 and there was significant private sector ownership and a strong foreign presence. Financial deepening, measured by the ratio of M3 to GDP, increased from 5 per cent in 1990 to 30 per cent in 2000. Banking system assets almost doubled during the period, increasing from about 30 per cent to 57 per cent of GDP. The credit to the private sector increased 10 per cent per year peaking at 23 per cent of GDP in 1998. The financial system emerging from this structural transformation was relatively well capitalized.

However, the banking system had become heavily dollarized, making it very vulnerable to the elimination of the hard peg; more than 80 per cent of credits to both the private and public sector and close to 85 per cent of deposits were denominated in foreign currency (Figure 11.6). Although the banks did not have a currency mismatch, they were subject to credit risk as most lending in foreign exchange was to borrowers with local currency income (due to the government’s convertibility promise).

Figure 11.6Argentina: share of foreign currency loans and deposits, 1994-20041

(in per cent)

Source: BCRA.

1. 2005 data through June.

Under the Convertibility Law, the regulators did not impose prudential rules to differentiate between Argentinian peso vs. US dollar exposures, thus fostering the dollarization of the financial intermediation. As noted in Daseking et al. (2004) when doubts about the peg encourage holdings of dollar deposits, households and firms may still have an incentive to borrow in dollars, if they expect to be bailed out (or simply default) in the event of a large devaluation. Ex post, borrowers seem to have anticipated government intervention to limit their exposure in the event of a devaluation (as indeed happened as a result of the asymmetric pesoization).

The prolonged economic recession combined with slippages in fiscal policy and political instability led to capital flight. The authorities’ policy response (including a financial transactions tax and changes in the decade-old convertibility plan) exacerbated these concerns. Following the deposit runs, the authorities announced a series of restrictions on the cash withdrawals of deposits (the ‘corralito’) in December 2001 (Figure 11.7). The measures included (i) a Arg$250 per week limit on deposit withdrawals per bank account; (ii) limits on bank transactions to payment by cheques, credit and debit cards and interbank transfers; (iii) gradual dollarization of banks’ assets by only allowing the rollover of peso loans into dollar loans; and (iv) prohibition of transfers of funds abroad without prior Central Bank approval.

Figure 11.7Argentina: total banking system deposits, 2001-51

(public and private, local and foreign currency)

Source: BCRA.

1. Foreign currency deposits for 2002 converted at market exchange rates (Bloomberg).

Although the prudential regulations put in place during the 1990s should have been sufficient to protect banks from most risks, they were unable to counter the impact of the extraordinary events that took place after the collapse of the currency board arrangement in 2002. On 21 December 2001 the Argentinian government declared a bank holiday and restricted the withdrawal of deposits. It then forced an asymmetric conversion of deposits and loans at exchange rates that were lower than market rates. This resulted in large losses for depositors and large windfall gains for corporate and household borrowers. The asymmetric conversion of deposits and loans led to a hole in bank balance sheets which was then compensated by government bonds; however, the large gap between the market and book value of the compensation bonds (Bodens) implied a significant loss for the banks.

In the second half of 2002, deposit outflows reversed and the deposit base began to grow. This turnaround reflected the combined impact of exchange rate stability, high domestic interest rates and effective full protection of depositors in case of bank closures. In the face of this stability, the authorities began easing deposit restrictions and in December 2002, the authorities announced the lifting of all restrictions on the withdrawal of sight deposits.

By end-June 2005, bank deposits, in real terms, are 20 per cent higher than pre-crisis levels (Figure 11.8). However, the restructuring of deposits (for the third time in Argentinian history) has eroded credibility, and households and institutions have not been saving through the domestic banking system. The recovery in deposits is mainly due to the increase in public deposits and in sight deposits for transaction purposes. Private sector deposits are still 15 per cent below their December 2001 level; FCDs have been increasing slowly and account for 8.5 per cent of total deposits as of June 2005 (Figure 11.9).

Figure 11.8Argentina: total bank deposits

(in billions of constant 1999 Argentinian pesos)

Source: BCRA.

Figure 11.9Argentina: bank deposits, 1994-2005

(in billions of Argentinian pesos)

Source: BCRA.

