Dollarization and bitcoinization in El Salvador, why one mandate was enforced and the other wasn’t

Introduction

El Salvador is the only country that has imposed two distinct legal tenders upon its own economy in the course of a single generation. The Ley de Integración Monetaria withdrew the colón from circulation on the first of January of 2001 and installed the dollar of the United States in its place [16]. Twenty years later the Ley Bitcoin granted the status of legal tender to a decentralized digital asset, with a mandate of obligatory acceptance that no sovereign State had established before [15]. The same legislative body approved the two measures, through comparable procedures, and against the opinion of the majority of the population. Neither produced the benefits that its promoters announced.

Their institutional results, nevertheless, were the opposite. Dollarization remains in force twenty five years later, and there exists no precedent of a complete dedollarization in any country that adopted the measure officially [14]. The Ley Bitcoin lost the status of legal tender on the twenty ninth of January of 2025, through Decreto Legislativo № 199, that is to say, in less than four years, and by decision of the same legislative majority that had approved it [21].

The present article asks why, and the question requires a distinction that the existing literature has not made. The institutionalization of a monetary measure is not the same thing as its success. The studies that have compared the two Salvadoran impositions conclude, with reason, that neither transformed the economy of the country [9]. That conclusion leaves the problem intact, given that the two measures failed in matters of their announced objectives and only one of them was revoked. If success and institutional permanence are treated as equivalent, the divergence disappears and there is nothing left to explain.

The response that I sustain does not attribute the divergence to the volatility of Bitcoin considered in isolation, nor to the structural weaknesses of the Salvadoran economy, nor to the authoritarian character of the government that promoted the measure, although each of those factors intervenes. I identify three conditions that operate jointly, whether the mandate ratifies a unit of account that is already operating or attempts the creation of one by decree, the quality of the imposed asset in that function, and the enforceability of the obligation given the population upon which it is placed. The third is the principal contribution of the present research and has received no attention in the literature. The Ley de Integración Monetaria placed its mandate upon a dozen licensed banks. Article 7 of the Ley Bitcoin placed it upon every commercial economic agent, in an economy where informal employment reaches 69 per cent [13].

The scientific novelty of the present research consists, therefore, in establishing, through the comparison of two sovereign monetary impositions carried out in the same country, that legal coercion constitutes neither a sufficient nor a necessary condition for currency substitution, given that its effect finds itself conditioned by the preexistence of a network of use, by the relative quality of the asset as unit of account, and by the enforceability of the mandate over the target population. That finding exceeds the case examined, at a moment in which more than one hundred central banks develop projects of retail digital currency, given that the technical superiority of these instruments over Bitcoin does not modify the condition of enforceability.

Section 1 constructs the theoretical framework and identifies the gap, section 2 exposes the method and its limitations, section 3 develops the comparison and the three conditions, and section 4 confronts the rival explanations and derives the implications.

1. Theoretical Framework

1.1. The Network Foundation of Currency Substitution

The literature on currency substitution was constituted, in its greater part, during the nineteen nineties, around the voluntary abandonment of the national currency by the economic agents of a country in response to inflation. Calvo and Végh showed that the substitution operates through the successive functions of money, given that the agents abandon the national currency in first place as store of value, afterwards as unit of account, and only at the end as medium of exchange [10]. Guseinov formulates a convergent distinction useful here, given that what is substituted in an official procedure of dollarization is not necessarily what has already been substituted in effective economic practice [2].

The mechanism that governs the process was formulated by Guidotti and Rodríguez in a manner that has become classical, the substitution operates as a Gresham's law in reverse, that is to say, the good money displaces the bad one, given that the agents can choose [12]. The formulation carries an implication that the literature on cryptocurrencies has not incorporated, given that the theory predicts substitution only when the entering asset dominates the displaced one in the relevant monetary function.

What this literature describes in aggregate terms received a formal foundation in the models that treat the use of a currency as a coordination problem. Matsuyama, Kiyotaki and Matsui demonstrated that the international use of a currency admits multiple equilibria, given that the utility of holding it depends on the proportion of agents that already accept it, and that the resulting configuration is determined by the history of the interactions that preceded it and not by the intrinsic properties of the currency [17].

