UNIVERSIDAD DEL CEMA

Buenos Aires
Argentina

Serie
DOCUMENTOS DE TRABAJO

Área: Economía e Historia

DOLLARIZATION DYNAMICS: A COMMENT

Emilio Ocampo y Nicolás Cachanosky

Junio 2023
Nro. 855

https://ucema.edu.ar/publicaciones/doc_trabajo.php
UCEMA: Av. Córdoba 374, C1054AAP Buenos Aires, Argentina
ISSN 1668-4575 (impreso), ISSN 1668-4583 (en línea)
Editor: Jorge M. Streb; asistente editorial: Valeria Dowding <ved@ucema.edu.ar>
 DOLLARIZATION DYNAMICS: A COMMENT

Emilio Ocampo and Nicolás Cachanosky1

Abstract

We analyze a recent paper that claims that dollarizing an economy in the presence of a “dollar
shortage” will provoke an immediate sharp reduction in real output and welfare. We find many
problems with the model that supports this conclusion: confusion about the nature of a dollar
shortage and its practical implications, invalid assumptions, invariant calibration in the
presence of a regime change and lack of empirical testing. In our opinion, the paper does not
make a valuable contribution to the dollarization debate nor provide useful guidance to
policymakers. The proposed model is based on unrealistic assumptions and its predictions are
contradicted by the available evidence. A more promising and useful line of research would
have been to investigate what happens to a dollarized economy in a scenario of exchange rate
overshooting and how it compares in relation to other stabilization plans.

JEL Codes: E10, E42, E50, F30, F33.

Keywords: Dollarization, Dynamics, Dollar shortage

1 Emilio Ocampo is Professor of Finance and Economic History at Universidad del CEMA (UCEMA) in Buenos Aires and
Nicolás Cachanosky is Associate Professor of Economics and Director of The Center for Free Enterprise at The University of
Texas at El Paso. We received valuable comments from Jorge M. Streb and an anonymous referee. Any mistakes are our sole
responsibility. The views expressed here do not necessarily represent those of Universidad del CEMA (UCEMA), The Center
for Free Enterprise at The University of Texas El Paso or The University of Texas at El Paso.

1
 Dollarization Dynamics: A Comment

The “shortage of dollars” seems unfortunately to mean all

Howard S. Ellis, The Dollar Shortage in Theory and Fact (1948)

1. Introduction

In a recent paper, Caravello, Martinez Bruera and Werning (CMBW, 2023) set out to analyze a

supposedly overlooked cost of dollarization: the recession induced by dollarizing an economy that faces

a “dollar shortage.”2 Based on a simple two-sector model with a representative agent and fixed and

flexible prices, to which they later add agent heterogeneity and financial frictions, the authors claim that

under such scenario, a dollarization would resemble an immediate sudden-stop and generate an initial

non-trivial reduction in output and welfare. Their model predicts that convergence to equilibrium could

take more than five years.

We think there are several issues with this paper. The first concerns the nature and implications of a

dollar shortage, the key underlying assumption of CMBW’s model.3 CMBW provide an imprecise

definition of what is a “dollar shortage.” Moreover, they assume it is not only an initial condition but

also a permanent one. However, the empirical evidence suggests that dollar shortages, even under

CMBW’s definition, are transitory and engendered by policy distortions. The implications in one case

or the other are very different. Secondly, the results of CMBW’s model are based on another key

assumption: with a dollar shortage it is only possible to dollarize at a “unfavorable rate.” In other words,

2 The following comments apply to the version dated June 2, 2023. A slightly different version was later published by the
NBER.
3 The concept of dollar shortage was used extensively in the post-war international economics literature (see for example Ellis,

1948, Graham, 1949, and Kindleberger, 1950).

2
they assume that in a scenario of free capital mobility and no foreign exchange restrictions, the

dollarization exchange rate must be significantly higher than the prevailing market exchange rate.

However, if the dollar shortage is permanent (as assumed by CMBW), this necessarily implies that

market participants behave irrationally. Third, the empirical evidence suggests that there is no univocal

relationship between a “dollar shortage” (as defined) and an “unfavorable” nominal exchange rate

(which they authors do no define but we assume means implying a real exchange rate significantly than

its long-term equilibrium value). The latter could exist without the former, and the former, if transitory

in nature, cannot lead to the latter.4 Therefore, the benchmark case of CMBW’s model –dollarization

cum dollar shortage– is empirically irrelevant and theoretically questionably. It would have been more

useful to analyze what would happen in a scenario of dollarization with prior overshooting and low

money balances and under such scenario to evaluate how dollarization compares to alternative

stabilization plans.

Fourth, CMBW built a model to suggest the likelihood of potentially “dire consequences” of dollarizing

an economy facing a “dollar shortage” but nowhere in their 52-page paper there is a discussion of

whether the model’s predictions fit the facts.5 Also, the model is calibrated with parameters that do not

consider that dollarization is a regime change in the sense given by Sargent (1982).6 Therefore, model

parameters cannot be assumed to remain invariant (Lucas, 1976). CMBW implicitly assume that the

only structural change brought about by dollarization is that the nominal exchange rate remains

inalterably fixed. On a minor note, CMBW only mention five papers on dollarization and omit many

valuable works in a vast literature produced on the subject since 1993 (see for example some of the

works cited in Ocampo, 2023).7

4 If, as we argue here, the dollar shortage is transitory (i.e., the borrowing constraint is not permanent), then the market
exchange rate cannot be “unfavorable” unless there is overshooting (Dornbusch, 1976).
5 In a footnote CMBW incorrectly assert that “Ecuador dollarized in 2000-01.” Dollarization was announced on January 10,

2000, officially closed by June 8. By the end of September approximately 97% of all existing sucres had been taken out of
circulation. In other words, the Ecuadorean economy was fully dollarized by the end of 2000.
6 From a policy standpoint, dollarization must be accompanied by other structural reforms. It wouldn’t make sense without

them. This was the case in Ecuador and in Argentina with the Convertibility Plan.
7 Also, CMBW incorrectly cite the paper by Goldfajn and Olivares (2000) by adding two commenters as authors. It suggests

they did not even read the paper.

