1 Should Argentina Dollarize or Float? The Pros and Cons of Alternative Exchange Rate Regimes and their Implications for Domestic and Foreign Debt Restructuring/Reduction by Nouriel Roubini Stern School of Business, New York University First Draft: December 2, 2001 2 The opinions presented in this paper are solely those of the author; the usual disclaimer applies. As Argentina is in the middle of a major financial turmoil and the currency board regime is on the verge of collapse, the country is considering whether it should formally dollarize, i.e. give up the peso altogether and formally adopt the dollar as its currency for all contracts and transactions. In this paper I compare and discuss the foreign exchange currency regime options available to Argentina (section 1) and then discuss the formal criteria to assess a country’s readiness for dollarization (section 2). Alternative exchange rate regimes have different implications for the amount of domestic and foreign debt restructuring and/or debt reduction that the country will have to undertake in the near future to restore medium term debt sustainability. Thus, while discussing these alternative currency regimes, I will also consider the implications of these alternatives for the amount of debt reduction that will be required to restore the medium term sustainability of the debt of the country (section 1). The main conclusions of the paper are as follows: 1. The currency board regime and “convertibility” are effectively dead as severe capital controls and restrictions on bank deposits (a partial freeze) have been imposed. The country will have to move soon to a new foreign currency regime. 2. The only feasible currency regimes are either a move to a float (flexible exchange rate) or dollarization. 3. Dollarization could occur at the current parity or after a final devaluation. 4. In the short run the balance sheet effects of a depreciation/devaluation under a move to a float or a “dollarization cum devaluation” would be significant. Thus, a greater debt writedown of government and private sector foreign currency liabilities would become necessary if the change in currency regime is associated with a nominal and real depreciation. Such debt writedown will be particularly messy and complex to achieve for the private sector agents’ dollar liabilities with the risk of severe real costs deriving from disorderly bankruptcies. Thus, some creative ex-ante solutions to reduce such real costs are discussed and explored in the paper. 5. On the other hand, a real depreciation is necessary to restore competitiveness as the overall evidence suggests that currency is overvalued and the fundamental real exchange rate is significantly depreciated relative to its current value. Thus, dollarizing at the current parity would be undesirable. 6. Balance sheet effects deriving from the real depreciation necessary to change relative prices (real depreciation) would occur, in amounts similar to the regimes of float or dollarization after final devaluation, even if Argentina dollarizes without first devaluing; such effects would occur via the price deflation necessary to change relative prices. Thus, dollarizing without devaluing will not prevent such balance sheet effects from occurring and causing financial distress for agents whose relative terms of trade have worsened for good. Thus, the argument that dollarizing at the current parity avoids balance sheet effects is flawed. 7. Extensive capital and exchange controls and a freeze of bank liabilities will be necessary for quite a protracted period of time regardless of which new currency regime is chosen. The loss of credibility in the financial system and the severe real and financial distress of a wide range of financial and non-financial institutions will 3 require extensive controls to prevent a disorderly run on assets. But the restructuring of various claims will be messy and prolonged in any case and scenario. 8. The country does not satisfy most of the criteria for optimal long run dollarization. Thus, from a long run perspective, dollarization risks to have more costs than benefits even if, in the short run, dollarization may appear to be less risky than an outright move to a float. As dollarization at the current parity does not prevent the balance effects from occurring over time and it does not even solve the short run competitiveness problem while it has also undesirable long run consequences, it is the least desirable option. 9. The risks of a float are a possible return to high inflation and disorderly balance sheet effects if the nominal and real exchange rates overshoot. But dollarization would not be an optimal long-run strategy, even with a final devaluation before dollarization. Thus, to float is the better strategy that the country should choose. While monetary policy would be constrained in a float regime, in this regime the ability to change the real exchange rate via a nominal depreciation would not be undermined even with widespread liability dollarization. At the same time, the risks of a flexible exchange rate regime could be minimized and monetary stability and low inflation ensured via inflation targeting and the choice of a credible, independent and conservative central banker. 10. Regardless of the new currency regime and the amount of debt restructuring/reduction, radical economic reforms will have to be undertaken to ensure long run fiscal discipline, openness to trade and structural changes in the functioning of the state and greater structural flexibility of the economy. Section 1. Should Argentina dollarize or move to a float? Pros and cons of alternative exchange rate regimes in Argentina and implications for debt restructuring/reduction of these currency regime options Exchange rate regime options available to Argentina today As the currency board regime with convertibility is effectively imploding given the imposition of capital/exchange controls and the partial freeze on bank deposits on December 1, 2001, the exchange rate regime of Argentina will have to change soon. The main options regarding the new foreign exchange regime available to Argentina are as follows: 1. Move to a floating exchange rate regime while maintaining full capital mobility. A likely sub-option in this case is to move to a float but impose temporary capital/exchange controls to avoid a free fall and overshooting of the currency under flexible rates. 2. Dollarize at the current exchange rate of one peso per dollar 3. Dollarize after having devalued the currency to regain competitiveness. Note that both options 2 and 3 may be accompanied by capital and exchange controls and partial bank deposit freeze, such as those introduced on December 1, 2001, to prevent a bank run from leading to a collapse of the peg and the implosion of the banking system. 4 4. 5. Adopt a dollar-peso standard at the current exchange rate parity and create a secondary non-convertible currency unit (the Lecops or Patacones) that will provide seignorage but whose value/exchange rate will depreciate over time. Maintain the current currency board but try to avoid the collapse of this regime through the type of capital controls and the bank freeze introduced on December 1, 2001. Among these options, it is clear that the last one, maintaining the currency board with convertibility, is the least feasible as the currency board regime is now effectively dead with the run on the currency and the bank in the last weeks of November 2001. The only way the authorities could try to temporarily salvage the currency board’s peg parity would be to maintain and extend the draconian capital and exchange controls and bank deposit freeze imposed on December 1. But this solution means the effective death of the most crucial components of the currency board system (the “convertibility” of peso into dollars without any capital account restrictions). So, as of December 1, 2001, convertibility is defunct and so is, for all practical purposed, the currency board regime. While the controls may give some short term breathing time in terms of allowing a smoother move to a different exchange rate regime, a different regime will emerge quite soon. One is therefore effectively left with the options of floating or dollarizing; the other option, the one of the creation of a secondary local currency – the Lecops/Patacones option – is also highly unlikely to be feasible for reasons to be discussed above. Therefore, float and dollarization as the only realistic and feasible options. If dollarization is chosen, the issue will be whether to dollarize at the current parity of one peso per dollar or whether to dollarize after having devalued the currency one last time. The arguments for dollarizing only after devaluing are twofold: 1. if there is currently a competitiveness problem in Argentina, one final devaluation will be necessary to change the real exchange rate, regain competitiveness and help restoring economic growth; 2. devaluing before dollarizing will allow the country to save some forex reserves after having dollarized. Since the current stock of reserves barely covers the currency in circulation, all those reserves would be used in the exchange of reserves for US $ notes necessary to dollarize and none would be left for lender of last resort support of banks and for liquidity buffer needs for public debt management purposes. Devaluing before dollarizing would instead allow to maintain some liquid forex reserves to be used to reduce future liquidity/rollover risks of banks and the government. Balance sheet effects under a move to floating exchange rates as compared to devaluation followed by dollarization The main argument against dollarizing after devaluing (as opposed to dollarizing at the current parity) is that devaluation will cause severe balance sheet effects for the government, the banks, the non-financial firms and all other economic agents that have open foreign currency liability positions. Thus, compared to dollarization at the current parity, devaluation plus dollarization may cause a greater amount of bankruptcies among financial and non-financial firms (especially firms/households with dollar liabilities in the non traded sector) and will require a greater writedown of their foreign currency liabilities. The size of these balance sheet effects 5 and the ensuing financial distress will depend on the size of the real depreciation that a nominal depreciation/devaluation causes. Such balance sheet effect will also imply that, in the short run, the amount of haircut on sovereign foreign currency debt will be greater than those that would occur if the government debt is restructured but the country dollarizes at the current parity. While the two options of float or dollarization after devaluation will have different implications and consequences, the two alternatives are similar in one dimension: if - and this is a big if - the rate of real depreciation after a move to a float is equal to that that would be obtained with the devalue & dollarize option, the balance sheet effects of the two options would be identical. I.e. the amount of financial distress for banks, firms and government deriving from the real increase in the real value of the foreign debt after a devaluation will be similar under the two regimes. Indeed, one potential difference between the dollarizing at the current parity option and the two other options (float or devalue and dollarize) is that, in the short run, the additional balance sheet effects of a devaluation on all agents with net dollar liabilities, would be avoided if Argentina dollarizes at the current parity; but as we will see below, over the medium term, dollarizing at the current parity will not prevent such balance sheet effects from occurring as they depend on a necessary and unavoidable change in relative prices. Moving to a float or devaluing & dollarizing differ, however, in a number of ways: 1. The same amount of real depreciation in the two regimes may be consistent with very different amounts of nominal depreciation and inflation in the two alternatives. If a move to a float were to lead to a sharp fall of the currency and a nominal depreciation in significant excess of what is required to restore competitiveness (a risk that should not be underestimated), the pass-through of such a depreciation to domestic inflation may become very large. Under one extreme scenario, a complete loss of monetary credibility may lead to high inflation, or even hyperinflation, under a move to a float. 