Although credit has been growing briskly in 2004-5, credit to GDP remains low at about 10 per cent of GDP compared with 25 per cent before the crisis (Figure 11.10). In 2004, the total loans to the private sector increased by about 25 per cent, mainly in terms of short-term commercial (overdrafts and promissory notes) and consumer loans. Foreign currency loans are 9.5 per cent of total loans as of June 2005. According to regulations imposed after the crisis, banks can only lend in foreign currency to exporters or to borrowers that have an income denominated in foreign currency.

Figure 11.10Argentina: financial intermediation and financial deepening

(in per cent of GDP)

Source: BCRA.

The sustained decline in deposit and credit dollarization can be attributable to a number of factors:

  • By avoiding major price instability following the crisis, Argentina was able to restore the function of the peso as a means of payment and store of value. The country experienced a strong recovery following the disorderly exit of the convertibility regime at the end of 2001, and has concluded the largest sovereign debt restructuring in recent history.

  • The implementation of prudent macroeconomic policies has been the main factor behind the return of confidence. In addition, a positive international economic environment supported the recovery and public finances through high commodity prices. Domestic interest rates remained low and the peso nominal exchange rate remained broadly constant against the dollar through central bank interventions.

  • Through limits on the extension of dollar credits, the authorities have been trying to promote the peso as the currency for financial intermediation. Thus, banks have little incentive to mobilize dollar deposits as these can only be on-lent to exporters or borrowers with foreign currency income. On the other hand, there has been limited demand for loans from exporters as the sector is enjoying cash-rich positions due to high commodity prices.

Although the banking system has been gradually recovering due to the extension of regulatory forbearance, it mainly functions as a narrow banking system focused largely on payments and very short-term financial intermediation. As noted in de la Torre, Levy Yeyati and Schmukler (2003), after the collapse of the currency board and the forced conversion, Argentina, which previously had financial intermediation without a flexible currency, now has flexibility with very limited banking. In other words, currency mismatches have been replaced by maturity mismatches.

Finally, the manner in which Argentina exited convertibility had major implications on the enforcement of property rights and contracts.19 The credibility in financial institutions and the investment climate in general are unlikely to recover quickly due to frequent discretionary changes in contracts and regulatory frameworks.

11.5 Lessons from country experiences and policy implications

The experiences of these countries show that forced de-dollarization imposes a large and persistent cost on most sections of an economy. There is an immediate cost, as existing contracts are re-denominated and the availability of liquidity is disrupted. There is also an immediate but non-transparent redistribution of wealth. The longer-term costs may be even more substantial. Savers who wished to keep assets in dollars in the domestic financial system will now, if possible, switch to the nearest alternatives, namely assets abroad and consumption. Banks may prefer to concentrate on fee-based business rather than intermediation if they believe that the government will intervene ex post in their agreements with depositors or borrowers. Borrowers who could afford (at least in cash flow terms) financing denominated in foreign currency will have to scale back projects. The persistent consequences may include low intermediation, low savings and investment, capital flight and high-risk premia.

The governments that undertake forced de-dollarization may well be aware of these risks. They may also be aware that de-dollarization and similar actions (such as securitization of deposits) may have highly arbitrary redistribution effects. As far as the authors are aware, no government has chosen this path except in exceptional circumstances when the alternatives looked even worse.

The proximate cause of forced de-dollarization is usually a foreign exchange crisis; the country is running out of foreign exchange to meet its external obligations and perhaps its domestic dollarized obligations. The main underlying cause seen in the examples was an unsustainable fiscal situation, which generally led to the accumulation of a large stock of foreign currency-denominated government debt. The immediate trigger for the crisis may, however, be political (as in the case of Pakistan). Once fear of a crisis begins to intensify, there is likely to be a run on FCDs (and possibly local currency deposits also), which will drain off more foreign currency liquidity.

A common thread in the several country experiences is that deposit dollarization – and forced de-dollarization of deposits – is relatively easy to cope with if that is the only issue. The costs of de-dollarization are much higher and the chances of success are lower when credit (and other contracts) is dollarized. Four factors come into play:

• First, credit contracts are much more complex than deposit contracts. A loan agreement typically addresses numerous contingencies and redenominating a loan contract requires also complex renegotiating of pricing (the risk premium). Such renegotiation takes time, especially in the uncertain environment in which mandatory de-dollarization is undertaken. Until lenders and borrowers have renegotiated the outstanding stock of redenominated loans, they are unlikely to agree on new financing. Hence, enterprises will be starved of credit.