3
Finally, although CMBW’s model supposedly has universal applicability, it was built with Argentina

in mind (the authors are Argentine) and is specifically calibrated for Argentina. CMBW note that

“recent proposals for dollarization, previously rejected, have resurfaced and gained traction in

Argentina in the midst of the presidential race and escalating inflation.” The objective of CMBW’s

paper is to discredit such proposals and warn about the potentially “dire consequences” of dollarization.

For all the above reasons and others that we will explain in more detail below, we believe CMBW’s

paper does not make a meaningful contribution to the dollarization debate nor offers valuable guidance

to policymakers. Their mathematical model is based on a definition of a dollar shortage that is

conceptually unsound and empirically irrelevant. Even though, according to most estimates, private

sector dollar holdings amount to more than 40% of GDP, CMBW claim that the country faces a dollar

shortage that they estimate at 2.5% of GDP imposes a permanent binding constraint on the economy.

Only with a set of unrealistic initial conditions and questionable micro-foundations the model built by

CMBW can yield its apocalyptic predictions. A more promising and useful line of research would have

been to investigate which monetary policy framework and foreign exchange regime can deliver superior

results in a scenario in which, to stabilize the economy and establish a solid foundation for sustainable

growth, policymakers must exit an unsustainable regime in a relatively short period of time due to the

electoral calendar, but a) have no credibility and no effective commitment device other than

dollarization to reduce time inconsistency (see Ocampo, 2023), and b) face heightened inflationary

expectations.

2. Nature and Practical Implications of a Dollar Shortage

The most important prediction of CMBW’s model –that dollarization will inevitably lead to a sharp and

lasting contraction in output– assumes the existence of a “dollar shortage” that constrains the economy

for an indefinite period. In fact, CMBW assert that the “most important” parameter of their “simple”

model is “is the size of the shortfall in initial dollars relative to its steady state value (p.4).” Therefore,

it is critical to understand the nature and practical implications of a “dollar shortage.”

4
The closest CMBW get to a definition of the term is when they argue that “due to a shortage of foreign

reserves or a lack of foreign credit, the conversion of domestic currency to dollars at an unfavorable

rate [sic]… leaving the initial balance of foreign currency below its long-run steady-state value.” But

what is an “unfavorable rate”? The authors don’t provide an answer either. Presumably, it is a rate that

implies an historically high real exchange rate and/or a rate that is not warranted by macroeconomic

fundamentals (i.e., a non-market exchange rate.) Later CMBW clarify that “dollars are scarce in the

precise sense that money balances are below their steady state (p.11).” In the calibration section, they

add that “at the pre-dollarization exchange rate, there are not enough dollars to convert all existing pesos

(p.22).” They omit to clarify whether the pre-dollarization exchange rate is the one artificially set by

the central bank below its market clearing level or some other rate. 8

CMBW’s model predicts that in the presence of such “dollar shortage”, a dollarization would be

immediately followed by a sharp and lasting recession.

“In our model, if initial balances are large enough the economy reaches a steady-state

equilibrium immediately, seamlessly switch from domestic to foreign currency.

However, this is no longer the case when dollars are initially scarce, due to limited

reserves, limited credit and low pre-existing dollar holdings. In this scarcity situation,

the economy undergoes a non-trivial transition towards its steady state (p.3).”

In summary, according to CMBW, a “dollar shortage” occurs if: a) international reserves are close to

zero, or significantly below the monetary base when converted at a “favorable” (prevailing? official?

market?) exchange rate, and b) there is no access to external financing (“bt = 0 for all t ≥ 0”, p.8).

Although not a necessary condition for a dollar shortage, CMBW also assume that aggregate money

balances pre-dollarization, as measured by M0, are below their steady state level (and even lower after

dollarization).9

8 Things get more complicated when the foreign exchange market is tightly controlled by the authorities, as is currently the
case in Argentina. It is hard to define what the “prevailing” exchange rate is today, as in effect, there are multiple exchange
rates.
9 In fact, in CMBW’s model pre-dollarization money balances significantly below their steady state level would seem to be

more important that the exchange rate at which they are dollarized.

5
Two consequences are derived from this set of initial conditions. First, a dollarization must necessarily

be implemented at an “unfavorable” exchange rate. Second, post-dollarization money balances, as

measured by M0/GDP, will be significantly below their steady state. Third, only an adjustment of money

balances, which requires a sharp drop in the consumption of tradables and recurrent trade surpluses, can

bring back the economy back to equilibrium. As CMBW explain, “individuals cut their consumption to

build up their money balances (p.14).” In the benchmark case, this brings about “potentially dire

consequences”: an immediate and lasting contraction in output and welfare. Convergence to equilibrium

takes a long time. Under all scenarios CMBW’s model predicts that it will take more than four years

for consumption and output to return to their steady state equilibrium.

3. Model Assumptions, Calibration. Empirical Testing and Extensions

Assumptions

A key assumption underlying CMBW’s model is that the effects of an initial “dollar shortage” are

lasting, even after dollarization, due to a permanent borrowing constraint. This is a strong assumption.