2. If under a float the currency initially falls more than the amount of devaluation in the devalue & dollarize option, the short run (and most likely long run) real depreciation of the exchange rate may be larger than under the alternative regime option. Thus, balance sheet effects, and the ensuing distress of firms, banks and government, would be much larger under a float in the short run, and in the medium long run too if the medium run depreciation is greater under a float than under the devalue & dollarize alternative. Thus, assuming that Argentina needs to devalue (either via a float or in the process of dollarizing) in order to regain competitiveness (rather than dollarizing at the current parity), the risks of a float relative to devalue & dollarize are a free fall of currency and high inflation and more severe balance sheet effects if such a free fall occurs. A comparison of the main pros and cons of alternative forex regime options More in general, the benefit of a float, compared to either one of the two dollarization options is the potential, in the medium-long run, of using the nominal exchange rate to absorb domestic and external shocks that require a real depreciation. The main risk of a float is that Argentina will go back to severe monetary instability and high inflation. Thus, if a float is chosen, the issue is whether Argentina could avoid both the short run overshooting of nominal exchange rates and a perverse sharp fall of its currency value that would spike inflation. I.e., the issue is what would be the nominal anchor for inflation expectations under a float? Inflation 6 targeting and the choice of a credible conservative central banker (as in Brazil under Fraga) is the option most likely to succeed if a return to high inflation has to be prevented. Given the historical poor policy credibility of monetary policy authorities in Argentina, other monetary regimes under a float (a Taylor rule, monetary aggregate targeting, et cetera) are less likely to be effective than a credible inflation-targeting regime to ensure persistent low inflation. If instead dollarization is chosen, the choice between dollarizing at the current parity and dollarizing at a devalued currency level depends on the two factors discussed above. If it is believed that Argentina has a competitiveness problem, dollarization after a final devaluation is better. Similarly, the latter option dominates also when one considers that it would allow the country to maintain a buffer of forex reserves after dollarization to be used for lender of last resort support of banks and avoid future rollover crises on the public debt. The main disadvantage of devaluing and dollarizing (as well as of a float) compared to dollarizing at current parity is that the short balance sheet effect on all dollar-liability agents will be larger and require significant further debt writedown of debts of firms and banks, in addition to those of the government. Thus, in comparing dollarization at the current parity with the two other alternatives, the issue becomes of how to deal with such short run balance sheet effects. Here are some issues to be considered. Debt restructuring/reduction implications of alternative exchange rate regimes First of all, it is now clear that, under any new exchange rate regime, a semi-coercive restructuring of public debt will occur with certainty; how coercive such restructuring will be is uncertain but the two main options are: 1. The one currently favored by the government, .e. maintain the face value of the claims and reduce the coupon/interest rate at below market rates; this option being similar to the debt restructuring in Pakistan and Ukraine. 2. a reduction in the face value of the claims (a formal haircut of principal) on top of a capping of the coupon rate as in the cases of Russia and Ecuador. While both options imply the recognition that the NPV of the original claims is reduced, i.e. there is effective debt reduction in both cases, the latter option would imply a greater NPV haircut for investors. Thus, the choice between the two options depends on an assessment of how “insolvent” Argentina is, i.e. how much debt relief is necessary to achieve medium term debt sustainability. Second, regardless of how large the haircut of the debt is, if a depreciation occurs (either under a float or a devalue & dollarize option) the initial balance sheet effects of such real depreciation would be larger for all agents with net dollar liabilities. Sovereign debt restructuring implications After a depreciation/devaluation, the government real debt would increase for two reasons: 1. The direct effect on the real value of the dollar debt deriving from the devaluation; 2. The increased fiscal cost of bailing banks and other institutions that may become bankrupt after a devaluation. Dealing with the greater distress of the government debt position deriving from the depreciation/devaluation would be, however painful, much more simple than dealing with the 7 distress of private sector agents (banks, firms, households). In fact, as the government is anyhow restructuring its debt regardless of the currency regime, any extra real debt burden costs deriving from a real depreciation could be in principle dealt with a greater haircut on the value of the government debt. Note that this “greater haircut” strategy would necessarily require a greater haircut of the debt held by foreigners (non–residents) to reduce the real burden of the debt for the government and the country. In fact, if devaluation leads to a greater direct debt burden for the government and greater indirect burden (via bankruptcies of banks and these institutions liabilities are guaranteed by the government), a capital levy on the domestic banks (via a greater haircut of domestically held debt) would only increase the implicit liabilities of the government (if depositors of banks are truly guaranteed and are not to take a real hit) and the need to raise further revenues to pay for these additional extra costs. In other terms, any capital levy that hits deposit-guaranteed domestic banks is just an inter-government accounting reshuffling of liabilities: an haircut on debt held by banks increase their hole on their asset side and thus increases the implicit liabilities of the government (who will need to recap these banks) with no net change in the debt burden for the government. Thus, to get a true haircut and reduction in the government and country’s debt burden, a greater hit will have to be taken by the non-resident holders of the country debt. Whether this can be legally and practically done is a complex issue that we will not address here. Of course, the debt burden for the government from financial distress of banks (under a haircut of domestic debt) could be reduced if domestic depositors take a hit as well, i.e. deposits are not, ex-post, fully guaranteed and part of the haircut of the assets of the bank (the restructuring of the government debt they hold) is taken by the holders of such bank liabilities rather than the government. The hit on domestic depositors (that is starting with the current freeze on deposits) is like a capital levy on domestic holders of assets. The capital levy on domestic agents (households, firms and banks) may then be reduced if foreign holders of domestic government and private debt take a bigger hit. Thus, the government could reduce its burden by increasing the burden on foreign holders of government explicit and implicit debt. The explicit government debt will be reduced via a greater haircut; the implicit debt (the one deriving from cross border interbank claims on the domestic banking system and other foreigners deposits in the banking system) could be reduced if such deposits are not guaranteed ex-ante and ex-post and take some of the hit once bank assets are reduced via a capital levy on their holdings of government debt. While all these options are messy and risk to further undermine the financial system and the private agents’ confidence in it, in practice the message is that, if the government and the country has, on net, to reduce its debt burden, any capital levy on debt held by domestic residents, will only imply a reshuffle of wealth and liability between government, banks, firms and households. A true debt reduction for the country means a greater levy on foreign (i.e. non residents) held debt. And since any currency regime option that requires a real devaluation of the currency will increase the direct and indirect debt burden of the government, for any degree of pre-devaluation insolvency of the government, a greater haircut of foreign held debt becomes necessary if a depreciation/devaluation takes place. 8 Private sector/agents debt restructuring/reduction implications (and the Hausmann proposal) Dealing with the financial distress and balance sheet effects of a devaluation on banks, private firms and households becomes much more messy and difficult than dealing with the extra distress of such a devaluation for the government. Potentially a large amount of private agents may become financially distress and bankrupt after a devaluation. Such distress will depend on two factors: the amount of net foreign currency liabilities of each agent; the size of the real depreciation of the currency for each specific agent/firm. While dealing with the extra distress for the government is “easy” in the sense that the liabilities of this agent have to be restructured anyhow, with or without a change in the currency regime, restructuring the liabilities of thousand of private agents becomes a prolonged and complex issue as the examples of Indonesia, Korea and other financial crises episodes show. One approach to deal with such balance sheet effects of a real devaluation and the ensuing financial distress of private agents is to eliminate it ex-ante via a systematic reduction of real dollar debts (as in the Hausmann proposal) as opposed to dealing with it ex-post via costly bankruptcy/restructuring procedures. The argument for an ex-ante solution is compelling: if a real devaluation will lead to the widespread bankruptcy of private agents (households with dollar mortgages, non-traded sector firms with dollar liabilities, banks with net foreign currency exposure and banks with no direct foreign currency exposure but that will be distressed anyhow once the dollar loans to distressed firms and households become non performing after a devaluation), then the real resource costs of restructuring, reduction of debts and liquidation of assets may be massive leading to a severe credit crunch and a sharper fall in output than socially necessary or desirable. Since these are real resource and output costs that can be and should be avoided, it is better to deal with them exante by reducing the real value of dollar debts than ex-post through messy, costly and protracted bankruptcy/restructuring procedures. One way to solve this balance sheet problem ex-ante is via the Hausmann proposal of turning all dollar debts into domestic currency debt and indexing the returns of these claims to the inflation rate. This way, the cost of any real depreciation, that would have increased the real burden for the holders of this dollar debt, is shifted instead to the creditors/holders of these claims. I.e., if a real depreciation would make such debtors insolvent and lead them to default on these debts anyhow, it may be less costly and more efficient to achieve such a real debt reduction via an ex-ante transformation of such foreign currency liabilities into peso, inflation-indexed liabilities. This way any real debt burden increase deriving from real, but not nominal, real depreciation is automatically shifted to the creditors rather than to the debtors. This Hausmann solution to the problem of the balance sheet effects of a devaluation has many practical shortcomings that have been already widely discussed (for example, many foreign investors are institutionally prohibited to hold foreign currency, i.e. non-dollar, assets) but it is conceptually a logical and sensible way to avoid the inefficient costs of bankruptcy and restructuring that would derive from depreciation-induced financial distress. Its main conceptual, as opposed to practical, shortcomings are different ones. First, this solution reduces the real burden of dollar liabilities across the board for all debtors regardless of their ability to pay it. For example, take two firms with the same amount of dollar debt, one being in the traded sector and the other being in the non-traded sector. After a real devaluation the latter is bankrupt while the former is not and would be able to continue servicing its debts. There is no reason to provide the same debt burden reduction to both firms; 9 the debt reduction should be agent/firm specific and tailored to the real distress that the real devaluation implies for the specific agent. And paradoxically, unnecessary debt reduction for firms that are able to service their foreign currency debt would possibly restrict their ability to borrow in the future and/or would increase the cost of future borrowing. Second, across the board debt relief gives “unfair” debt reduction to firms/agents that happen to have borrowed in foreign currency rather than in local currency. Take two agents/firms with same borrowing patterns, one that borrowed at lower dollar interest rates in foreign currency and the other who borrowed at the higher local currency interest rates in local currency. When all dollar liabilities are turned into real peso liabilities, the first agent/firm benefits twice: it was servicing its debt at lower nominal and real interest rates to begin with having borrowed in dollars and now it obtains another debt reduction via the transformation of such debt into real peso debt. Instead, the latter firm that paid higher interest rates in pesos to begin with does not get any debt relief. Thus, the across the board transformation of dollar debt into real peso debt has severe distributional effects. Such redistribution of wealth is not just from domestic agents to foreign holders of these claims but also redistribution from local agents who had dollar liabilities to those who were holding dollar claims against such agents. While across-the-board ex-ante solutions to the balance sheet effects of real depreciations are, as discussed above, “unfair” in a number of dimensions, the alternative process, i.e. a firm/agent-specific restructuring requires much more costly out-of-court or in-court bankruptcy/restructuring procedures that are likely to be long, messy and highly costly and eventually of little benefit to debtors and creditors. Ideas for partially selective as opposed to across the board private debt restructurings Then, one sensible compromise would be to make the debt relief more individual specific but with some across-the-board features. For example, since all mortgages are in dollar in Argentina and, as in the case of Mexico in 1995, millions of households holding such mortgages would stop paying them given their inability to service a larger real debt after a severe depreciation, one could give across the board relief to households with mortgages (maybe only excluding high income households). Then, the greater amount of distress for the banks and financial institutions who hold these dollar claims would be socialized, as in the case of Mexico, via a government recapitalization of these banks. At the end, the higher costs of bailing out or recapping the banks would be still borne by the average household (as taxes will have to be raised over time to service these greater government liabilities) unless the government imposes a greater capital levy (haircut) on the foreign holders of its debt. Similarly, the relief for non-financial firms could be tailored to their financial distress after a real depreciation. For example, firms in the traded sector would not need debt relief as their dollar liabilities are automatically insured via the higher price of their goods (export prices or local currency prices of import competing goods) and greater amount of competitiveness after a real depreciation. Firms in the non-traded sector instead would experience much more financial distress if they have large amounts of dollar liabilities. Thus, a stronger case can be made for giving across the board, ex-ante debt relief for these firms’ dollar debts. Finally, there is the issue of how to deal with the financial distress of the banking system after a real devaluation. Here, the first point to notice is that while Argentine banks are formally foreign currency hedged, as the amount of their foreign currency liabilities is on aggregate, equal to the amount of their foreign currency assets (as they borrowed in dollars and lent to domestic 10 agents in dollars in equal proportions), their effective foreign currency exposure is massive and their apparent currency hedging faulty for two reasons: 1. By borrowing in dollars and lending in dollars, they just transferred the currency risk from themselves to the firms and households who borrowed from them. But once the devaluation occurs, many among such firms and households will go bankrupt and their loans will become non-performing; thus, the banks themselves would be in turn distressed. 