The consequence that I extract from this body of theory structures the first condition of my comparison. A monetary network is not created by the announcement of its existence, and a mandate that arrives after the coordination has occurred performs a function completely distinct from the one performed by a mandate that attempts to initiate it.

1.2. Hysteresis, Asset Quality, and the Limits of the Framework

The second element concerns what occurs after the substitution, and here the contribution of Uribe results determinant. Uribe demonstrated that the process presents hysteresis, that is to say, that once the migration towards the foreign currency has been produced the agents do not return to the national one even if the inflation that motivated it disappears, given that the costs of learning and of network coordination cannot be recovered [26]. Levy Yeyati confirmed in addition that there exist no precedents of a complete dedollarization in a country that adopted the measure officially [14].

The mechanism establishes an asymmetry decisive for my comparison. The hysteresis protects a substitution that has occurred, and it protects nothing when the substitution has not occurred. A mandate that never generated a network of use does not accumulate the sunk costs that would make its reversal expensive, and results therefore revocable at a very reduced political price. On the quality of the asset capable of generating such a network, Belousov defines Bitcoin as an artificial or surrogate form of money which interacts with the traditional monetary system without managing to substitute it in the function of measure of value [1].

The whole of this theory shares, however, a presupposition rarely made explicit, namely that the substitution is a decision of the economic agents, and it ceases to be adequate when the substitution is not selected but ordered. The literature has treated the official dollarization as the formalization of a substitution that had already occurred, and in that configuration the presupposition of voluntariness is conserved because the mandate arrives after the decisions. But the Salvadoran case of 2001 does not adjust itself to it, and Rudenkov has documented this point in his comparison with Ecuador, showing that El Salvador did not face a financial crisis before the adoption of the dollar [4]. The case of 2021 adjusts itself even less. In both, the norm precedes the behaviour it seeks to produce.

The literature on full dollarization does not offer conditions that permit the anticipation of whether an imposed mandate will consolidate itself or whether it will remain as a norm without effect [14]. That gap constitutes the scientific problem that the present article addresses.

1.3. Enforcement Capacity as a Monetary Variable

There exists a third element the framework requires and has not incorporated. A monetary mandate is a norm, and a norm produces effects only if the authority that dictates it can verify its observance and penalize its breach. When the addressee is a supervised banking institution, verification and penalty result trivial. When the addressee is a commercial establishment not registered before any authority, the verification is impossible and the penalty does not exist.

The monetary literature has treated informality with growing sophistication, but in matters of a question distinct from mine. Ohnsorge, Okawa and Yu establish that a broad informal sector weakens the effectiveness of public policies and the capacity of the government to generate fiscal revenues [20], and the literature on monetary transmission habitually ignores that a large part of the economy operates outside of the formal system. All of that literature refers, nevertheless, to the transmission of an incentive, that is to say, to the capacity of the authority to modify the behaviour of the agents through relative prices. What the Salvadoran case obligates me to consider is a prior question, whether that population can be reached by the mandates of the authority at all. And in a dollarized economy the question is not merely analytical, given that El Salvador disposes neither of an interest rate of its own nor of an exchange rate, so that the two channels which the literature analyses do not exist here. The legal mandate is the only monetary instrument that remains available to the State, and precisely for this reason the condition of enforceability converts itself here into the determinant one.

The literature has approached enforcement from the side of the causes and not from the side of the consequences. Robinson, Vostroknutov and Vostroknutova model the decision to abandon the formal institutions as a calculation that weighs their burden against the enforcement capability of the State [22]. There the enforcement capacity is a parameter within the decision of an agent who observes the norm, evaluates the risk and opts for the exit. The situation that I examine is distinct, given that a considerable part of the population upon which Article 7 fell evaluated no risk, because the authority that dictated the norm could not identify them.