Under this scenario, CMBW claim that the exchange rate for dollarization will be “unfavorable” and

significantly higher than the market exchange rate prevailing immediately before dollarization. Given

that they also assume free capital mobility (p.6), this proposition implies that participants in the foreign

exchange market behave irrationally. Otherwise, the market exchange rate pre-dollarization could not

be significantly below the level implied by full convertibility. Second, there could be reasons other than

a dollar shortage (as defined by CMBW) for a country to dollarize at an “unfavorable” exchange rate

(e.g., pre-dollarization overshooting a la Dornbusch, 1976). Finally, under a dollar shortage scenario,

any stabilization plan that uses the nominal exchange rate as an anchor would face similar dynamics to

dollarization. 10 Third, in CMBW’s model the dollar shortage can only correct itself through an

adjustment in money balances which requires a trade surplus and a reduction in aggregate consumption

(see pp.8-9, 11, 14).

10 CMBW seem to be taking a Nirvana approach to the problem of an “unfavorable rate”. In this context, a nirvana fallacy
consists in comparing the results of a realistic policy framework with those of an idealized one instead of those of a realistic
alternative (see Demsetz, 1969).

6
CMBW assert that Argentina currently faces “a dollar shortage” (p.2). However, the lack of foreign

reserves at the central bank is the direct consequence of an unsustainable foreign exchange policy.11 It

is therefore a transitory phenomenon, unlike the chronic dollar shortage that Europe faced in the

immediate postwar period under the Bretton Woods system. 12 No structural limitation prevents

Argentina’s central bank from accumulating foreign reserves, but a wrong-headed policy that can be

reversed. Basic microeconomics teaches us that if the government artificially sets the price of a good

below its equilibrium level (i.e., where demand and supply meet) this will inevitably create a shortage.

It is very easy to correct this problem: remove price controls and let the market reach its equilibrium.

More importantly, even if net foreign reserves are low immediately after the liberalization of the foreign

exchange market, the exchange rate will not necessarily overshoot.13 Market participants do not look at

the current level of foreign reserves but at the level achievable with sustainable policies.14 The empirical

evidence suggests that this is the case. With respect to the borrowing constraint, it is evident that

Argentina has limited access to voluntary sources of financing. However, if Argentina’s long history of

defaults teaches us anything, is that borrowing constraints do not last forever.

On December 7, 2015, net international reserves at the Argentine Central Bank (BCRA, 2016, p.35)

amounted to US$6.8MM, while the monetary base converted at the free-market exchange rate amounted

to US$44 billion. At the time, the sovereign had no access to the international capital markets due to

the refusal of Cristina Kirchner’s government to accept the rulings of a New York judge. According to

CMBW’s definition, Argentina faced a dollar shortage. However, ten days later, after a new government

took office, the BCRA relaxed controls on foreign exchange transactions and let the peso float. The

official exchange rate moved from 9 AR$/US$ on December 7 to 13 AR$/US$ pesos on December 31,

2015 (BCRA, 2015). By the 31st of January 2016, net foreign reserves had increased to US$ 8.5 billion.

By April, the government reached a settlement with holdout creditors which reopened the capital

11 For refutation of the hypothesis that there are not enough dollars to effect a dollarization at a “favorable” exchange rate see
the report by Anker (2023).
12 In both cases however, the shortage is created by the monetary authority. “The ‘shortage’ of dollars in the foreign world is

therefore solely and simply explicable in the fact that the dollar is made available to residents of foreign countries, in
necessarily rationed amounts, at lower prices in their respective currencies than in a free market they would be prepared to pay
for the given supply.” (Graham, 1949, pp.3-4).
13 Overshooting is possible and perfectly rational under certain circumstances (see Dornbusch, 1976).
14 This is particularly true when the market expects a regime change, which is the case in Argentina today.

7
markets for the sovereign. In other words, the much feared “dollar shortage” quickly disappeared.

Argentina was able to tap the markets again within two quarters and to issue a 100-year bond at a fixed

annual coupon of 7.125% within six quarters. It is important to keep these facts in mind given the slow

convergence (more than four years) predicted by CMBW’s model.

To conclude, a “dollar shortage” as (defined by CMBW) is not a useful concept to analyze the potential

impact of dollarization. First, it is not permanent. Second, a “dollar shortage” may or may not lead to

an “unfavorable exchange rate”, and an “unfavorable exchange rate” may or may not be caused by a

“dollar shortage.” In our view, it would have been more productive for CMBW to consider building a

model that predicts: a) what would happen to an economy that, for whatever reason, dollarized in a

scenario of overshooting, and b) under such circumstances, in which way the results of a dollarization

would differ from alternative stabilization plans (with fixed or flexible exchange rates.)15 Regarding the

first question there is a precedent that CMBW chose to ignore even though they mention it as “a country

that famously went down this extreme path”: Ecuador.

At the time of dollarization, January 2000, Ecuador was in default of its Brady bonds, its banking system

was mostly bankrupt, and the inflation rate was running at an annualized rate close to 400% per annum.

On December 31, 1999, the prevailing market exchange rate was 20,000 sucres per dollar.16 Although

at such rate the Ecuadorean central bank had more than enough foreign reserves to “buy” the entire

monetary base, it was insolvent on a mark-to-market basis (i.e., it had a negative net worth).17 On

January 10, 2000, President Mahuad announced the dollarization of the economy at a conversion rate

of 25,000 sucres per dollar. This implied a 20% depreciation of the domestic currency in only 10 days.