2. The bank liabilities, 60% in dollar and 40% in pesos, could be instantaneously transformed into dollars by depositors, thus suddenly leading to a currency mismatch between assets and liabilities. Then, banks can only hedge this risk by selling assets and buying the foreign reserves of the central bank, thus undermining the currency board. Also, the forex reserves of the central bank are much less than the peso deposits that can be instantaneously transformed into dollar deposits. Thus, banks have no way to truly hedge such a sudden currency shift of their liabilities. The above arguments suggest that a real depreciation will severely affect the financial conditions of the banking system and could be a cause of severe financial distress for them. The distribution of such losses between the bank shareholders, the government, the banks depositors and the foreign holders of claims on such banks will be a most messy process. Rather than having widespread bank insolvencies and bankruptcies and/or a huge increase in government liabilities, it may be better to deal with this problem ex-ante. If firms in the non-traded sector and households holding mortgages will get debt relief (either ex-ante or ex-post via an initial suspension of their servicing of these debt) and if the bank holdings of government debt will take a hit (either via a loss in their market value or a formal haircut), bank liabilities will have to be reduced or otherwise the government will have to pick up the bill. This means a combination of levies on shareholders of the banks, on foreign holders of interbank claims and on depositors (as in the case of a freeze that leads to some real haircut of claims; note that, so far, the December 1 freeze does not have yet such features). Since the objective should be to maintain the viability of a sound private banking system while minimizing the implicit liabilities of the government, an equitable distribution of costs would imply that all claimants will take a hit: shareholders, depositors and foreign creditors on the liability side of the banks’ balance sheet. At the same time, the amount of debt relief to be provided to the banks borrowers in distress (non-traded sector firms and mortgage burdened households) will have to be restricted (but not so much as to cause widespread private sector bankruptcies) if severe distress of such bank claims and bank insolvencies have to be avoided. Current need for a bank deposit freeze and extensive capital and exchange controls under any alternative scenario for Argentina. Note also that the second problem discussed above (currency mismatch of the banks) and the risk of an implicit or explicit capital levy on deposits is also behind the bank run that has been observed in Argentina in the last weeks. The December 1, 2001 decision to partially freeze deposits is only a temporary solution to this severe problem. If the freeze would have included a prohibition to switch peso deposits into dollars, such capital controls would imply that once the depreciation occurs, the capital levy hits the peso depositors rather than the government and/or the owners of the bank. Thus, this channel of financial distress could be plugged even the effect 11 of a real depreciation on non-performing loans of distressed non-traded sector firms and mortgage burdened households would not be plugged. But since the measures announced on December 1 imply that effectively all deposits are turned into dollar deposits (the decree says that they can be but the capping of deposit rates implies that all depositors will switch their deposits to dollar deposits, an effective dollarization of all bank liabilities), the currency mismatch of banks is now severely worsened unless the country formally dollarizes. In fact, now banks effectively have all their liabilities in dollars while 40% of their assets are still in local currency, these mostly being their holdings of peso government bonds and some local currency loans. Formal dollarization would turn all banks assets and liabilities into dollar ones but if a dollarization is preceded by a one time depreciation, the balance effects of this depreciation for firms and households would lead to a spike in non-performing loans and financial distress for banks. Note also that a broad bank freeze, one even more extensive than the one decided on December 1, will have to be maintained for a while even if the country dollarizes at either the current parity or a lower parity. In the latter case, the freeze is necessary to deal with the balance sheet effect mess of a devaluation in a situation where a bank panic and run is already under way. Even if the country were to dollarize at the current parity, an extensive freeze as well as capital controls to avoid a run on the foreign reserves of the central bank by panicky banks, firms and depositors worried about default risk /capital levy, as opposed to currency risk, will be necessary for a while, as in the case of Ecuador. This means that, for the time being, convertibility and capital mobility are dead, and neither banks, households nor firms will be allowed to access the foreign reserves of the central banks or freely move assets abroad (strict capital and exchange controls). Given the potential for disorderly runs, capital flight and roll-off runs on public debt while the messy transition to a different currency regime and a restructuring of government and private agents liabilities occurs, extensive capital/exchange controls and an extensive freeze of bank deposits cannot be avoided at this point. This also means that the discussion on any alternative exchange regime, maintaining the current peg, moving to a float or dollarizing with or without previous devaluation will have to be made while strict capital and exchange controls, as well as extensive bank freezes, will be in place. A formal run on reserves, a collapse of the peg and move to a float triggered by a speculative attack may be temporarily avoided via extensive capital controls on all private resident and non-resident economic agents. These controls will have to be extensive as in the current conditions, the incentive to run, hedge and cover currency risk are massive. But a move to a float may not be avoidable if there are leakages in the capital control regime that allow residents and non residents to access the central bank reserves at the current parity. While the costs of shorting the peso are now very high, given the extremely high, short term interest rates, this may not prevent a run on the reserves (if capital controls are not imposed) as the expected benefits of covering currency positions are higher than any sustainable short-term interest rate. Note that even in Turkey, overnight interest rates of 5000% on February 21, 2001 did not prevent a further run on reserves and the move to a float that day as such rates implied only a 2% expected daily return on local currency assets. Under those conditions, more than a 10% probability of a 20% overnight devaluation is enough to make profitable shorting the currency and buying reserves even with such extreme overnight rates. Thus, Argentina can in the very short run avoid the fate of Turkey, a sudden market triggered collapse of the peg, only via extensive and persistent capital controls. 12 Arguments against dollarization at the current parity The discussion above makes evident the potentially disorderly effects on balance sheets of a depreciation under either one of the depreciation scenarios (float or devalue & dollarize). Everything else equal, this would suggests that the alternative of dollarizing at the current parity may make sense as such balance sheet effects may be avoided. But the dollarization at the current parity may not make sense for many important reasons: 1. Balance sheet effects deriving from external shocks requiring a change in relative prices (real exchange rate) cannot be avoided, in the medium run, even in a dollarization without devaluation regime. 2. If, as likely, the country has a competitiveness problem, a real depreciation is necessary to restore growth; otherwise, growth will stagnate, default risk will increase and a more severe debt crisis may recur down the line. Also, if the country will face in the future real shocks that require a real depreciation, dollarization will not be optimal and nominal exchange rate flexibility (i.e. a floating exchange rate regime and monetary policy independence via the existence of a local currency) may be required to absorb such shocks. 3. Severe capital and exchange controls, as well as a bank freeze, will have to be maintained to have an orderly restructuring of domestic and foreign claims of government, firms, banks and households. I.e. the transition to a dollarized system (even without a final devaluation) will be messy. 4. The country does not satisfy the traditional criteria for readiness for, and optimality of, dollarization. Balance sheet effects will occur even with dollarization at the current parity The first point is the most important. We have so far conducted the discussion under the maintained assumption that there would be severe balance sheet effects under the two alternatives where the currency is depreciated/devalued but no balance sheet effects if the country dollarizes at the current parity. But, over time, the balance sheet effects are the same, regardless of which currency regime is chosen and cannot be avoided even in the dollarization without devaluation regime. The reasons why dollarizing at the current parity does not prevent balance sheet effect are already well known in the theoretical literature on the subject (see Calvo (2001), Cespedes, Chang and Velasco (2001)) but are worth reminding as there is a widespread, and wrong perception, that they can be avoided if Argentina dollarizes at the current parity. The argument is as follows: suppose that a country experiences a shock that requires a real depreciation such as a negative term of trade shock or a sharp fall in the demand for its exports or any other real demand or supply shock that requires a real depreciation. Under flexible exchange rates, such depreciation will occur via a nominal depreciation. Under a fixed exchange rate or in a dollarized regime, such a real depreciation will occur through a fall in the relative price of domestic goods, for example a fall in the relative price of non-traded goods. Thus, regardless of the currency regime, the balance sheet effect deriving from the increase in the real value of dollar debt (once the relative price change occurs) cannot be avoided. In a dollarized regime, the fall in the relative price of non-traded goods will imply that the real value of dollar debts of non-traded firms whose output price is falling will be as large as 13 the one that would have occurred under flexible exchange rates with a real depreciation driven by the exchange rate adjustment rather than the good price adjustment. And, indeed, in Argentina, these balance sheet effect under fixed rates, are already underway for a couple of years as sustained price deflation has been the channel through which the real depreciation is occurring and the way the price of non-tradable is reduced. Thus, since Argentina needs a change in relative prices to regain competitiveness, the balance sheet effects of this required change in relative prices are already at work, even while no nominal devaluation has occurred yet. In other terms, if one is concerned about balance sheet effects of a real depreciation, one cannot avoid them by dollarizing at the current parity; they will, and they are already, emerging regardless of. It is correct that, while balance sheet effects deriving from a change in the fundamental real exchange rate do not depend on the currency regime, their short-run severity may be reduced if the country does not devalue or dollarizes without devaluing first. It has been argued that, when a country moves to a float or depreciates and dollarizes, the balance sheet effects occur all at once