• Second, deposits and, in particular, savings deposits may be less important for ongoing economic activity than credit. If an enterprise cannot roll over a line of credit, it will be in default and may be liquidated. Hence, output and employment drop, with a multiplicative effect on demand. If a household that is a net saver temporarily loses access to its bank accounts, it can carry on much as before.20

• Third, the extension of foreign currency loans by domestic banks implies that banks create FCDs; a balance of payments inflow that provides banks with some foreign assets is likely to have a multiplicative effect on both FCDs and foreign currency lending. Hence, forced de-dollarization of both deposits and credit will disrupt both sides of the banks’ balance sheets. It could also change the monetary transmission process because the supply of and demand for loans could be affected by the breach and the change in banks’ portfolio.21 In contrast, without foreign currency credit, the stock of FCDs is closely linked to balance of payments developments, which limits the (ex post) magnitude of fluctuations. Furthermore, de-dollarization of FCDs alone should not have a first-round impact on the supply of credit and monetary transmission.

• Finally, those who have borrowed in foreign currency constitute a strong lobby group, whose vital interests may be imperiled in the turbulent circumstances surrounding forced de-dollarization. Their interests may be in conflict with those of depositors, a conflict that may be resolved by passing the burden onto some other sector. The Argentinian authorities, for example, responded to such pressures by choosing asymmetric pesoization of deposits and loans. Borrowers benefited (at least initially), but banks’ capital was largely eliminated, which in turn led to a sharp decline in extension of credit.

Since forced de-dollarization has not been undertaken except in very difficult circumstances, it is hard to separate its effects from those of other events. The outcome in some cases seems to have been relatively benign in that: (i) the economy ‘bounced back’ quickly and, in particular, the output loss was limited; and (ii) de-dollarization persisted beyond the emergency and indeed even when restrictions were lifted. The experience of Pakistan, it can be argued, meets both criteria. To date, the experience of Argentina meets (ii) but not (i); conclusions are less definite because less time has elapsed.

Even in the often extreme and perhaps chaotic circumstances in which the authorities choose de-dollarization ‘the hard way’, some policy principles can be followed to increase the chances of success and to minimize the negative side effects. The varied experiences described here, combined with the application of some general economic reasoning, suggest a number of such principles. Many of them would apply also to efforts to induce voluntary, slow de-dollarization, but are at least as important when conversion is compulsory. Moreover, several of the recommendations listed below apply in almost any scenario involving forced de-dollarization, while others may be suitable only in certain restrictive circumstances or if very carefully implemented.

Discourage foreign currency lending

As argued above, foreign currency lending, especially to non-export-oriented sectors and households, creates additional prudential and macroeconomic concerns. Hence, the arguments for discouraging foreign currency lending may be stronger than those for discouraging FCDs. Many countries have an outright ban on bank lending to residents, or at least for non-export related business. One could also imagine adapting other regulations, such as the setting of risk weights for capital requirements or limits on lending to households relative to household income, to discourage foreign currency lending beyond what would be called for based on purely microeconomic prudential considerations.

Remove regulatory incentives for dollarization

Dollarization has often been favoured by various regulatory provisions that disadvantage saving and financing in the local currency (as was most clearly the case in Pakistan). Removing these incentives should at least slow the pace of further dollarization or the recurrence of dollarization once de-dollarization has been achieved by other means. The main incentives affect savers and intermediaries (institutionalized incentives for foreign currency borrowing are less common):

Depositors. FCDs should not be treated more favourably than local currency deposits in such matters as the coverage of for deposit insurance and the prioritization of claims in the event of bank resolution; these factors may be most salient when de-dollarization is undertaken in the context of a banking crisis. Fiscal provisions also need to be designed to be at least neutral between currency denominations. For example, account needs to be taken of capital gains on FCDs in case of depreciation (following forced de-dollarization, savers may give weight to the possibility of an exchange rate crisis, and anticipate large capital gains on FCDs), and income tax provisions may need to allow for differences in nominal yields.22

Banks. Prudential and other regulations should not be distorted in ways that favour foreign currency intermediation, and indeed that allow for the complex of risks that they can engender. Thus, for example, the risk weight on bank assets, reserve requirements and liquidity requirements need to be chosen in this light.