At this nominal rate, the implicit real exchange rate was twice the average level for the period 1994-

15 In one of the extensions to their model, CMBW claim that if the price of the non-tradable good is too high at the outset, the
transitional dynamics of the model would be similar under a fixed exchange regime or a currency board (p.4).
16 The sucre had been floating since February 1999.
17 According to official figures, as of December 31, 1999, Ecuador’s net foreign reserves amounted to US$ 808 billion, and

the monetary base converted into dollars at the market exchange rate to US$ 770 billion. Technical insolvency is not necessarily
a problem for a central bank, particularly if a large portion of its liabilities is long term or owed to multilaterals. The problem
in Ecuador in January 2000 was that half of the assets in central bank’s balance sheet were government debt.

8
1998. However, approximately 25% of the exchange rate overshooting took place during 1999, before

dollarization.18

Several points worth mentioning. First, by May 2000, i.e., less than two quarters of dollarization,

Ecuador had reached an agreement with the IMF and the Paris Club and by August it had completed a

restructuring of its outstanding debt. Second, within four quarters of dollarization, net foreign reserves

had increased by 23%. Third, monetization increased rapidly without pushing the economy into a

recession. Fourth, the banking system was able to perform its intermediation role more efficiently after

dollarization. Total credit to the private sector almost doubled in US$ between December 1999 and

December 2004. More importantly, the percentage of low-income households with access to credit

increased from 24.4% in 1999 to 27,3% in 2005/6. The dollar amount of loans to this group almost

tripled in real terms and doubled as a percentage of household income (see Bebczuk, 2008).

Even if we consider that the exchange rate at which Ecuador dollarized meets the definition of

“unfavorable,” the overshooting reflected an unprecedented economic, financial and political crisis that

unfolded during 1999 and reached its climax in January 2000 more than a scarcity of dollars at the

central bank in January 2000 (whatever was left of the banking system was highly dollarized).19

Argentina’s Convertibility offers two interesting episodes to test CMBW’s theory. The most obvious

one is its establishment on March 31, 1991. Despite what the Convertibility Law required, the central

bank did not at the time have enough liquid dollar foreign reserves to “buy” the entire monetary base.20

Also, the sovereign was in default of its external bank debt and did not have access to the international

capital markets (in February 1991 the country risk premium was 1600 basis points). Finally, following

several bouts of hyperinflation in 1989 and 1990, money balances were at an historically low level

(M0/GDP was at 2,8%). However, it is hard to argue that Convertibility was effected at an “unfavorably”

high exchange rate (in real terms it was significantly below the historical average).21 According to

18 At the closing exchange rate on 31 December 1999, the real exchange rate index (1994=100) was at 173.37. At the end of
January 2000, at 25,000 sucres per dollar, the index was at 206.51. In comparison, the index average for 1994-1998 was 99.57.
19 See Jacome (2004) for a discussion of the factors that contributed to exchange rate overshooting in Ecuador in 1999.
20 The available data does not allow us to estimate a mark-to-market balance sheet for Argentina’s central bank as of March

31, 1991, but technical insolvency cannot be discarded. The monetary authorities were much less transparent than currently.
21 According to some estimates, the real exchange rate in 1991 diverged as much from its long-term equilibrium value –almost

25%– as in 1980. See Espert and Maino (2000).

9
CMBW’s model, the adjustment of money balances should have led to a sharp contraction in

consumption and a surplus in the trade balance. Instead, consumption grew 12,6% in 1991 and 10,8%

in 1992 (GDP at 8,9% and 8,7% respectively), while the trade balance surplus fell slightly in 1991 and

turned negative in 1992. Net reserves doubled in the first four quarters following Convertibility and

Argentina had regained full access to the international capital markets eleven quarter later.22

The second and less obvious precedent is the traumatic exit of Convertibility between December 2001

and January 2002. The Argentine government managed to engineer the set of initial conditions imagined

by CMBW: a sovereign debt default that imposed a lasting borrowing constraint, a devaluation that led

to an overshooting of the exchange rate to “unfavorable” levels (the nominal exchange rate remained

practically unchanged at 3 pesos per dollar from January 2003 until December 2006), a sharp reduction

of money balances below their steady state equilibrium through a partial confiscation of dollar

denominated deposits, and an asymmetric “pesification” of bank liabilities that crippled the banking

system and rendered it unable to fulfill its basic intermediation function. The outcome of this policy

mix matches the predictions of CMBW’s model: real wages and consumption dropped, an unprecedent

string of trade and current account surpluses followed and it took real GDP more than four years to

return to its previous peak despite a significant increase in the terms of trade. The exit from

Convertibility was a non-virtuous regime change that entailed not only breaking a hard peg but also the

rule of law. It generated results that are the opposite of those that would result from implementing a

dollarization with structural reforms.

Argentina’s current situation in some respects is worse than in 1991 –according to some estimates, net

foreign reserves at the central bank are negative– but it is also easier to reverse. History shows, that if

the incoming government announces a credible regime change (such as dollarization accompanied by

a program of structural reforms), Argentina could regain access to the international capital markets in a

relatively short time.23

22Ministerio de Economía (1993).

10
As we have already explained, the concept of a permanent “dollar shortage” does not seem to have

much relevance theoretically or empirically. Therefore, the base case scenario for which the model

yields its predictions –dollarization cum dollar shortage– is unrealistic.24 Also, CMBW take a very

simplistic view of how dollarization is implemented. There is no “one size fits all” dollarization. Each

country dollarizes its own way given a set of economic, financial, political, legal and institutional

constraints. For the purposes of CMBW’s model, implementation details cannot be assumed away. For

example, in Ecuador, the US dollar replaced the sucre as legal tender by a law approved by Congress.