and lead to instantaneous distress for banks and firms while if the currency peg is maintained or dollarization without devaluation occurs, the real devaluation occurs only slowly over time via price deflation so that firms and banks have time (the years that it takes for the real depreciation via deflation to occur) to adjust to the shock. But this time to adjust may be only of limited benefit as firms knowing that the real price of their good will fall over time will reduce investment and output in expectation of lower prices for their goods. Thus, slowing the adjustment of balance sheet and the required real resource allocation via a gradual emergence of the balance sheet effects may not be either feasible nor optimal; if the real price is expected to fall the real debt burden is expected to be higher regardless of whether the price adjustment has occurred instantaneously or not. Thus, delaying the adjustment may not solve the problem caused by the required change in relative prices. The only case that can be made for delaying the adjustment and smoothing it over time via deflation rather than nominal exchange rate adjustment is that, if a float occurs, there may be an unnecessary overshooting of the real exchange rate (relative to its new fundamental value) that is driven by the attempt of all agents to hedge their currency liability exposure; this overshooting may then cause in the short run much more massive and socially inefficient balance sheet effects of the devaluation that will in turn cause more negative effects on investment and output than those that would occur if the real depreciation was closer to what is required by the smaller change in the “fundamental” real exchange rate. In other terms, if the move to a float were to be disorderly and lead to an overshooting of the real exchange rate (as observed in all episodes of currency crises in the 1990s), avoiding such overshooting via a dollarization without initial devaluation may be preferable (indeed, Roubini, Perri, Schneider-Kisselev and Cavallo (2001) provide a theoretical model and empirical evidence of this overshooting and its real costs). But if this is true, there is another way to prevent an overshooting of the real exchange rate if the dollarization option is preferred: first to devalue by the amount necessary to change the real exchange rate to its new equilibrium/fundamental value and then dollarize. Thus, the above logic suggests that dollarization without a devaluation is dominated by devaluation & dollarization (or a float if the overshooting can be somehow avoided) as long as a change in relative prices is required. 14 Is the Argentina peso overvalued? Is a real depreciation necessary and feasible under flex rate in Argentina? This leads us to the second problem with the option of dollarizing at the current parity. If the real exchange rate is overvalued and a real depreciation is necessary, it makes sense to achieve this real depreciation right away since postponing the balance sheet effects via persistent deflation is as costly as dealing with them right away through a nominal and real depreciation. Then, the question becomes of whether the peso is overvalued today and whether a real depreciation is necessary to restore competitiveness and growth. Some argue that competitiveness is not a big issue in Argentina today. The arguments are a combination of the following points: there was not much overvaluation in the first place according to some measures; price deflation is already leading to real depreciation; wages are falling in nominal and real terms faster than can be seen in actual numbers; export were growing fast until recently and the trade balance is already improving; a small open economy with given terms of trade cannot affect them with a nominal depreciation; and the export to GDP ratio is small (about 10%) so that a real depreciation will not stimulate net exports or growth very much. I am not convinced by the above arguments and I believe that a significant real depreciation is necessary. First of all, while different measures of the real exchange rate show different degrees of overvaluation, several of them estimate it to be currently around 20%. Second, as the country has gone through three years of severe recession and low investment, the fundamental real exchange rate may be more depreciated today than a measure based on past historical average of the real exchange rate would suggest. Only a sharp depreciation will reduce the real price of physical assets and capital in Argentina to the point that it will be convenient for foreign investors to buy such assets and start investing into more productive capacity in the country. Also, only a large real depreciation, even beyond the long run equilibrium value, will stimulate competitiveness and growth. Argentina, as many other emerging market economies that experienced a crisis, may need an “undervalued” currency for a while to restore growth. Thus, a large real depreciation is necessary. Third, prices and wages are falling but they are falling a very slow rate; it would take a decade for a 20% overvaluation to be undone by deflation if prices and wages are falling at a 2% rate per year. Fourth, the trade balance and current account of Argentina are improving but only because the country is in a severe recession and thus imports and investment are sharply down. While both Turkey and Argentina were running a 4% of GDP current account deficit in 2001, Turkey was growing at 6% that year while Argentina’s GDP was falling. Thus, the fullemployment trade deficit and current account deficit of Argentina are much larger than their recession-depressed levels. Fifth, while a small open economy exporting raw materials cannot affect its terms of trade, it can affect the domestic relative price of traded versus non-traded goods, i.e. the real exchange rate, via a nominal depreciation; indeed commodity exporters such as Australia, New Zealand and Canada successfully adjusted to these terms of trade shocks in the last few years through a nominal – and real - depreciation of their currencies. Finally, while current trade shares are small, this does not imply that real depreciation would not help. For one thing it did in the case of Russia, a country with as small a trade share as Argentina. While exports may not grow very fast at first, they may over time as relative prices change in their favor. Also, import competing sectors of the economy that are now subject to 15 sharp import good competition will benefit from the real depreciation and will grow, in the same way in which unprofitable import competing firms in Russia became highly profitable after the ruble depreciation. And, as discussed above, a nominal depreciation will reduce the relative price of non-traded to traded goods and will incentivate over time the allocation of resources and investment in the traded sector. Also, if the country will face in the future real shocks that require a real depreciation, dollarization (under either one of its two variants) will not be optimal and nominal exchange rate flexibility may be necessary: Argentina will need monetary policy independence via the existence of a local currency and the flexibility of a floating exchange rate. This is the main argument for moving to a flexible exchange rate regime. In any case, the discussion above suggests that, even if the country were to decide to dollarize, the competitiveness argument would suggest that dollarization should occur after devaluation rather than at the current parity. Why the Lecops/Patacones solution will not work The need for a real depreciation in Argentina is also the reason why the solution currently favored by some in Argentina, the creation of a dual monetary regime with a Dollar-Peso fixed rate standard and a parallel currency (the Lecops/Patacones standard) - whose value would depreciate over time - would not work. The current creation of the Lecops to pay for central government and provincial government wages and pensions would work, as a disguised way to create seigniorage, if the country did not have a competitiveness problem. The issuance of the Lecops would imply that holders of this asset will see its value reduced over time (and indeed Lecops are already trading at a 90 cents on the dollar discount relative to their face value), an effective depreciation of this new currency whose creation restores real seigniorage revenues to the government. Then, in an ideal scenario, the private sector would remain on a dollar-peso standard at the current parity and there would be not balance sheet effects and no real wage changes. Only government workers and pensioners would take a hit (a reduction in real government spending necessary to restore the fiscal balance of the government) both through the cut in their nominal incomes (the 13% cut in wages and pensions) and the reduction in the real value of the Lecops they are paid with. Eventually, as the supply of Lecops increases, the right – currently in place - to pay taxes at the full face value of these assets would have to be phased out; otherwise as the nominal and real exchange rate of these assets falls, the government ability to extract seigniorage from these assets would be reduced. But if the country suffers of a competitiveness problem, this clever attempt to create seigniorage without a depreciation will fail. If a change in relative prices and real wages is necessary to restore growth, such real depreciation has to occur either through a nominal depreciation or through persistent and costly deflation. Since the latter, as argued above, is not desirable, Argentina cannot get itself out of its competitiveness, growth, debt and solvency problems just by issuing Lecops. This easy solution to Argentina’s problems is too clever to be right. Monetary policy independence in Argentina with a float A number of authors challenge this argument in favor of floating exchange rates by arguing that, in the case of Argentina, there is no independent monetary policy nor there can be any under floating exchange rates since the country is so heavily dollarized. The argument is that, while other emerging markets, such as Brazil, Chile, Mexico (as well as others in Asia and 16 in other regions) may have benefited from flexible exchange rates and are able to live reasonable well under them, Argentina could not benefit from flexible exchange rates and would have no monetary policy autonomy because of its extensive liability dollarization. For example, it has been argued Brazil may have financial indexation but liability dollarization there is limited (for example via legal restrictions to writing local financial contracts in foreign currency); thus, there is room for independent monetary policy in Brazil but not in Argentina. A somewhat different variant of the same argument is that Argentina, given its history of monetary mismanagement and high inflation, has never had an independent monetary policy and it will never will. These arguments appear to be flawed after some closer inspection. First of all, a nominal depreciation under flexible exchange rates will lead to a real depreciation and a change in the relative price of traded and non-traded goods regardless of the amount of liability dollarization; this point is clear from economic theory and practice. This real depreciation depends on the fact that, if nominal wages are downward inflexible but real wages are not downward inflexible, a nominal depreciation will reduce real wages and will increase real competitiveness regardless of liability dollarization. So unless, one wants to argue that real wages are downward inflexible in Argentina, there is room for a nominal depreciation to lead to a real depreciation. Indeed, as recent evidence show, both real and nominal wage are downward flexible in Argentina. So, this argument against flexible exchange rates is not valid. Second, one could argue that, given the history of high inflation in Argentina until 1990, any nominal depreciation will lead to higher inflation with no real depreciation, another way of saying that a nominal depreciation would not reduce real wages. But even this argument is flawed. In all episodes of a currency collapse in the 1990s (Mexico, Korea, Indonesia, Thailand, Russia, Brazil, Turkey) a nominal depreciation has been associated with a real depreciation. Actually, the surprise is why the pass-through to domestic inflation has been so small in all these episodes and why inflation rates have not surged (or they have surged for a year and then returned to single digits). Even in countries with a past history of high inflation and/or indexation (Mexico, Brazil, Russia) the pass-through has been small and inflation either did not surge or sharply fell to single digits after the initial surge. If countries such as Mexico, Brazil, Russia and others were able to design under float monetary regimes (such as inflation targeting) that provided an anchor for inflation expectations in spite of their history of high inflation, why shouldn’t Argentina be able to do so? After all, if the pass-through and the surge of inflation after a devaluation depend on a country history of inflation, Argentina has had a fixed rate and low inflation for much longer (over a decade now) than Mexico, Russia and Brazil did (about 5 years each on average). Thus, the view that Argentina would move to high inflation after a move to a float is not warranted and this argument is independent of whether the country’s financial assets are dollarized. The crucial factor is rather whether real wages are downward flexible or not. And all the evidence for Argentina is that they are downward flexible. In this regard, the fact that many private sector wages are today set in dollars is irrelevant; the issue is more whether, after a depreciation that requires a reduction in dollar wages, such real wages would fall. Since they are falling today both in nominal and