Do not heavily penalize one sector

Policies need to achieve, and be seen to achieve, a reasonable degree of burden sharing. This is partly a matter of equity and welfare maximization; if there is diminishing marginal utility, the spreading of costs reduces the total welfare loss. More concretely, if one sector or section of society suffers disproportionately, it will have great difficulty restoring normal operations, or it will have especially little trust in economic institutions and the commitment of the authorities to follow policies consistently. Hence, the economy as a whole is likely to take longer to return towards equilibrium following forced de-dollarization. Furthermore, the most disadvantaged groups are likely to put up fierce political resistance, leading to a policy reversal.

Do not decapitalize the banks

Circumstances under which forced de-dollarization is imposed almost invariably involve a range of economic shocks, such as a balance of payments or exchange rate crisis or a fiscal crisis. These shocks will in themselves occasion a deterioration in the quality of banks’ assets and impose other costs on banks (for example, due to a shortage of liquidity or higher interest rates on their short-term liabilities while returns on longer-term assets are slow to adjust). A weakened banking system is likely to curtail lending and thus propagate and exacerbate the effects of the shocks. Policies that have the effect of reducing the capitalization of a large share of the banking sector will then make the situation worse. If banks have to conserve capital by concentrating on short-term and low-risk assets, the availability of working capital and investment financing will dry up and the corporate sector will suffer severely, and economic activity will contract. A vicious circle of lower output, non-performing loans and bank losses may result. In the end the government will almost certainly have to recapitalize the banks and then go through the complex process of reprivatizing them.

Do not forbid new FCDs

The episodes related here suggest that there is little danger in allowing the opening of new FCDs, free of restrictions, provided that positive incentives for dollarization have been removed. The experience of forced de-dollarization will reduce the supply of savings into FCDs, so a strong resurgence in dollarization is unlikely so long as macroeconomic stabilization is on track, local currency interest rates are allowed to adjust and regulations favouring FCDs are removed. FCDs in moderation can be helpful by providing banks with foreign currency liquidity, for example, for financing lending to exporters.

Do not use the counterpart to FCDs as usable reserves

Forced de-dollarization has often been the outcome of the authorities’ use of FCDs to finance capital flight or imports directly, or through the government’s fiscal operations. To avoid the recurrence of an emergency, the counterpart of (remaining or new) FCDs needs to be kept in relatively liquid form, such as in the foreign assets of commercial banks or the central bank.

Allow two-way exchange rate flexibility

Exchange rate flexibility may help discourage dollarization by exposing all market participants to the risk of valuation losses; there are no ‘one-way bets’. Furthermore, such flexibility is almost a precondition for the development of an efficient market for hedging instruments; if the exchange rate is tightly controlled or if it moves in only one direction, the differences in opinion and information among participants may be too small to drive active market trading. Prices and availability in an illiquid market are likely to be unfavourable, which feeds back into low participation.

11.6 Concluding remarks

Forced de-dollarization is highly disruptive. The mandatory re-denomination of obligations involves the breach of a multitude of contracts and the government revealing a willingness to overrule the property rights of much of the population. Dollarization is itself largely a manifestation of distrust of government policies, which forced de-dollarization confirms. Much time and many favourable circumstances will be needed to re-establish trust and, in particular, confidence that another round of de-dollarization or more extreme expropriation will not be forthcoming.

Given these risks, forced de-dollarization will be politically very difficult to undertake except in extreme circumstances. In principle, compulsory currency conversion could be a useful policy tool in ‘normal’ times to accelerate de-dollarization in the context of an overall strategy to revive demand for the local currency. In such circumstances, it may be easier to determine an appropriate conversion rate and less severe restrictions on withdrawals may be manageable. However, in a country where economic policies have induced significant dollarization, economic agents are unlikely to be quickly convinced that forced de-dollarization signals the start of a new, better policy regime and that the infringement of property rights will not be repeated. Hence, even in good circumstances, forced de-dollarization is likely to prompt some capital flight and disintermediation.

Forced de-dollarization is not a panacea: if faith in the national currency is not restored, re-dollarization in the form of FCDs or capital flight will be unavoidable and these costs will have been incurred without benefit. Other measures are needed to reduce the associated costs and improve the odds that the economy will recover rapidly. At a microeconomic, institutional level, preferable policies are those that minimize the breach of contract and ensure that such breaches that are imposed are transparent and equitable. At a macroeconomic level, policy rules and decision-making mechanisms need to be established that promote stabilization of domestic monetary conditions. Thus, whether a country attempts to de-dollarize ‘softly’ or is forced to de-dollarize ‘the hard way’, the underlying causes of dollarization need to be addressed if the effort is to be worthwhile.