Ecuadoreans had a fixed period to exchange their sucres for dollars. Initially, the rate of dollarization

was relatively slow: after four months only 30% of the total sucres in circulation had been exchanged.

As a result, over a period of nine months the economy had a bi-monetary system in which dollar and

sucre bills co-existed (however, the banking system was fully dollarized).25 El Salvador implemented

dollarization differently. Instead of replacing the domestic currency –the colón– it simply made the

dollar legal tender. In essence, the law created a two-currency economy (the banking system was fully

dollarized). In this case, the dollarization curve was flatter than in Ecuador but, after four months, only

30% of the colones in circulation had been swapped for dollars. It took two years for the dollarization

rate to reach 90%. This means that a main assumption of CMBW’s model –instantaneous dollarization–

is not only unlikely irrespective of the format chosen but also inconsistent with the empirical evidence.

Dollarization must be implemented at a market exchange rate, i.e., the rate determined by the interaction

of supply and demand once all restrictions on foreign exchange transactions and capital movements

have been lifted. This must necessarily be the most “favorable” exchange rate, even if there is

overshooting. In the case of Argentina, current economic policy, which in the last two years has led to

a strong appreciation of the peso in real terms, is unsustainably and must necessarily be corrected, with

or without dollarization. Otherwise, it will be impossible to stabilize the economy. To put the onus of

liberalizing foreign exchange markets on dollarization is to fall into a Nirvana fallacy.26

24 This is not an assumption about agent behavior within the model but about the scenario for which the model yield predictions.
25 Ecuador also minted fractional coins (less than one dollar) of an uncertain legal status.
26 See note 9 above.

11
CMBW assume the terms of trade remain unchanged but ignore the fact that the initial mega devaluation

generates a significant one time increase in the domestic price of the tradable good. Presumably this

would also lead to an increase in output, which would in turn increase the trade balance and lead to

higher inflows of dollars. However, for the benchmark case the model’s dynamics “are solved without

reference to the non-tradable sector” (p.3). The reason is that CMWB assume that the “output of the

tradable good is given by a constant endowment” (p.7). In other words, the production of the tradable

good does not require labor nor technology and has zero price elasticities. The model parameters remain

invariant even though in the real-world producers of the tradable good would have an incentive to

increase production. This is a very strong assumption that not only seems to defy basic microeconomics

but also magnifies the predictions of their model.

Also, in an economy with a high degree of de facto dollarization (like Argentina today or Ecuador in

January 2000) it would be highly unlikely for the private sector to have a dollar shortage even under

the conditions spelled out by CMBW. In other words, the shortage of dollars affects mostly the central

bank. However, in CMBW’s model, it is the private sector that must adjust money balances to corrects

the dollar shortage. See for example the following statement.

“Proposition 1 shows that dollarizing when dollars are scarce effectively induces a

“sudden stop” that lowers consumption of tradables throughout, creating a surplus in

the current account to accumulate dollars… Intuitively, individuals cut their

consumption to build up their money balances. Put differently, the shortage of money

increases the shadow interest rate, inducing individuals to save. Collectively, these

efforts create a current account and trade surplus that increases dollar money balances

(p.14).”

There is a contradiction in CMBW’s argument. A dollar shortage as defined supposedly requires an

increase in money balances (which reduces consumption). But the private sector, which accounts for

80% of money balances, doesn’t really have a dollar shortage (at least in Argentina). For the argument

to work in the base case, requires two other strong assumptions: a) pre-dollarization dollar holdings are

12
highly concentrated (significantly more than peso holdings), and b) the banking system will never be

able to intermediate those holdings (i.e., private agents with a dollar surplus cannot lend them to private

agents with a dollar deficit). CMBW explicitly make this assumption when extending their model to

include heterogenous agents and incomplete markets (pp.4-5).

Finally, as we have made it clear in several articles (Ocampo and Cachanosky, 2023), in the case of

Argentina, it would not be advisable to move from the current highly regulated and controlled FX

regime to a new hard currency regime (dollarization or other) without a transition period, which as a

first step would involve the liberalization of the foreign exchange market and the lifting of price controls.

With the current relative price structure, the economy will not reach its full potential. Therefore,

imposing a monetary straitjacket on this structure would make it more difficult to achieve such goal.

To conclude, CMBW’s model predicts that a deep and lasting recession (up to five years or more) would

follow if a country dollarized at an “unfavorable” exchange rate in the presence of: a) a permanent

borrowing constraint (i.e., no foreign borrowing or investment), b) historically low levels of

monetization, c) inflexible prices, d) a highly concentrated tradable sector with zero supply elasticities,

e) a banking system that cannot intermediate savings. In this imaginary world, the only mechanism that

allows the economy to return to steady state equilibrium is a lasting reduction in the consumption of

tradables. It would seem CMBW chose their assumptions about the initial conditions so that the model

could yield a pre-specified set of predictions.

We agree that unrealistic assumptions are not problematic if a model yields verifiably accurate

predictions. In this case, not only some of the model’s assumptions about agent behavior are unrealistic

(e.g., no use technology and zero elasticities in the production of tradables) but, more importantly, also

about initial conditions. In other words, the model’s predictions are only valid in an imaginary world

that has little connection with reality. And even though CMBW did not empirically test their model, the

available evidence seems to contradict its apocalyptic predictions. When reading CMBW’s paper we

are reminded of Caballero’s warning about “quantitative mathematical formalizations of a precise but

13
largely irrelevant world” that are “useless as a tool for understanding significant events and dangerous

for policy guidance (Caballero, 2010, p.92, 100).”