real terms given the recession and the need to reduce unit labor costs, there is not reason why they would not fall in real dollar terms after a devaluation. Formal partial de-indexation of wages to the currency may occur over time and anyhow wages are already informally de-indexed. In conclusion, while the ability of Argentina to conduct an independent monetary policy under a float may be limited (but it is also limited in many other emerging market economies 17 with flex rates), it ability to engineer a change in relative prices and the real exchange rate under a float would not be limited even with liability dollarization. Domestic and external shocks that lead to a nominal depreciation in an emerging market with flexible exchange rates lead to a real depreciation even if monetary policy independence is limited (see Brazil in 2001 whose monetary policy has been straight-jacketed by the inflation target but whose currency value has fallen in nominal and real term over the year as driven by market forces). So, flex rates can provide shock absorption benefits to an emerging market economy regardless of whether its monetary policy flexibility is somewhat limited by the need to maintain policy credibility and avoid excessive inflationary fall of its currency. The relation between default and currency risk in Argentina A related issue in the debate between floating and dollarizing is whether the country default risk is high today because of the currency risk and would sharply fall once the country dollarizes. First of all, one thing is clear and self-evident. While dollarization eliminates the currency risk, it does not automatically reduce the country risk; the latter reflects mostly the ability (and willingness) to pay of the government and its private agents. Thus, to a first approximation, it is not affected by the currency risk. If Argentina is insolvent, it will remain insolvent with or without dollarization and will have to restructure/reduces its foreign debt regardless of its choice of the currency regime. And indeed the government is currently attempting this restructuring even while trying to maintain the current peg. Also, it is clear that, while dollarization by definition eliminates the currency risk and the risk of a currency crisis, it neither prevents and avoids debt crises deriving from liquidity runs on banks or rollover runs on the short term government debt. Dollarization also does not eliminate debt crises deriving from excessive debt (insolvency and semi-insolvency debt crises) as the experience of Panama with default and Brady restructurings suggests. So, dollarization does not eliminate liquidity/rollover risk nor insolvency risk. The more sophisticated argument in favor of dollarization suggests that, while country/default risk and rollover risk would not be eliminated with dollarization, it would be significantly reduced if the currency risk is eliminated. The argument goes as follows: the current default risk premium is high (a spread of over 30% on the dollar debt of Argentina in early December 2001) in part because of the currency risk. In fact, if a currency collapse leads to large balance sheet effects and banks/firm/government experience distress because of these balance sheet effects, the country risk will be higher because firms, banks and government with a higher burden of their dollar debts after devaluation are more likely to default. Thus, eliminating the currency risk would also reduce the country risk but reducing the risk of disruptive balance sheet effects. However, this argument is not convincing for a number of reasons: 1. Even if the default risk was partially related to the currency risk, the empirical issue is how much. One can argue that many agents in Argentina, including the government, are borderline insolvent regardless of the currency risk. So, the reduction of country risk is not going to be significant once dollarization occurs and currency risk is eliminated. After all, Ecuador’s country spread is still one of the highest among emerging markets (at 15% the third highest after Argentina and Nigeria) even after dollarization and any reduction in this spread relative to its level in the default period has more to do with the economic reforms and better macro policies recently implemented rather than dollarization per se. 18 2. If, as argued above, balance sheet effect will be large even if Argentina dollarizes at the current parity, the government’s real debt burden as well as the one of many other private agents will be large and increasing over time as deflation increased this debt burden and makes the balance sheet effects emerge. Thus, eliminating the currency risk would not eliminate the balance sheet effects deriving from the needed change in relative prices for some agents in the economy. Thus, elimination of the currency risk would not decrease such default risk premia. 3. If the country has structural rigidities, such as labor market rigidities that lead only to a slow reduction in nominal wages after real shocks that require them to fall in nominal and real terms, dollarizing as opposed to moving to a float, would exacerbate the effects of these rigidities and cause an increase in country risk. In other terms, if a real shock requires a fall in real wages and a real depreciation but these occur only slowly, via deflation in a dollarized economy, the negative effect of such rigidities on employment, output and growth may increase the country risk under dollarization. An economy with negative growth or low growth will experience a faster increase in its debt to GDP ratio and an increase in its default risk. So, being stuck in a regime that ensures low long run growth and inability to absorb shocks with a nominal depreciation may lead to a more unsustainable debt path and greater defaults over time. The above discussion suggests that the frequently arguments that country risk would fall after dollarization because of the elimination of the currency risk are not very convincing. Quite to the contrary, the country risk may actually increase after dollarization. Capital controls and banking freeze in the move to dollarization Let us consider next the third issue above, whether dollarization at the current parity can restore confidence and prevent the ongoing bank run and attack on the reserves of the central bank. I will argue that strict capital and exchange controls, as well as a broader bank freeze, will have to be maintained to have an orderly restructuring of domestic and foreign claims of the government, firms, banks and households even if the country decided to dollarize at the current parity. The reason is that, at this point, the transition to a dollarized system (even without a final devaluation) will be messy and disorderly as the confidence in the safety of the banking system, convertibility and the peg has been undermined and as economic agents realize that a costly restructuring of the liabilities of the government and of many private agents is unavoidable. First, note the run on the banks is not the result of irrational panic. The asset side of the balance sheet of the domestic banks has lost value in a number of ways while the real cost of bank liabilities has been going up because of: 1. The semi-coercive placing of government bonds on banks balance sheets whose mark-to-market value is falling with the rise of government bonds rates while the domestic debt restructuring imposes interest rate caps on these bank assets. 2. The increase in non-performing loans after three years of a severe recession. 3. The realization that many firms and households that hold dollar liabilities to banks will be severely distressed if a real depreciation occur and will also be 19 4. 5. distressed if the nominal depreciation will not occur (as the balance sheet effects of the real devaluation via deflation unfold). The mismatch between the currency denomination of assets (where a fraction is still in pesos) and that of liabilities in a situation where agents can (and will soon) instantaneously switch peso deposits into dollar deposits. The increase in the cost of banks liabilities, as overnight rates were surging and peso deposit rates were also surging made inevitable the need to cap the interest rate on peso deposit even if those policy actions undermined the confidence in the banking system. At the same time that banks’ solvency was being undermined, all agents in the non-traded sectors (households holding dollar liabilities such a mortgages and firms in the non traded sector) were also facing increased financial distress as the ongoing change in relative prices (price deflation) was increasing their effective real interest rates and the real burden of their dollar debts. Thus, the growing concern that widespread bankruptcies would become unavoidable. Finally, at the government level an already high debt burden and the failure to reduce fiscal deficit to levels compatible with a sustainable debt dynamics was leading to the need to restructure debt obligations on coercive terms. Also, the currency board was becoming increasingly unsustainable and not credible given all the external shocks hitting the economy and the policy mistakes that increasingly undermined its credibility. Thus, the increasing loss of forex reserves as domestic and foreign agents started to hedge to cover their currency exposure. At the end only drastic capital controls and a freeze on bank liabilities may prevent a financial meltdown and a twin currency and banking crisis. But, even if the current turmoil will lead the authorities to move to dollarization without a devaluation, the damage to the balance sheets of most economic agents is already severe and the confidence in the domestic financial system seriously undermined. Also, dollarizing without devaluing will not solve the competitiveness problem while not preventing the balance sheet effects of the ongoing real depreciation via price deflation from taking place. All these vulnerabilities that undermined investors confidence imply that extensive capital controls and restrictions on bank liabilities will be necessary even if the dollarization option will be chosen. Does Argentina satisfy the criteria for dollarization? No The final issues in discussing the arguments for dollarization is whether Argentina satisfies the traditional and non-traditional criteria to assess whether dollarization makes sense from a long run perspective (see the second part of this paper for a more detailed discussion of such criteria). These criteria are useful to assess whether dollarization would be optimal in the long run (compared to the alternative of a float) regardless of whether a final devaluation before dollarization occurs. In summary, it does not seem that Argentina makes an overall good case for dollarization from a long-run perspective. Consider the following shortcomings of dollarization. First, the Argentinean business cycle is not highly correlated with the U.S. one as the country is mostly a producer and exporter of commodities and raw materials. Indeed, the Fed sharp monetary tightening in the 1999-2000 period, at the time when the US was growing very fast while Argentina was in a recession, shows how inappropriate it may be for Argentina to adopt the monetary policies of the U.S.. 20 Second, Argentina is a closed economy that trades very little with the US. Its export to GDP ratios is barely above 10% and only about a quarter of its exports go to the U.S.. While trade liberalization, FTAA and/or bilateral trade expanding agreements with the U.S. may increase over time its trade integration with the U.S. this will occur only slowly over the long run. Third, dollarizing will imply that the exchange rate of the Argentina (its nominal and real exchange rate) will be tied to the U.S. dollar; if the dollar keeps on strengthening relative to the Euro and the Yen, further losses of competitiveness of Argentina relative to these trading areas will ensue. Also, if as likely, most Latin American economies maintain flexible exchange rates and their currencies depreciate over time relative to the US dollar, further losses of competitiveness of Argentina’s traded goods sectors compared to its regional partners will result. Already the sharp fall of the Brazilian peso this year, on top of the 1999 depreciation, and of other Latin American currencies has lead to further nominal and real appreciation of the Argentine peso relative to these currencies and to severe stress on the Mercosur free trade arrangements. Such divergence of relative prices may occur in the long run if these currencies remain on a float while Argentina dollarizes. Fourth, Argentina will keep on being buffeted over time by real external and domestic shocks that require a real depreciation; for example, terms of trade shocks. If such depreciation cannot be achieved via a nominal depreciation because of dollarization, it will have to be achieved via painful domestic price and wage deflation. Fifth, Argentina’s labor markets are rigid and nominal wages are downward inflexible in the short run. Thus, any real shock requiring a fall in real wages may cause instead unemployment and low growth if real wages cannot be reduced via a nominal depreciation. Hong Kong was able to absorb the 1998 overvaluation shock by reducing rapidly nominal wages in a very flexible labor market and thus restore growth by 1999. Argentina, instead, has been stuck in a recession for three years now. Sixth, while room for independent monetary policy is limited in emerging markets in general and heavily dollarized economies in particular, this does not mean that Argentina would have no flexibility to use monetary policy to partially absorb shocks that require a real depreciation. Using monetary policy to allow a nominal and real depreciation under a float may be necessary and useful to dampen the real output effects of such shocks, While liability dollarization further constraints the autonomy of monetary policy, as long as real wages are not downward inflexible, monetary policy can be used to allow a nominal and real depreciation and stimulate demand and output. At the same time, as in many other emerging market economies, inflation credibility can be maintained through alternative monetary regimes such as a formal inflation-targeting regime. Seventh, the fiscal position of the country is still severely imbalanced and the debt situation unsustainable. Thus, dollarization may not prevent further debt and rollover crises. Eight, the soundness of the banking system is by now undermined (see the discussion above) while, at the same time, dollarization will further limit the ability of the central bank to provide lender of last support to the banking system. While such support is limited even under flexible exchange rates in emerging markets as liquidity injections may lead to sharp capital flight and currency depreciation, the loss of all forex reserves in the passage to dollarization is a further loss of lender of last support resources that is specific to dollarized economies. How will lender of last resort (LOLR) support will be provided in a dollarized economy? This is not clear. 