    AlexanderW.T.Baliño and C.Enoch (1995) ‘The Adoption of Indirect Instruments of Monetary Policy’ IMF Occasional Paper No. 126 (Washington, DC: International Monetary Fund).

    BaliñoT.A.Bennett and E.Borensztein (1999) ‘Monetary Policy in Dollarized Economies’ IMF Occasional Paper No. 171 (Washington, DC: International Monetary Fund).

    Bonacorsi di PattiE. and D.Hardy (2005) ‘Financial Sector Liberalization Bank Privatization and Efficiency: Evidence from Pakistan’ Journal of Banking and FinanceVol. 29 pp. 2381-406.

    BufmanG. and L.Leiderman (1995) ‘Israel’s Stabilization: Some Important Policy Lessons’ in R.Dornbusch and S.Edwards (eds). Reform Recovery and Growth: Latin America and the Middle East (Chicago: University of Chicago Press for NBER).

    CalvoG. (2002) ‘On Dollarization’ The Economics of TransitionVol. 10No. 2 pp. 393-403.

    ChopraA. (1994) ‘Monetary Policy and Financial Sector Reform’ in L.EbrillA.ChopraC.ChristofidesP.MylonasI.Otker-Robe and G.Schwartz‘Poland: The Path to a Market Economy’ IMF Occasional Paper No. 113 (Washington, DC: International Monetary Fund).

    DasekingC.A.GhoshT.Lane and A.Thomas (2004) ‘Lessons from the Crisis in Argentina’ IMF Occasional Paper No. 336 (Washington, DC: International Monetary Fund).

    de la TorreA.E.Levy Yeyati and S.Schmukler (2003) ‘Living and Dying with Hard Pegs: The Rise and Fall of Argentina’s Currency Board’ EconomíaVol. 5No. 2 pp. 43-9.

    de NicolóG.P.Honohan and A.Ize (2005) ‘Dollarization of Bank Deposits: Causes and Consequences’ Journal of Banking and FinanceVol. 29No. 7 pp. 1697-727.

    de NicolóG.P.Honohan and A.Ize (2003) ‘Dollarization of Banking System: Good or Bad?’ World Bank Policy Research Paper No. 3116.

    GalindoA. and L.Leiderman (2003) ‘Living with Dollarization and the Route to Dedollarization’ Inter-american Development Bank Working Paper No. 526 (Washington DC).

    GuideA.-M.D.HoelscherA.IzeD.Marston and Nicoló (2004) ‘Financial Stability in Dollarized Economies’ IMF Occasional Paper 230 (Washington, DC: International Monetary Fund).

    HandyA. (1998) ‘Egypt Beyond Stabilization Toward a Dynamic Market Economy’ IMF Occasional Paper No. 163 (Washington, DC: International Monetary Fund).

    HavrylyshynO. and C.Beddies (2003) ‘Dollarization in the Former Soviet Union: from hysteria to hysteresis’ Comparative Economic StudiesVol. 45No. 3 (September).

    HerreraL. O. and R.Valde’s (2005) ‘Dedollarization indexation and nominalization: The Chilean Experience’ The Journal of Policy ReformVol. 8No. 4 pp. 281-312.

    HusainI. (2003) Economic Management in Pakistan 1999-2002 (Oxford: Oxford University Press).

    HyderZ. (2003) ‘Workers’ Remittances Resident FCAs and Kerb Premium: A Cointegration Analysis’ State Bank of Pakistan Working Paper No. 3.

    International Monetary Fund (2004a) ‘Monetary Policy Implementation at Different Stages of Market Development Country Cases and Appendices’ supplementary information prepared by staff of the Monetary and Financial Systems Department.

    International Monetary Fund (2004b) Pakistan: Financial System Stability Assessment including Reports on the Observance of Standards and Codes on the following topics: Monetary and Financial Policy Transparency Banking Supervision and Securities Regulation Country Report No. 04/215available at (July).

    International Monetary Fund (2004c) Pakistan: 2004 Article IV Consultation Ninth Review Under the Three-Year Arrangement Under the Poverty Reduction and Growth Facility and Request for Waiver of Performance Criteria – Staff Report; Staff Statement; Public Information Notice and Press Release on the Executive Board Discussion; and Statement by the Executive Director for Pakistan Country Report No. 04/411available at (December).