Calibration

CMBW calibrate their model for Argentina and claim that “that the transitional dynamics” (i.e., costs)

of dollarization “may be significant.” They recognize that given the “large uncertainty regarding the

implementation of a dollarization policy, any calibration is admittedly speculative (p.22).”27 We fully

agree with this last statement and would add that the speculative calibration of their model puts into

question the validity and relevance of its predictions.

CMBW do not explicitly quantify the dollar shortfall in terms of the difference in the exchange rate

before and after dollarization but indirectly through dollarized money balances. For Argentina, CMBW

estimate the steady state money balances (M0/GDP) at 5%. They then compare this figure with the

levels immediately before and immediately after dollarization, which they estimate at 4% and 2,5%

respectively. The “dollarization cum dollar shortage” scenario is calibrated using the latter figure, which

CMBW consider “a reasonable starting point” (p.22). This would imply a devaluation of the peso of

approximately 50%. In other words, they imply dollarization would be effected at the current “informal”

(blue) market exchange rate (without having previously eliminated all restrictions). As mentioned in

the previous sub-section, dollarization must be effected at the market exchange rate once all restrictions

are lifted which does not necessarily have to equal the “informal” exchange rate pre-dollarization.28

CMBW claim that the large amount of dollars currently held by the private sector is an irrelevant factor

and that the sharp and immediate reduction in the dollar value of M0 at the “unfavorable” rate would

negatively affect the public sector, as it would have “to pay the private sector dollars in exchange for

pesos that are useless (p.22).” Even if we accepted the validity of the dollar shortage concept, there are

27 When all the local currency is exchanged for dollars, the central bank no longer has a monetary liability (see Beckerman,
2001, p.25). In such circumstances, monetary circulation is hard to estimate because the inflow of physical dollars is not
directly measurable (see Vera, 2007).
28 As of April,2023, the IMF estimated a real exchange rate overvaluation of between 10% and 25% (IMF, 2023, pp.6,53).

14
two strong assumptions implicit in this statement that are questionable. First, that dollarization occurs

instantaneously. Second, that existing private sector US dollar holdings are irrelevant.

As mentioned in the previous sub-section, even if the dollar were to be imposed as exclusive legal

tender (as was the case in Ecuador), dollarization would only occur gradually (how fast would depend

on several factors, the most important of which is the conversion exchange rate.) It took nine months in

Ecuador to complete dollarization and almost two years in El Salvador.

With respect to the second assumption, CMBW explicitly dismiss the argument that Argentines are

already heavily dollarized (p.24). According to the latest official statistics, as of December 31, 2022,

total liquidity denominated in US$ or other foreign currencies outside of Argentina or the Argentine

banking system amounted to US$ 243 billion (see INDEC, 2023, p.6). This figure does not include

undeclared assets. According to some estimates, Argentines hold approximately US$200 billion –some

of which are undeclared for tax purposes– in dollar bills within the borders of Argentina (Gorodisch,

2021). CMBW claim that these “aggregate dollar holdings is [sic] not the correct measure of liquidity.”

Why would aggregate peso holdings be the correct measure for money balances but not aggregate dollar

holdings? Supposedly because the distribution of the latter is concentrated among “affluent” individuals.

However, the same can be said about peso holdings.29 Contrary to CMBW’s assertion, dollarized money

balance in Argentina would not have to adjust upwards following dollarization because Argentines do

not have a shortage of dollars. History proves that with a credible regime change, a portion of this

liquidity (that portion already declared to the tax authorities) would be repatriated and/or deposited in

domestic banks which would lead to a monetary and credit expansion. However, this is impossible in

CMBW’s world, in which the banking system is never able to fulfill its normal functions, i.e., it cannot

intermediate savings. But both the theory and the evidence suggest that after dollarization the level of

financial development increases significantly. For example, whereas in 2002 the ratio of bank credit to

the private sector to GDP in Argentina and Ecuador were similar, in 2022 it was twice as large in the

29For example, according to official statistics, as of March 31, 2023, 22% of the holders of peso denominated time deposits
in the Argentine banking system accounted for 86% of total peso denominated time deposits (Information available at
https://www.bcra.gob.ar/PublicacionesEstadisticas/Cuadros_estandarizados_series_estadisticas.asp ).

15
latter. The assumption that financial frictions and markets are invariant to dollarization does not make

sense.

Contrary to CMBW’s assertion, instead of a sudden stop, dollarization would most likely provoke a

massive inflow of dollars and a second monetary expansion through the banking system (this is what

happened in Argentina under Convertibility and in Ecuador after dollarization). Also, the private sector

currently accounts for 80% of total bank deposits in Argentina. Therefore, any pre-dollarization

monetary imbalance, must be corrected by decisions taken by private agents. Finally, following

dollarization, base money (M0) will cease to have the meaning and significance usually accorded to it.

In fact, it will become increasingly difficult to estimate it with precision.30 Therefore, whether M0 rises

after dollarization, rapidly or gradually, or not at all to a steady state level would be irrelevant. A broader

monetary aggregate such as M1 or M2 would be a more appropriate input for the model.

It is worth mentioning that when calibrating their model CMBW do not consider the “cepo” (current

restrictions on foreign exchange transactions and capital movements) to characterize the initial

conditions. They implicitly assume that foreign exchange market is completely liberalized at t=0 or that

the prevailing exchange rate is a market rate. The fact that is not the case, raises one of the most complex

issues that the incoming government will have to resolve to stabilize the economy.