21 Panama prides itself for not having any LOLR facility but the country is an international financial center, most of its banks are foreign owned and the IMF has been effectively providing support to the country as Panama has been the most assiduous client of the IMF for the last 25 years (over 12 IMF programs). In Ecuador, that recently dollarized, most of the banking system is still insolvent and in the hands of the government while the recapitalization of the banks has not fully occurred yet. Also, the issue of how to pay for the implicit liabilities to the government from the losses incurred by the Ecuadorian banking system has not been solved yet. And there are no meaningful lender of last resort resource today. Finally, limiting future government liabilities via partial and market priced deposit insurance may not prevent future bank runs. Ninth, the amount of labor mobility between Argentina and the US is very limited as there is no free and unrestricted migration. And the amount of implicit fiscal federalism deriving from workers’ remittances (important for many central American economies) is absent in the case of Argentina. Finally, the loss of seignorage deriving from dollarization, while not being huge is not economically insignificant as it amounts to about 0.5% of GDP per year. On the positive side, Argentina satisfies some of the non-traditional criteria for optimal dollarization. First, the economy is already partially dollarized; so moving to full dollarization will be relatively easy. Second, Argentina has already a currency board with all the costs of the lack of monetary independence while paying the costs of the partial credibility of the currency board peg. Thus, giving up altogether monetary autonomy with dollarization will imply small marginal costs relative to the current regime of a currency board. Third, given its history of poor monetary credibility and high inflation as well as liability dollarization, Argentina has very limited monetary policy autonomy. So, at the margin the full loss of monetary autonomy would be relatively small while the benefits of low inflation would be permanently locked in. Fourth, dollarization may force the political system to undertake the macro and structural reforms required for dollarization to succeed. But the dollarization straitjacket (as Ulysses tied to the mast to resist the lure of the Sirens’ song) may or may not incentives the necessary economic reforms. For example, Panama has been fiscally irresponsible for decades, in spite of the dollarization straitjacket, eventually defaulted on its foreign debt and has permanently relied on the financial IMF life support. Fifth, Argentina does have enough gross forex reserves so far to dollarize if it wishes to do so. But given the amount of foreign liabilities owed to the IMF, net international reserves are lower and net reserves may not be sufficient to allow dollarization unless a final devaluation is engineered before the dollarization. Finally, large sections of Argentina’s population may support dollarization; thus, there is significant political support for dollarization. Considering both pros and cons, a careful assessment suggests that the cons dominate the pros and that Argentina does not make a good candidate for successful dollarization from a long run perspective. This suggests that Argentina may be better off by floating its currency; it would then have to avoid currency overshooting and ensure monetary stability and low inflation via inflation targeting and the choice of a credible, independent and conservative central banker. The risks of a float are high, especially in the short run; they are the possible return to high inflation and disorderly balance sheet effects if the nominal and real exchange rates overshoot. But 22 dollarization may be a more risky long-run strategy, even with a final devaluation before dollarization. Some may argue that even Ecuador did not make an obvious case for dollarization as its fiscal balances were in shambles, its debt unsustainable, it banks bankrupt and other dollarization criteria not satisfied. And some argue that, in spite of not satisfying these formal criteria for dollarization, dollarization has be so far a success in Ecuador. But the jury on whether dollarization will be successful in the long run in Ecuador is still out. Fiscal conditions have improved but not enough; positive terms of trade shocks, such as the increase in the price of oil until recently as well as new production and distribution of oil via a new pipeline have improved the country prospects; the country defaulted on its foreign debt and reduced the principal value of its foreign debt but, in spite of that, the foreign debt to GDP ratio is still close to 100%; financial conditions in the banking system have improved, the bank deposits freeze has been phased out but the financial system still remains very fragile, many banks are still under government control, the bank recapitalization still to be finished (and the resources to finance it not yet found) and the issue of how to provide lender of last support under dollarization is still unresolved; finally, significant IMF and other multilateral support to Ecuador has significantly helped to finance the transition to dollarization. Thus, the relative improvement in Ecuador’s conditions in the last year does not yet prove that this is a successful case of dollarization. And any inferences from Ecuador to Argentina would be a matter of speculative guesswork. Need for sound monetary and fiscal policies and structural reforms regardless of the currency regimes and the amount of debt restructuring Regardless of the new currency regime and the amount of debt restructuring/reduction, radical economic reforms will have to be undertaken to ensure long run fiscal discipline, openness to trade and structural changes that ensure changes in the functioning of the state and a greater structural flexibility of the economy. A better currency regime may help and debt restructuring may help to reduce the burden of foreign debt but, unless the political system is able to ensure sound money, sound fiscal balances and structural reforms that open the economy and provide the necessary efficiency and flexibility of the economy, any monetary regime will be bound to fail and recurrent financial and debt crises may occur. It is better to float in Argentina Subject to the above caveat that sound macro and structural policies are necessary in any regime, overall a move to a float makes more sense than dollarization for Argentina. The risk of overshooting and high inflation may be limited with an alternative monetary regime (inflation targeting) while the balance sheet effects of the real depreciation can be addressed through a larger writedown of sovereign debt and a selective writedown of some private debt. Some monetary autonomy, however restricted, will be maintained under a float and the nominal exchange rate adjustment will allow a change in real exchange rates when internal and external shocks require it. Part 2. Criteria to assess whether a country is ready for dollarization 23 Next, in this section we analyze and discuss the criteria to assess whether a country is ready for dollarization. The application of this analysis to Argentina has been already provided in the section above. Monetary, financial and fiscal factors Policy credibility Countries where policy makers have historically suffered from a lack of policy credibility, especially in the monetary area, may benefit from the discipline imposed by a rule-based monetary regime such as a currency board or dollarization. The inability of discretionary policy makers to credibly commit to monetary stability and discipline has been suggested as one of the strongest arguments in favor of the rigid monetary discipline imposed by dollarization. There are sometimes partial exceptions: monetary unions such as EMU resulted more from the desire to create stronger economic and political ties than just the need to provide monetary and inflation stability. But emerging market economies that desire to import the credibility of a low inflation anchor country may consider dollarization as a strong discipline device. The lack of policy credibility can be measured by looking at a number of variables such as: the country’s experience with inflation; a past history of exchange rate instability and crises; the existence of previous financial and banking crises; the degree of unofficial dollarization; the country’s inability to borrow long-term in domestic currency; the country’s fiscal record; the spread between local dollar-denominated and local currency denominated interest rates. Inflation experience The role of recent inflation history is ambiguous. On one side, countries that need the most rules (such a currency board or dollarization) are those that have had a history of monetary instability; and indeed, countries such as Argentina and others switched to a currency board as a way to break out of a cycle of high inflation. On the other side, a recent low inflation performance signals that the country is credibly able to commit to the monetary discipline required to pursue price stability. Most likely, dollarization is most appropriate for countries that have sinned in the past (have had high monetary instability) but now have competent and stable governments with deep popular support that are determined to commit to rigid monetary rules to maintain long-run policy stability. In this dimension Argentina appears to fare better than Brazil and other Central American countries. Current exchange rate regime While a country could dollarize starting from any exchange rate regime, a successful experience with currency boards or credibly fixed exchange rates signals that the country has already shown its commitment to a stable currency, has proven its willingness to pay any costs associated with fixed exchange rates and is thus unlikely to experience further large costs from giving up altogether a national currency. Also, the additional transitional cost of moving to dollarization from fixed rates or currency boards are lower than when starting from more flexible exchange rate regimes. A transitional currency board stage should not, however, be a necessary criterion for dollarization. As currency boards may imply some costs (the risk of eventual devaluation) without the full credibility benefits of dollarization, it may make sense for a country considering dollarization to avoid a transitional stage of currency board and move directly to the elimination of the national currency. Experiences such as EMU suggest that the path to monetary 24 union and dollarization does not necessarily go through a linear transition involving greater exchange fixity: EMU was implemented even as the 1992-93 ERM crisis led to a significant formal widening of the ERM exchange rate bands. Reserve coverage of monetary base A minimum criterion for dollarization is that the foreign exchange reserves of the dollarizing country should at least cover the monetary base (or the currency in circulation). However, some countries that may otherwise be good candidates for dollarization may not satisfy this requirement in which case they may consider borrowing the necessary reserves from official or private creditors. While this solution to the need to convert the money base into dollars is technically possible, the increased foreign currency liabilities of the central bank may undermine the credibility of the dollarization. In estimating whether foreign currency reserves are sufficient to cover base money, one should look at usable reserves. For example, Costa Rica includes among its forex reserves over $300m credits to Nicaragua and Honduras that have not been serviced in a long time and are unlikely to be serviced ever. In addition to monetary base, one may want to include other (domestic and foreign currency) liabilities of the central bank in the aggregate that needs to be covered by foreign reserves. By these criteria, a number of Central American countries do not appear to have sufficient forex reserves to be able to dollarize without further foreign currency borrowings. For example, the Dominican Republic does not even have enough foreign reserves to cover the currency outside the deposit money banks; Belize and Costa Rica have reserves below the monetary base; Guatemala has reserves below the sum of the monetary base and central bank bonds (issued for the purpose of open market operations). Such reserve coverage looks even worse if we include the foreign liabilities of the central bank. Conversely, Argentina, El Salvador, Honduras and Nicaragua have enough reserves to cover monetary