The formulation of the condition results, in my analysis, simple. The enforceability of a monetary mandate depends on the regulatory visibility of the population upon which the obligation is placed, and that visibility varies enormously between an economy with a predominant formal sector and one in which the majority of the economic units operate outside of every registry [20]. Where the visibility is complete, the mandate can be applied and its observance verified. Where it is minimal, the mandate produces what the sociology of institutions denominates a decoupling between the formal structure and the effective practice [19]. I do not import here a foreign framework into monetary analysis, I make explicit a condition of application that monetary analysis has been assuming as satisfied.

2. Method

The present article employs a within case comparison of diachronic character, that is to say, between two events that occurred in the same unit of analysis at distinct moments in time. The two impositions occurred in the same country, under the same constitutional order, with the same legislative procedure, and upon populations comparable in their socioeconomic composition, so that the factors which ordinarily contaminate a comparison between countries find themselves controlled by construction. What varies is not the context. What varies is the design of the mandate and its outcome. The technique is the one denominated structured and focused comparison, which interrogates both cases through the same questions, formulated in advance. The dimensions appear in Table 1 and were derived from the theory of currency substitution and from the literature on institutional enforcement, not from the observation of the cases, a precision necessary in order to avoid the reproach that they were selected because they produce the desired result.

The dependent variable requires an explicit definition, given that a good part of the confusion of the existing literature originates in its absence. The variable that I seek to explain is not the economic success of the measure, nor the fulfilment of its promoters' promises. It is the institutional persistence of the mandate, that is to say, whether the norm remained part of the monetary order of the country or whether it was eliminated from it. The revocation of the legal tender status of Bitcoin through Decreto Legislativo № 199 operates here as an observed condition of reversal [21]. It is not an experiment and I do not present it as such, given that nobody assigned the treatment and there exists no control group. What it provides is sufficient for a comparison of this nature, an unambiguous observation of the dependent variable in one of the two cases, produced by the same political actor that had approved the original mandate, which permits the discarding of explanations centred on the alternation of governments.

The sources employed are of three types, the primary legal instruments consulted in the Diario Oficial [15, 16, 21], the official statistical series of the Banco Central de Reserva de El Salvador and of the International Labour Organization [7, 8, 13], and the empirical literature based on representative surveys and field work, which provides the data on effective adoption that the official statistics do not register. Where the secondary sources present discrepancies I have gone to the national primary series.

Two limitations must be declared. In first place, the design operates with two observations and permits no statistical inference, given that its objective consists in the identification of conditions and not in the estimation of effects. In second place, twenty years separate the two events, and during that interval the digital infrastructure, the political regime and the international framework against money laundering experienced substantial changes. I treat this second limitation in section 3, given that the confounding factors of that interval operated in favour of the Ley Bitcoin and not against it, which converts the Salvadoran case into a hard case for my own argument.

3. The Two Impositions Compared

The two measures share characteristics which make the comparison productive. Both were approved by the Asamblea Legislativa in a matter of days and without meaningful parliamentary debate, both were rejected by the majority of the population, both were presented as instruments of long term development, and both failed in the delivery of the benefits announced [9, 18]. What the literature has not asked is why, sharing all of those characteristics, they produced opposite institutional results.

The question comes to be possible only if the institutionalization of a measure is separated from its success. Dollarization did not produce the growth, nor the investment, nor the credit conditions promised in 2001, given that the public debt continued growing until 77.9 per cent of gross domestic product (GDP) in 2019, against an initial cost of stocking of 2.1 per cent of GDP and a permanent loss of seigniorage [11]. It became institutionalized nevertheless. The question of the present section, therefore, is not whether these were good policies. It is why one of them converted itself into a part of the monetary order of the country and the other did not. Three conditions explain the divergence, and Table 1 summarizes the comparison in matters of each one.

Table 1. Two monetary mandates in El Salvador, conditions and outcomes

3.1. The Ratification of an Existing Unit of Account

The condition of ratification requires a correction of the common description of the Salvadoran case. It is frequently assumed that the dollarization of 2001 came to formalize a currency the population already held in a generalized manner. The data do not sustain that assumption. The deposits in foreign currency of the Salvadoran banking system amounted to 7.7 per cent before the reform, against a Central American average of 34.1 per cent [23]. El Salvador constituted, in matters of financial dollarization, one of the least dollarized economies of Central America.