Empirical Testing

According to CMBW, the dynamics of a dollarization in which the domestic currency is converted at

an “unfavorable” rate are similar to those of a sudden stop: “a temporary drop in the consumption of

tradable goods to accumulate of [sic] foreign currency.”31 An inflation rate significantly above the US

compensates for the initial overshooting of the exchange rate.

“When prices are not fully flexible a recession initially ensues—even when prices are

fully adjusted upon announcement of dollarization—because the exchange rate is

30 “It is not comparable with the monetary base before dollarization, since currency in circulation after dollarization consists
of dollar currency not issued by the Central Bank” (Beckerman, 2001, p.25) and Vera (2007).
31 An economy facing a CMBW’s dollar shortage is an economy that that is still suffering the consequence of a sudden stop.

16
 overvalued, relative to the flexible price equilibrium. This recession eventually gives

way to a boom as the economy accumulates dollars and the exchange rate falls behind

and becomes undervalued, relative to the flexible price equilibrium (p.3).”

Earlier in the paper, CMBW explain their logic: “tradable consumption falls as agents seek to save to

build up their stock of foreign currency. Put differently, the low real money balances raise the domestic

interest rate (or shadow rate), and this lowers spending (p.3).” Following this logic, CMBW argue that

“given that quantity of money starts below steady state and since the economy does not have access to

capital markets, a drop in tradable consumption is required to rebuild money balances.” According to

CMBW’s model convergence to equilibrium is very slow:

“The price of non-traded goods falls and gradually recovers back to steady state, as

money balances are gradually rebuilt… convergence takes a long time: even after 16

quarters (4 years), the stock of money is around 15 percentage points below steady state

(p.23).”

Neither of the above statements squares with the facts. Ecuador implemented dollarization at what could

be described as an “unfavorable rate” (but not due to a CMBW “dollar shortage”, as net foreign reserves

were enough to buy the monetary base at the market rate.) Therefore, the logical thing for CMBW to

do would have been to test their model against the empirical evidence. As the table below in the

Appendix shows, in the case of Ecuador, the key macroeconomic variables moved in a direction that is

the opposite of what CMBW’s model predicts (see the Appendix).

Extensions

CMBW extend their model in two ways. First, they assume that prices are not flexible and the initial

price of the non-tradable good is not reset with dollarization (i.e., it is too high). Second, they add agent

heterogeneity. With the first extension, the initial recession is deeper and longer lasting. No surprises

here. With respect to the second extension, they assume the economy is made up of two groups of

agents: “the first group are rentiers specialized in the tradable sector, living off their endowment of the

export good (e.g., commodities); the other group of agents is specialized the non-tradable sector,

17
providing labor to produce non-tradable goods.” Another critical assumption is that financial markets

are imperfect and incomplete, “so these groups cannot borrow from one another, or insure each other.”

Although according to CMBW their model supposedly has universal applicability, it was built with

Argentina in mind (the authors in fact are Argentine). Modeling the tradable sector in this manner is

questionable on empirical and theoretical grounds.32 It implicitly assumes that capital and technology

are irrelevant and a regime change such as dollarization would have no positive impact on incentives

and output.33 Basically, in the world imagined by CMBW the production of tradables has zero supply

elasticity. A similar view has inspired the misguided economic policies that have led to Argentina’s

economic decline.

Be it as it may, this extension of the CMBW model leads to a strong conclusion: “heterogeneity has

non-trivial implications for the aggregate dynamics, but most even more noteworthy is the breakdown

across groups. In particular, dollarization may disproportionally hurt the group specialized in the non-

tradable sector.” In other words, according to CMBW “dollarization cum dollar shortage” does not only

lead to a sharp and lasting output reduction but also greater inequality, benefiting the “landed oligarchy”

at the expense of the masses of urban workers. Ecuador may not serve as a proper empirical test of this

hypothesis since its tradable sector is mostly controlled by the State. However, it may be worth pointing

out that inequality fell after dollarization.

4. Conclusion

In summary, we find that CMBW’s contribution to the dollarization debate is limited. Their

mathematical model is based on a definition of a dollar shortage that is conceptually unsound and

empirically irrelevant. It predicts what would happen in an imaginary world that differs from reality in

very fundamental ways. Unrealistic assumptions are not a problem if a model’s predictions are

32See for example Mundlak, Cavallo and Domenech (1989) for an alternative view.
33CMBW assume no change in the terms of trade but ignore the fact that the initial mega devaluation generates a significant
one time increase in the domestic price of the tradable good, and presumably, would also lead to an increase in output, which
would in turn increase the trade balance and lead to higher inflows of dollars. However, for the benchmark case the model’s
dynamics “are solved without reference to the non-tradable sector” (p.3). In fact, CMWB assume that “output of the tradable
good is given by a constant endowment” (p.7). The model parameters remain invariant even though producers of the tradable
good would have an incentive to increase production. This assumption defies basic microeconomics.

18
verifiably accurate. This is not the case with CMBW’s model. Its predictions are contradicted by the

available evidence. CMBW’s apocalyptic warning of potentially “dire consequences” of dollarizing the

Argentine economy is groundless and offers no valuable guidance to policymakers. A more promising

and useful line of research would have been to investigate which monetary policy framework and

foreign exchange regime can deliver superior results in a scenario in which, to stabilize the economy

and establish a solid foundation for sustainable growth, policymakers must exit an unsustainable

repressed regime but lack credibility and limited ability to build it in the short time afforded by the

electoral calendar. This is the challenge the next government of Argentina will face. The probability of

failure is high. There are no easy solutions.