base and the bonds of the central bank. Obviously, if the reserve coverage goes beyond the monetary base and the other liabilities of the central bank, the country will have resources to provide partial coverage of the liabilities of the banking system, i.e. to provide some lender of last resort services. This issue is discussed in more detail below. Soundness of the banking system The existence of a sound, competitive, well-supervised and well-regulated banking system is an important condition for a successful dollarization. Weak banking system may experience systemic crises that are fiscally costly and may, in the absence of a strong lender of last resort facility, lead to financial panic and serious economic distress. Such fiscal costs of systemic banking crises are larger the larger is the financial sector relative to the size of the economy. A large presence of foreign banks in a dollarizing economy may help as it will: (a) reduce the risk of banking crises; (b) provide implicit lender of last resort support through home country head offices. Some have argued that a weak banking system should not be an excuse for delaying dollarization as poor financial systems pose serious problems regardless of the exchange rate regime and banking fragility may be used as an excuse to delay reforms and monetary stability. However, the credibility of dollarization may be undermined if a systemically fragile banking system implies large fiscal costs of a bailout of the banking sector, especially given the more limited lender of last resort resources available in a dollarized economy. Extent of informal dollarization 25 The greater is the degree of current unofficial dollarization, the smaller the benefits of exchange rate devaluation and the greater are the potential benefits of formal dollarization. If the dollar is already used as a unit of account, means of payment and store of value, the costs of a transition to formal dollarization will be minimized. Moreover, in an economy where a large part of the liabilities of the financial, corporate and household sector are already in foreign currency, a currency devaluation will have contractionary effects and may lead to severe financial distress for foreign currency borrowers. Thus, liability dollarization increases the benefits of full dollarization. In particular, countries that are not able to borrow long-term in their own currency (“original sin” countries in the Hausmann terminology) may also gain from the financial deepening (such as development of capital market for long term finance) associated with formal dollarization. Such financial deepening has been argued to potentially be an important benefit of dollarization. Ability to provide lender of last resort functions after dollarization While a dollarized country is generally more restricted in its ability to provide lender of last resort services to its banking system, such a function can be performed even in a dollarized economy through a variety of channels. First, if the foreign reserves are in excess of what is required to cover the monetary base, such excess reserves can be used to cover part of wider monetary aggregates such as demand deposits and other longer term liquid liabilities of the banking system. Second, the dollarized country could build-up liquid reserves via borrowing from the private sector (private contingent credit lines) or international financial institutions. Third, changes in reserve requirement ratios may provide further liquidity to a banking system under pressure. Fourth, under a seigniorage revenue-sharing arrangement, the discounted value of the stream of future seigniorage payments could be used as a collateral for lines of credit with private and/or official creditors. Conceptually, other revenue streams or assets could also be used as a collateral for such credit lines even if one would need to consider the implications of such collateralization for the overall sustainability of the public debt of the country. Some have argued that the actual ability to provide lender of last resort functions under a flexible exchange rate regime may also be effectively limited by the availability of foreign exchange reserves when the banking system has large foreign currency liabilities. Attempts to systematically rescue a banking system under distress via massive liquidity injections (as opposed to limited liquidity support to selected banks under pressure) may cause inflation, currency devaluations and/or a run on the reserves of the country as banks face a significant rollover risk for their foreign liabilities. Thus some have argued that experiences such as those of Indonesia suggest that, under conditions of limited fiscal resources for a banking system bailout, lender of last resort support through excessive liquidity creation under a depreciating currency may actually be destabilizing, especially in an environment in which there is weak credibility of the monetary regime. While the argument on the limits of lending of last resort under flexible exchange rates may have some validity with countries with low reserves, high foreign liabilities of the banking system and a fiscal inability to address a systemic banking crisis, the argument has less validity for monetary stable economies with significant amounts of foreign reserves and relatively low foreign currency denominated debt. Moreover, in discussing lenders of last resort one should conceptually separate the provision of liquidity to specific banks when they are under distress and there is risk of contagion to sound banks from the fiscal problem for the country deriving from systemic distress of a large fraction of the banking system. The latter problem is essentially a fiscal one and a credible lender of last 26 resort means that the fiscal authority has enough resources to cover, over time, the fiscal cost of the bailout of the financial system. Provision of short-term liquidity may plug the panic deriving from systemic distress but cannot solve the fiscal problem of a large banking system bailout. On the other hand, when banking distress is not systemic but the problem of individual banks can lead to panic, bank runs and contagion to sound banks, provision of short term liquidity can avoid runs without necessarily implying large and long-term fiscal costs. In this latter case, it is of fundamental importance for a central bank to have enough resources to plug any liquidity shock, guarantee depositors that liquidity is available and prevent a destabilizing run. While such liquidity support is feasible when a country has its own currency, this stabilizing lending of last resort is limited by the available foreign exchange reserves in a dollarized economy. Thus, the ability to prevent self-fulfilling bank runs may be compromised in a dollarized economy if central bank foreign currency liquidity is not available in sufficient amounts. Revenue cost of seigniorage loss Dollarization that occurs without seigniorage-sharing with the United States would imply a revenue cost in the form of seigniorage revenue loss. For countries in which seigniorage accounts for a significant fraction of government revenues, such loss has serious fiscal consequences and needs to be compensated by an increase in non-seigniorage revenues. If seigniorage revenues are significant, this switch in sources of revenue may require tax reforms to reduce a structural reliance on seigniorage. In the absence of revenue-sharing, the seigniorage loss is partly reduced if the dollarized country imposes non-remunerated reserve requirements on its banking system (essentially another form of taxation of banks) and thus the central bank can earn the interest rate on the non-currency component of the monetary base. For countries such as Costa Rica, Dominican Republic, Guatemala, Nicaragua and Belize seigniorage revenues have averaged around 1% of GDP (and a much larger fraction of central government revenues) in the 1990s; for Argentina, the corresponding figure is smaller and close to 0.6% of GDP. Central bank solvency in the absence of seigniorage sharing Another aspect of the loss of seigniorage to be considered is how such a loss affects the solvency of the central bank of a dollarizing economy. The discounted value of future seigniorage is an asset for the central banks that does not appear in the current balance sheets of central banks. This implies that central banks officially have often negative net worth; this may not be a problem when a country has its own currency since the discounted value of the stream of seigniorage revenues is a substantial asset that is not shown in such balance sheets. For example, the balance sheet of the central banks of Costa Rica and the Dominican Republic shows negative net worth while it is barely positive in Guatemala and Honduras. However, all these balance sheets would show solvency if the asset value of future seigniorage revenues was properly accounted for. This apparent insolvency of central banks becomes a more serious issue when a country dollarizes and seigniorage is not shared with the anchor country: in that case a negative net worth of a central bank may be a real form of insolvency. The ability of a central bank to provide credible lender of last resort services may thus be further undermined. Strategies to make a central bank solvent under dollarization could include: implicit seigniorage from nonremunerated reserve requirements of the banking system; explicit seigniorage revenue sharing with the anchor country used to recapitalize the central bank; or other forms of recapitalization of the central bank. 27 One radical solution to this problem would be not to have a central bank as a dollarized country may not need one; the experience of Panama is consistent with this view. The Panamanian approach is, however, very radical as the country does not have reserve requirements, lender of last resort, discount window liquidity or deposit insurance. While Panamanian authorities praise the totally hands-off market discipline approach to banking regulation and supervision, it is not obvious that this approach may work for other countries considering dollarization. In fact, Panama is a major financial center with a significant foreign bank presence; thus, Panama is effectively free riding on the lender of last resort, liquidity access, supervision and regulation that its foreign banks receive from home bank authorities and head offices. It is thus highly likely that central banking functions (such as supervision/regulation, deposit insurance, lender of last resort and required reserve setting) will need to be provided even in a dollarized economy, especially if such functions are not provided by the central bank of the anchor country. Thus, if a central bank is to operate even after dollarization and provide the above services, the solvency of such an institution in a dollarized economy is an issue that needs to be addressed. State of public finances The smaller is the budget deficit and the stock of public debt, the smaller is the risk that dollarization might fail. In fact, unsustainable fiscal conditions may eventually tempt policy makers to reverse dollarization, return to a domestic currency so as to be able to run again the printing presses (regain access to the inflation tax). Severe fiscal problems may also undermine the confidence of the public in the fiscal authorities and lead to a foreign debt-related financial crisis (a point elaborated in more detail in the next section). Fiscal soundness may also limit the risk that the United States may be seen implicitly or explicitly as prepared bail out the dollarized country. In the process that led to EMU, the Maastricht fiscal criteria (on deficits and debt) as well as explicit rules (in the EMU Treaty) against bail-out of member countries were requested by Germany to ensure that a fiscal crisis in a member country would not lead to bail-outs or negatively affect interest rates and currency values of the monetary union. Two recent episodes suggest that the anchor country should be concerned about the fiscal soundness of a dollarizing economy. First, the external value of the Euro fell following the recent news that the EU would allow Italy to breach its fiscal targets for the year 2000; fiscal problems in one member country affected the external value of the common currency. Second, in the first half of 1995 the US dollar sharply fell following the fiscal and debt refinancing problem faced by Mexico after the devaluation of its currency. While financial problems in a dollarized economy should only affect the country risk spreads for the country experiencing difficulties, the possibility of contagion to the anchor country cannot be altogether excluded, especially if the dollarized country is relatively large. Fiscal problems in a small open economy (such as those in Central America) are unlikely to affect US asset prices (as Panama’s experience in the 1980s suggest). However, fiscal and financial problems in a relatively larger economy such as Argentina and Mexico could have negative spillovers for U.S. asset markets. Such negative externality is minimized if fiscal soundness is pursued by a dollarizing economy. External debt and financing requirements The stock of the external debt of a dollarizing country’s and its external debt servicing requirements will affect the success of dollarization. While it is correct that phenomena such as “sudden stops” of capital inflows and sharp reversals of capital flows (that have contributed to 28 some recent financial crises) may be reduced if a country gives up its domestic currency, the possibility of financial crises associated with excessive external debt positions cannot be ruled out. Just as cities and counties in a monetary area like the US can go bankrupt and experience a financial crisis, so can dollarized economies. For example, Panama had relatively large fiscal deficits that