What did find itself already in place was something distinct and, for the purposes of the present argument, more relevant. Between 1989 and 2000 the country operated a currency board in everything except in the name. The exchange rate converged around 8.75 colones per dollar in 1990 and was maintained until the very moment of dollarization, every colón issued had to count with a full backing in dollar reserves, and the design of that arrangement was prepared with the advice of the architect of the currency board of Hong Kong [11]. Inflation fell from 23.5 per cent in 1989 to below one per cent in 1999 under that regime, and de Rosa states in a direct form that the arrangement generated the conditions that later on made dollarization possible [11].

The distinction results essential. What existed before 2001 was not a population that held dollars, it was a price system already denominated against the dollar through an administered anchor of eleven years. The Salvadoran producers fixed their prices and calculated their costs against a rate that did not move. The Ley de Integración Monetaria did not ask of the economic agents the adoption of a new unit of account. It removed the intermediary between the unit of account they were already using and the currency that stood behind it. The Ley Bitcoin asked of an entire economy the incorporation into its price system of an asset that could move by double digits in a single week, directed at a population with no prior contact with it and no reason to want it. No jurisdiction had granted such status before, given that the review of foreign experience in matters of legal regulation of cryptocurrency shows models ranging from prohibition to regulated tolerance [5].

3.2. The Quality of the Imposed Asset as Unit of Account

The theory of currency substitution establishes a condition that the literature on cryptocurrency adoption has systematically ignored, substitution occurs when the entering currency dominates the displaced one in the function of unit of account, and not when the State declares that it should occur [10, 12]. The formulation allows the evaluation of the two mandates against the same criterion.

In the case of 2001 the condition was satisfied without difficulty, given that the dollar offered a stability the Banco Central de Reserva could not guarantee by its own means, and offered it in the three functions of money at the same time [11]. The Salvadoran economic agents did not have to be convinced. They had been operating against the dollar for eleven years.

In the case of 2021 the condition was not satisfied in any of the three functions. Bitcoin presents severe limitations as medium of exchange, given the costs and the confirmation times of the blockchain, and the Lightning Network is a partial solution not exempt from its own limitations [9]. As unit of account it results directly unusable, because a merchant cannot denominate prices in a magnitude that moves by double digits within a single week. And as store of value it retains a certain functionality, but a volatile one, and it retains it against the dollar, that is to say against the currency that the Ley Bitcoin was supposed to complement [1]. The Russian literature on the expansion of dollar denominated stablecoins points in the same direction, given that the digital instruments which do reach populations outside the formal banking system extend the reach of the dollar unit of account instead of displacing it [3]. The asset that the law imposed was inferior, in the function that matters for a price system, to the asset that already occupied it.

The empirical evidence confirms what the theory anticipated. Alvarez, Argente and Van Patten documented that 60.66 per cent of the users of Chivo abandoned the application after redeeming the initial incentive of thirty dollars, and that only 11.4 per cent of the firms had ever completed a sale in Bitcoin [6]. The survey of the Universidad Centroamericana José Simeón Cañas published in January of 2024 established that 88 per cent of the persons consulted had not carried out any transaction in Bitcoin during the whole of 2023 [25]. And the remittances transferred through digital wallets, the principal promise of the policy, descended from 1.34 per cent of the total in 2023 to 1.13 per cent the following year [18].

Here I arrive at the point that connects this condition with the outcome. The mechanism of hysteresis explains why the dollarization of 2001 results irreversible, and it explains it better than the arguments of political economy usually invoked [26]. But the mechanism never activated itself in the case of Bitcoin, for a simple reason. There was no substitution to sustain. Nobody had paid the costs of learning, no network of commercial acceptance had formed, and there existed therefore no accumulated investment that the reversal of 2025 could destroy. The Ley Bitcoin was revocable because it had never taken hold.

3.3. The Enforceability of the Mandate

The third condition is the one that I consider decisive, and it is also the one that the literature has left untouched. The two laws imposed an obligation, but they did not impose it upon the same subjects, and this difference determined whether the obligation could be enforced or whether it could not.