    IzeA. and A.Powell (2004) ‘Prudential Responses to De Facto Dollarization’ IMF Working Paper 04/66 (Washington, DC: International Monetary Fund). Revised version published in the Journal of Policy ReformVol. 8No. 4 (2005) pp. 241-62.

    IzeA. and E.Levy Yeyati (2003) ‘Financial Dollarization’ Journal of International Economics Vol. 59 (March) pp. 323-47.

    LorieH. and Z.Iqbal (2005) ‘Pakistan’s Macroeconomic Adjustment and Resumption of Growth 1999-2004’ IMF Working Paper 05/139 (Washington, DC: International Monetary Fund).

    MirakhorA. and I.Zaidi (2004) ‘Foreign Currency Deposits and International Liquidity Shortages in Pakistan’ IMF Working Paper 04/167 (Washington, DC: International Monetary Fund).

    ReinhartC.K.Rogoff and M.Savastano (2003) ‘Addicted to Dollars’ NBER Working Paper No. 10015 (Cambridge, MA: National Bureau of Economic Research).

    SavastanoM. (1992) ‘The Pattern of Currency Substitution in Latin America: An Overview’ Revista de Análisis EconómicoVol. 7No. 1 pp. 29-72.

    State Bank of PakistanAnnual Report various issues.

    State Bank of PakistanFinancial Sector Assessment various issues.

    UribeM. (1997) ‘Hysteresis in a Simple Model of Currency Substitution’ Journal of Monetary EconomicsVol. 40 (September) pp. 185-202.


The authors would like to thank Herman Kamil for his contributions and Nada Oulidi for able research assistance.

Attention is on partial dollarization, where the national currency still plays some role, rather than official, complete dollarization, which raises a somewhat different set of issues.

The term ‘dollarization’ is used as shorthand for the use of a foreign currency. In many European countries, the euro plays the role of the favoured foreign currency.

De Nicoló, Honohan and Ize (2003), for example, test a model relating demand for a portfolio of domestic and foreign currency deposits to risk of inflation and real depreciation. Based on a sample of 75 countries, they find that these risks are positively and significantly correlated with observed values of deposit dollarization. Thus, dollarization appears to be at least in part the rational response of agents to a lack of monetary policy credibility.

An example of a possible hysteresis effect might be found in the payment system: suppose that, before dollarization, the system could handle payments in national currency only. If software is installed to handle foreign currency transactions between residents, transacting in foreign currency becomes permanently much easier.

See Savastano (1992) for a detailed analysis.

Exchange and capital controls, however, were removed later in August 1990.

In Bolivia, within three years of the decision in 1985 to re-legalize dollar accounts, 70 per cent of all savings deposited in commercial banks were denominated in dollars.

Using cross-country data, Ize and Levy Yeyati (2003) find that dollarization of deposits is significantly lower in countries where indexation is prevalent. See also Galindo and Leiderman (2003).

During this period Pakistan maintained a complex system of restrictions on the making of current and capital account payments.

Dollarization of transactions and lending was never widespread in Pakistan.

At their peak in the mid-1990s foreign currency bearer certificates constituted about 8 per cent of domestic government debt.

The stock of FCDs was $4.7 billion at end-1996, $6.3 billion at end-1997 and peaked at $7.2 billion in April 1998. They had fallen to $6 billion by June 1998.

The authorities also eventually renegotiated foreign debt.

The same conditions applied to foreign currency bearer bonds.

Initially the ‘Special US Dollar Bonds’ had a maturity of five years and a yield of LIBOR+2. Later the maturity was reduced to three years and the yield raised to LIBOR+4.

Holders of FCDs suffered a valuation loss because conversion was at the old, less depreciated exchange rate. However, because inflation was moderate, in real terms the loss of principal was small.

This is not to say that depositors will not protest vigorously and even violently to protect their interests.

Under dollarization, a bank that receives an FCD needs to find a matching foreign currency asset. Once assets and liabilities are de-dollarized, any extra deposit can be placed in a wider selection of assets.

Suppose that FCDs yield 4 per cent annually and local currency deposits yield 16 per cent. A tax rate of 25 per cent will reduce the spread, which determines the optimal portfolio allocation, from 12 percentage points to 9.

    Other Resources Citing This Publication