contributed to the build up of foreign debt up to the onset of the LDC debt crisis in 1982; it had reschedule its debt in the mid-1980s and eventually restructured/reduced the debt through a Brady deal following the late 1980s financial crisis (caused in part by U.S. financial sanctions). The possibility that excessive, short-term external debt may trigger a rollover/liquidity crisis or even lead to a country insolvency cannot be ruled out in a dollarized economy. “Sudden stops” may be dampened if the country follows prudent and disciplined fiscal policies but such financial crises cannot be ruled out altogether if external debt accumulation is excessive and public debt poorly managed. While a financial crisis may not significantly affect the anchor country if the dollarizing economy is small (and indeed the financial crisis in Panama had little effect on the U.S.), the anchor country should be concerned about the effects of such a crisis in a larger dollarizing economy (such as Argentina and Brazil). As the 1995 Mexican peso crisis episode suggests, financial crises in systemically significant emerging market economies can affect asset markets in the anchor country. Thus, the external debt and external financing requirements of a dollarizing economy are relevant factors in assessing whether the country is ready for a successful dollarization. Real and Trade Related Factors Ability to successfully pursue counter-cyclical monetary policy Some studies have suggested that some small open economies with a history of high inflation and high exchange rate volatility are effectively unable to use monetary policy for countercyclical purposes. A combination of unofficial dollarization, lack of policy credibility and wage indexation may render monetary policy ineffective to counter cyclical shocks. Worse, it has been argued that in some countries monetary policy may be pro-cyclical rather than counter-cyclical as negative external financial shocks may force monetary authorities to tighten monetary policy when a recession occurs to prevent excessive devaluation of the currency. Considering whether monetary policy has been able or unable to provide counter-cyclical output stabilization is thus a relevant criterion to assess whether dollarization will reduce the ability to smooth output fluctuations. It is thus a relevant criterion to assess the desirability of (if not country’s readiness to) dollarization. This ability will depend on a number of factors: how unofficially dollarized is the economy, what is the credibility of policy makers, what is the degree of indexation of wages, what is the degree of pass-through of exchange rates to domestic prices. Correlation of the business cycle with the US business cycle The need for exchange rate adjustment is reduced if a dollarized country’s business cycle is highly correlated with the one of the anchor country. If shocks hitting a common currency area are common to all the economies in the area, rather than being idiosyncratic national shocks, the need for currency adjustment is reduced and the monetary policy of the anchor country is more likely to be appropriate for the dollarized economy. The degree of synchronization of the business cycle, in turn depends on structural factors, such as the degree of trade integration and the similarity in production structure. Two caveats: 1. Synchronization is endogenous and can 29 increase as dollarization leads to greater trade and capital integration; 2. The degree of structural synchronization should be estimated separately from the role played by the exchange rate regime in driving a common business cycle. For example, under fixed rates, the business cycle of a follower country would be highly correlated to that of the leader regardless of structural interdependencies, just because the follower is forced to share the monetary policy of the leader. Trade integration with the US The greater is the share of a country’s exports and imports that is accounted for by trade with the United States, the better candidate is the country for dollarization. Greater trade integration implies greater synchronization of the business cycle of a dollarized economy with the anchor country. Also, trade integration is usually associated with greater financial and capital integration. By this criterion, Mexico, Canada and Central American countries are better candidates for dollarization than Argentina, Brazil and other Latin American economies. Vulnerability to terms of trade shocks The vulnerability to terms of trade shocks plays an ambiguous role. Such vulnerability is greater in countries whose exports are concentrated in a narrow range of, often primary commodity, exports. If the country is small and a price taker in the market for its export and import goods, it cannot modify its terms of trade (and negative shocks to the relative price of its exports) via currency devaluation. Thus, the benefits of dollarization are potentially larger for such small open economies that are price takers in international markets. On the other side, large shocks to the terms of trade may be need to (and will be successfully) absorbed via a nominal currency adjustment if the country has some export market power (and/or if the pass-through of depreciation to domestic prices is not full) so that a devaluation dampens the impact of terms of trade shocks on domestic relative prices. Diversified economies that are also large commodity exporters such as Canada and Australia have successfully used the currency tool to adjust to external terms of trade shocks. Instead, smaller open economies with lesser capacity to insulate domestic relative prices from terms of trade shocks may not gain as much from the option to devalue the currency following such external disturbances. However, even in such small open economies with given terms of trade, a nominal depreciation will lead to a reduction in the relative price of non-traded to traded goods and thus lead to a real depreciation of the currency that will benefit the traded sector. Thus, nominal exchange rate flexibility provided benefits even for such small open economies with exogenously given terms of trade. Note also that while a change in the real exchange rate can be obtained both via a change in the nominal exchange rate or through a change in the domestic price level, the adjustment costs of a real depreciation via price adjustment are usually considered to be larger than those implied by an exchange rate adjustment. Thus, the benefits of an adjustment of relative prices via exchange rate flexibility when the country has the ability to affect its real exchange rate via a nominal depreciation. An important caveat has to be kept in mind in considering whether a small country with no power over its terms of trade may benefit from exchange rate flexibility. While monetary policy may not affect its terms of trade, the country may still be able to partly use monetary policy to absorb external shocks, especially if non-tradable prices are sticky in the short run: the real exchange rate (as measured by the relative price of traded versus non traded goods) may be affected by monetary policy in the short-run. In this case, exchange rate flexibility will also 30 affect the domestic profitability of different sectors even if it does not affect the terms of trade of the country. Another factor to consider when discussing the role of terms of trade shocks is that fiscal revenues in many countries are significantly correlated with terms of trades: for example, oil and other commodity exports are a significant source of revenue for countries such as Mexico, Ecuador, Venezuela and Colombia. A real devaluation will thus tend to improve the fiscal balance of such a government as revenues are more related to the price of tradable goods while expenditure may be mostly on non-tradable goods; such fiscal adjustment is ruled out in a dollarized regime. Openness to trade Greater openness to trade, as measured by the ratio of exports and imports to GDP, may (with some caveats) strengthen the case for dollarization for a number of reasons. First, economies with large shares of international trade are usually small open economies that have little ability to affect their terms of trade. Second, the more open an economy, the greater the pass-through of nominal depreciation to domestic prices and thus the smaller the benefits of currency adjustment. Third, the larger is the share of the traded sector and the more diversified the export sector, the smaller is the change in the real exchange rate required to adjust to external shocks. On the other hand, one can argue that the benefits of currency adjustment may not be smaller for more open economies; while a given exchange rate depreciation will cause more inflation, it is also the case that less of a depreciation is required for the same counter-cyclical impact. The experience of Canada, Australia and other very open economies suggests that counter-cyclical policy may be quite effective in very open economies without causing more inflation. At the margin, greater openness will tip the balance in favor of dollarization if (given characteristics such as credibility, indexation and exchange rate pass-through), the effects of devaluation tend to be more inflationary than expansionary on output. Note that Mexico, Canada and Central American economies fare well on the openness scale while Argentina and Brazil are relatively more closed economies. Other Factors Flexibility of labor markets In the absence of the exchange rate mechanism, external shocks that require a change in real wages and/or mobility of labor across sectors will not have lasting effects on the rate of unemployment if there is enough flexibility in labor markets. This flexibility may need to take the form of downward flexibility of nominal wages (to induce a reduction in real wage if that is required), labor mobility across sectors and regions if changes in relative prices require a reallocation of factors of production, low hiring and firing costs to ensure labor market flexibility. In the absence of such labor market flexibility, negative external shocks may lead to protracted increases in the rate of unemployment and a permanent fall in the level of economic activity. Degree of labor migration Within a common currency area like the United States, adjustment to regional shocks (both transitory and permanent ones) is supported by inter-regional labor mobility. Labor will move 31 from regions negatively affected by economic disturbances to regions where economic activity is rising. In the case of countries considering dollarization, such free mobility of labor is ruled out as there are restrictions on cross-national labor migration. Still, in practice the degree of labor mobility may be significant as a number of countries in the America have a significant number of legal (and illegal) temporary and permanent migrant workers who can move between the United States and their country of origin. Examples are Mexico, El Salvador, other Central American countries and the Dominican Republic. The only important caveat is that, in the presence of significant labor rigidities (see bullet above), negative real shocks may lead to a significant increase in illegal immigration into the United States from a dollarized economy. Degree of capital mobility International mobility of real capital can, in part, substitute for the lack of labor mobility across countries. Shocks that require a movement across border of labor can be partly adjusted through movements of real capital. However, since the adjustment process working through the accumulation or decumulation of capital is slow, this adjustment mechanisms works only in the long run and is unable to shelter an economy from sudden external disturbances requiring a change in the relative returns to labor and capital. However, a higher degree of capital mobility in a dollarizing economy, as measured by flows of inward foreign direct investment (FDI), increases the likelihood that dollarization will be successful. Implicit or explicit fiscal federalism and income insurance schemes In a domestic currency union, idiosyncratic regional shock to output are partly compensated by a federal system of tax and transfers: a region hit by a negative shock will received federal transfers and pay less in taxes. Such an insurance mechanism partly shelters the regional income from shocks to regional output. Such an automatic insurance scheme cannot be at work in the case of a country that is dollarizing as monetary integration is not associated with fiscal integration. However, there are implicit forms of income insurance that may still be at work. If a dollarizing country has a large number of migrant workers in the anchor country, worker remittances may be an importance source of income for the dollarizing economy. For example, in El Salvador, remittances from citizens who work in the United States amount to over 10% of GDP per year. Also, a large stock of inward FDI and international equity portfolio diversification reduce the impact on the dollarizing country’s national income of negative output shocks as profit remittances to the anchor country automatically fall during recessions. Political factors To be successful dollarization requires a high level of public support; a broad and open political debate that shows a large popular and political support for dollarization will reduce the risk that political issues will derail dollarization. Countries that have deep political divisions, that have significant political minorities opposed to dollarization, that lack stable democratic institutions, that have a history of political turmoil may not be good candidates for dollarization. In fact, political support for dollarization would not be broad and there would be a greater probability that groups opposed to dollarization may at some point come to power and reverse the dollarization.