5. References

Anker Latin America (2023). “Special Report. Is dollarization viable?”, 24 May. Buenos Aires.

Bebczuk, R.N. (2008). “Dolarización y Pobreza en Ecuador,” CEDLAS, Working Papers 0066,

CEDLAS, Universidad Nacional de La Plata.

Beckerman, P. (2001). “Dollarization and Semi-Dollarization in Ecuador,” World Bank Working Paper

2643 (July).

BCRA (2016). Informe de Política Monetaria, Mayo 2016. Buenos Aires: BCRA. Available at

https://www.bcra.gob.ar/Pdfs/PoliticaMonetaria/IPM_Mayo_2016.pdf

Caballero, R. (2010). “Macroeconomics after the Crisis: Time to Deal with the Pretense-of-Knowledge

Syndrome.”, Journal of Economic Perspectives, Vol. 24, No. 4, (Fall), pp. 85-102.

Caravello, T. E., Martinez Bruera, P. and Werning, I. (2023). Dollarization Dynamics. NBER Working

Paper 31296 (June).

Demsetz, H. (1969). “Information and Efficiency: Another Viewpoint,” The Journal of Law &

Economics, Vol. 12, No. 1 (April), pp. 1- 22.

19
Dornbusch, R. (1976). “Expectations and Exchange Rate Dynamics,” Journal of Political Economy,

Vol. 84:6 (December), pp.1161-1176.

Gorodisch, M. (2021). “Cuántos dólares tienen los argentinos: los mayores tenedores fuera de EE.UU.”,

Cronista, September 26. Available at https://www.cronista.com/finanzas-mercados/cuantos-dolares-

tienen-los-argentinos-en-el-colchon-estiman-que-10-de-los-billetes-en-circulacion/

Ellis, H. S. (1948). “The Dollar Shortage in Theory and Fact,” The Canadian Journal of Economics and

Political Science / Revue canadienne d'Economique et de Science politique, Vol. 14, No. 3 (August),

pp. 358-372.

Espert, J.L. and Maino, R. (2000). “On the relationship between real exchange rate and public spending:

The case of Argentina,” mimeo.

Goldfajn, I. and Olivares, G. (2000). “Full Dollarization: The Case of Panama,” Economía Journal,

Vol.1-No.2 (Spring), pp. 101-156.

Graham, F.D. (1949). The cause and cure of the dollar shortage. Princeton, New Jersey: Princeton

University Press.

INDEC (2023). “Balanza de pagos, posición de inversión internacional y deuda externa. Cuarto

trimestre de 2022,” Informes Técnicos, Vol. 7, no 60. Buenos Aires: Ministerio de Economía.

Jacome H., L.I. (2004). “The Late 1990s Financial Crisis in Ecuador: Institutional Weaknesses, Fiscal

Rigidities, and Financial Dollarization at Work,” IMF Working Paper WP/04/12 (January).

Kindleberger, C. (1950). The Dollar Shortage. Cambridge, Massachusetts: MIT Press.

Lucas, R. E. (1976). “Econometric Policy Evaluation: A Critique”, pp. 19-46, in Brunner, K. and

Meltzer, A.H. (Eds.) Carnegie-Rochester Conference Series on Public Policy.

Ministerio de Economía (1993). Informe Económico. Primer Trimestre de 1993. Buenos Aires:

Ministerio de Economía.

20
Mundlak, Y., Cavallo, D. and Domenech, R. (1986). Agriculture and Economic Growth in Argentina,

1913-1984. Washington, D.C.: International Food Policy Research Institute.

Ocampo, E. (2023). “Dollarization as an Effective Commitment Device: The Case of Argentina”,

CEMA Serie Documentos de Trabajo, N. 848 (April).

Ocampo, E. and Cachanosky, N. (2022). Dolarización: Una solución para la Argentina. Buenos Aires:

Editorial Claridad.

Ocampo, E. and Cachanosky, N. (2023). Dolarización: Una solución para la Argentina. Available

online at https://dolarizacionargentina.substack.com

Sargent, T. J. (1982). “The End of Four Big Inflations,” pp. 41–98, in Hall, R.E. (Ed.) Inflation: Causes

and Effects. Chicago: University of Chicago Press/NBER.

Solt, F. (2020). “Measuring Income Inequality Across Countries and Over Time: The Standardized

World Income Inequality Database,” Social Science Quarterly 101(3):1183-1199. SWIID Version 9.3,

June 2022.

Vera L., W. (2007). “Medición del Circulante en Dolarización: Ecuador 2000-2007,” Cuestiones

Económicas, Vol.23, No.2:2-3.

21
6. Appendix: Macroeconomic Data for Ecuador

Table 1. CMBW model prediction versus Ecuador’s dollarization

Variable CMBW model prediction Dollarized Ecuador

Output Decreases Increased

Money balances Increase rapidly Increased gradually

Real Wages Decreases Increased

Non-tradable output Decreases Increased sharply

Trade balance Increases Decreased

Real exchange rate Immediate drop Appreciated rapidly

Figure 1. Monthly Index of economic activity

22
Figure 2. Tradable and non-Tradable output

Figure 3. Money balances (% of GDP)

23
Figure 4. Current account (millions of current US$), 1990 - 2005

Figure 5. Real exchange rate

24
 Figure 6. Real wages

Figure 7. GINI coefficient

Source: The source for Figure 7 is Solt (2020). For the rest is Banco Central del Ecuador (BCE

25