The Ley de Integración Monetaria placed its mandate upon the banking system. The process was executed with a velocity that results notable, given that thirty nine days passed between the presidential announcement at the end of 2000 and the formal extinction of the colón. The law fixed the exchange rate in an irrevocable form and obligated the Banco Central de Reserva and the institutions of the banking system to convert into dollars the totality of the accounts and of the physical cash they held [16, 24]. These were licensed entities under permanent supervision and exposed to the withdrawal of their operating licence in case of noncompliance. The Salvadoran State did not need to persuade the population of anything. It needed the compliance of approximately a dozen institutions that it could observe and sanction. The conversion was executed and completed, and the population did not have to make any decision, and in fact did not make it, given that at the moment of the adoption only two per cent declared itself in favour [24].

Article 7 of the Ley Bitcoin placed its mandate upon a completely distinct population. The text established that every economic agent had to accept Bitcoin as a form of payment when it was offered by the person acquiring a good or a service [15]. In an economy where informal employment reached 69 per cent [13] and only 28 per cent of the population holds a savings account in a formal financial institution [8], that mandate was directed at hundreds of thousands of economic units the State cannot observe, identify or sanction. The obligation existed in the Diario Oficial. It did not exist in any place where it could have been applied.

The Salvadoran Government understood this before the law entered into force. Immediately after the text became public, government officials declared that there would be no obligation to accept Bitcoin, invoking Article 12 of the law itself, which exempts those who in an evident and manifest form do not have access to the technologies that permit the execution of Bitcoin transactions [9, 15]. The press documented thirteen distinct moments of contradiction among officials on whether the use was obligatory or voluntary. Boos and Grigera characterize this configuration as a soft adoption and interpret it as a deliberate strategy destined to reduce the costs and the risks of the experiment [9].

My reading is a distinct one, and the difference between the two can be evaluated. If the nonenforcement had been a strategic choice, the Government would have retained the option of enforcing the mandate when it resulted convenient, and it never exercised that option during three years and a half. If it was a structural incapacity, what was observed is exactly what one would expect, an obligation announced, immediately qualified, never applied, and finally eliminated when its elimination acquires an external value in a negotiation. The comparison with 2001 discriminates between the two, because in 2001 the same State, with less technical capacity and less legislative control, did enforce a monetary mandate, and it enforced it because that mandate fell upon a population it could reach.

The interval of twenty years that separates the two impositions deserves a final observation, given that it could be invoked as a confounding factor. The infrastructure of 2021 was incomparably superior in telecommunications, in digital payment and in identification of persons, and the government of 2021 disposed of a legislative control that the government of 2001 did not possess. All of those differences operated in favour of the Ley Bitcoin. The mandate that failed is the one that counted with the better conditions of implementation, which converts the Salvadoran case into a hard case for my argument.

4. Rival Explanations and Theoretical Implications

4.1. The Rival Explanations

Before deriving implications that exceed the case of El Salvador I must confront the explanations that the literature has already proposed, given that if any of them accounted for the divergence my three conditions would result superfluous. The most developed alternative is the one Meijering and Rosales have formulated. According to these authors the Ley Bitcoin should not be understood as a policy of financial inclusion at all, given that it constituted a public relations instrument destined to capture the support of like minded constituencies abroad, and at the same time a mechanism for the channelling of public funds towards a circle of businessmen close to the President, through a company created with state resources which operates protected by the reserve of a private entity [18].

I do not dispute this account. It results convincing and compatible with my argument, given that the two explanations respond to distinct questions. Meijering and Rosales explain why the law was approved. The present article asks why it did not sustain itself. And here the asymmetry becomes visible. If the principal function of the Ley Bitcoin consisted in the extraction of public funds and in the projection of an international image, the derogation of January 2025 would have terminated both, and it was nevertheless the government itself, with fifty four of sixty seats in the Asamblea Legislativa, that proposed and approved it. An explanation centred on the benefits the law generated for the incumbent does not account for the decision of the incumbent to eliminate it. An explanation centred on the fact that the mandate was never operative does account for it, because what Decreto Legislativo № 199 eliminated was a legal fiction, and its elimination could therefore be exchanged for a financing agreement with the International Monetary Fund at a very reduced political cost [9, 21].

The second rival explanation is the one of Boos and Grigera, who sustain that the prior dollarization converted the adoption of Bitcoin into a low stakes gamble, given that El Salvador had already lost its monetary sovereignty, its lender of last resort and its seigniorage revenues [9]. The reasoning results correct in matters of costs. But an explanation of why a measure was cheap is not an explanation of why it did not persist. Cheap policies persist all the time, and the argument, taken to its conclusion, would predict the indefinite maintenance of the Ley Bitcoin precisely because maintaining it cost so little.

A third explanation can be constructed from the framework of endogenous institutions, in which the agents abandon the formal institutions when those institutions lack social appropriateness in their perception [22]. It would sustain that Article 7 was not observed because the mandate lacked legitimacy, and the surveys of the Universidad Centroamericana confirm that the Bitcoin policy was the most unpopular of the government [25]. The comparison with 2001 discards this reading, given that the dollarization counted with the support of two per cent of the population [24], which is to say it was as illegitimate or more, and it became institutionalized in any case. Perceived legitimacy does not discriminate between my two cases. Enforceability does.

4.2. The Conditions Beyond the Salvadoran Case

Once the rival explanations have been situated, the conditions can be stated in general form. A monetary mandate institutionalizes itself when it ratifies a unit of account already operating, when the imposed asset dominates the displaced one in that function, and when the obligation falls upon a population the State can observe and sanction. Otherwise it produces a decoupling between the formal structure and the effective practice [19]. The condition of enforceability offers the greatest transferability and has been the most neglected, in my judgement because the literature on monetary policy tends to assume a State capable of applying its own norms, an assumption reasonable in economies with low informality and not reasonable in the greater part of Latin America, of Africa and of Southeast Asia [20]. The relevant question for a government that considers a monetary mandate is not whether the technology functions nor whether the design is adequate. It is whether the State can reach the persons upon whom the obligation is being placed.

This has direct implications for the debate on retail central bank digital currencies. Those projects are technically superior to Bitcoin in every relevant dimension, given that they do not present volatility, they are issued by the monetary authority, and their settlement times can be designed for retail payments. Nothing of this modifies the condition of enforceability. A mandate of acceptance directed at merchants not registered before any authority is unenforceable for the same reason that Article 7 was, and the structure of informality of the target economy should therefore be treated as a design parameter and not as an implementation obstacle to be resolved later on.

It has been argued, finally, that the adoption of cryptoassets by peripheral States constitutes an attempt to gain leverage before the hegemony of the dollar [9]. The Salvadoran evidence does not sustain that reading, given that during the whole period of validity of the Ley Bitcoin the population continued operating in dollars [6].

Conclusion

The present article asked why two monetary impositions, carried out in the same country by comparable procedures and with equally disappointing results, produced opposite institutional outcomes. The dollarization of 2001 remains in force twenty five years later, while the Ley Bitcoin of 2021 lost its legal tender status after less than four years, by decision of the same legislative majority that had approved it.

The answer revolves around three conditions that operate jointly. The Ley de Integración Monetaria ratified a unit of account that had been functioning for eleven years, imposed an asset that dominated the displaced one in the three functions of money, and placed its obligation upon a dozen licensed institutions. The Ley Bitcoin did the opposite in each of the three.

The third condition constitutes, in my judgement, the principal contribution of the present research. The enforceability of a monetary mandate does not depend on the political will of the government that dictates it, nor on the technical quality of the imposed asset. It depends on whether the population upon which the obligation falls finds itself within the reach of observation and sanction of the State. What Decreto Legislativo № 199 eliminated in January of 2025 was not an operative monetary regime but a legal fiction, and its elimination could therefore be exchanged for an agreement with the International Monetary Fund at a very reduced political cost.

The present research leaves open two lines, the extension of the analysis to other cases of monetary imposition in economies with high informality, and the relation between the informal sector and monetary policy in general, which the literature has treated in matters of the transmission of interest rates and which also operates in matters of the very capacity of the State to impose the instrument in which its economy transacts.