Dollarization and multiple equilibria

Bank runs are a risk inherent to any monetary regime with a fractional reserve banking system. Official dollarization does not necessarily increase their likelihood. In fact, it can reduce it.

for the original version in Spanish click here

SUMMARY

One of the most frequently repeated objections to official dollarization is that it could provoke a bank run. At first glance, the argument seems compelling: replace the peso with the dollar and eliminate the lender of last resort, and the banking system will be left defenseless in the event of depositors panic. But this logic, grounded in a simplistic reading of the Diamond-Dybvig model, does not withstand historical scrutiny.

It is true that fractional reserve banking systems are inherently exposed to runs, since they operate with a structural mismatch between illiquid assets and liquid liabilities. And yes, under such conditions, expectations can become self-fulfilling. However, most bank runs throughout history have not been triggered by the viralization of imaginary fears, but by the existence of real vulnerabilities in banks’ balance sheets.

Argentina’s 2001 crisis left a deep scar. Still, the fear that dollarization will make such scenario more likely is overblown. The banking systems in Panama, Ecuador, and El Salvador have operated without a lender of last resort for more than two decades without a banking crisis despite several shocks. Financial stability under a dollarized regime is entirely feasible—provided the banking system is well-capitalized, well-supervised, and supported by responsible fiscal policy.

The risk of a run is not determined by whether a government chooses to issue its own currency or adopt the dollar. If banks are incentivized to originate toxic assets (gaucho banking) and/or fiscal disciplined is relaxed, they will be vulnerable with or without the peso. But if dollarization is implemented with strongly capitalized banks that are properly supervised and a primary fiscal balance is maintained, the risk of a run not only does not increase—it may actually fall. Bank runs aren’t caused by the medium of account used by the banking system. They’re caused by banker’s poor credit analysis and/or bad economic policies.

One of the most common objections to official dollarization is the possibility that it could trigger a devastating bank run. Although I have already explained why this risk is overblown (see here and here), given the persistence with which this objection is raised, some further clarifications are warranted.

For the purposes of the analysis that follows, I understand official dollarization as the decision taken by a government to a) grant legal tender status to the dollar, b) fix the amount of pesos in circulation, c) agree to exchange all those pesos at a fixed parity without a deadline, and d) establish the dollar as the sole unit of account for the banking system. Broadly speaking, this is the format adopted by El Salvador in 2001.

This is just one of several possible ways of implementing dollarization. As I have repeatedly pointed out, the regime that exists in Panama since 1904 is different from the one implemented in Ecuador in 2000, which in turn differs from the one in El Salvador since 2001, which is also distinct from the one implemented in Zimbabwe between 2009 and 2015. Any economist who disputes the viability of official dollarization, at a minimum should be able to explain how all these schemes differ and why it is relevant to the analysis. Otherwise, the discussion turns into a theoretical exercise without any practical relevance.

In a previous article, I argued that the viability of an official dollarization, as defined above, depends more on the sustainability of primary fiscal surpluses than on the level of reserves in the central bank (which does not mean that reserves are irrelevant). However, the discussion in Argentina always focuses on the latter. To some extent, this is understandable. The traumatic experience of 2001 left an indelible mark on the minds of depositors, bankers, and economists: the dreaded bank run.

A bank run is a risk that any economy operating with a fractional reserve banking system (explicit or implicit) faces. It will be greater or lesser depending on the institutional framework in which banks operate and the existence of “moral hazard.” Deposit insurance backed by a solvent state can reduce it, but not eliminate it (paradoxically it can also increase it). As Argentina’s 2001 experience showed, if such solvency does not exist or is questionable, the guarantee proves ineffective. On the other hand, as the 2008 crisis also demonstrated, as long as a leveraged, unregulated shadow banking system with strong incentives to originate poor-quality assets exists, the risk is far from trivial, even in developed economies like the United States.

Another important aspect is defining what is a “run.” There are two types of money in the economy: external and internal. The first is issued by the state through its central bank, while the second is created by a fractional reserve banking system, as has existed worldwide for at least two centuries. In most developed economies, internal money is three or four times external money. That is, most money circulating in the economy is "produced" by the private banking system based on liquidity and credit demand and reserve levels set by the central bank.

There is an asymmetry. Internal convertibility is at one-to-one conversion ratio: a peso in the bank is as valuable as a peso in your wallet, without discount—except during a deposit run. But how many pesos in your wallet equal one dollar? It depends. The exchange rate can be set by the central bank or the market.

An exchange rate run arises from doubts about external convertibility at the established parity—the central bank's capacity to convert pesos into dollars. If people perceive the central bank cannot repurchase all pesos at the set parity, either international reserves fall, or the exchange rate spikes, depending on the regime. External convertibility can be limited with exchange controls and/or restrictions on capital movements.

A deposit or banking run arises from doubts about internal convertibility—the ability to convert deposits to cash at a one-to-one ratio (without discount). Internal convertibility can be suspended via bank holidays or freezing deposits.

A crisis of internal convertibility occurs because people perceive deposits as inadequately backed. This may happen for two reasons: 1) banks have too many toxic assets (gaucho banking), or 2) banks are well-capitalized, but if panic spreads, they quickly run out of liquidity.

In the first scenario, banks have lent money to individuals or companies unable to repay principal and interest on time. If non-performing loans reach 10%, banks’ financial position becomes compromised. This situation provides valid reasons to doubt internal convertibility.

The second scenario has a theoretical basis on a famous paper by Diamond and Dybvig proposed that fractional reserve banking systems are always vulnerable to deposit runs due to structural mismatches: liabilities must be liquid, assets are not. If everyone demands liquidity simultaneously, banks cannot satisfy this demand. This implies that economies with fractional reserve banking systems face multiple equilibria: one with deposit runs and one without them. The first equilibrium, described by economists as "sunspot equilibrium," does not depend on fundamentals. Depositors panic not because they've analyzed a bank's balance sheet and found that their non-performing assets are too high, but because they perceive the panic of other depositors. Dybvig explained this behavior in his acceptance speech for the Nobel Prize in Economics:

“There can be an equilibrium in which everyone withdraws their money from the bank, whether they need it or not. We call that equilibrium a run on the bank. The outcome is bad for everyone. If you think everyone is trying to withdraw their money, you realize that if you wait, the bank will run out of money before you arrive. Therefore, you will also try to withdraw all your money.”

In other words, the viralization of depositors’ fears pushes the economy from a (good) equilibrium without a run on deposits to a (bad) equilibrium with a run on deposits and a collapsed banking system.

It sounds plausible, but the reality is more complex. As George Selgin has pointed out, most historical bank runs occurred due to actual financial vulnerabilities rather than spontaneous imaginary panics.

Those who argue that internal convertibility crises (bank runs) are more likely under dollarization than under a peso-based regime are, in reality, suggesting an external convertibility crisis when announcing dollarization. This can only happen if, at the established parity, there is excess demand for dollars (or excess supply of pesos).

If banks are well-capitalized with high-quality assets, there is no inherent reason why internal convertibility crises would be more probable under dollarization. Such a scenario could only arise following an exchange rate run, an external convertibility crisis.

To summarize, an exchange/banking run occurs when people discard external/internal money perceived as inadequately backed. With a 100% backed external monetary regime and 100% reserve banking, neither type could occur. Historically, such a combination has never existed. Even under the gold standard that existed in England from 1821 to 1914, external money was never 100% backed.

With a peso-based system, simultaneous exchange and bank runs are possible. Post-dollarization, by definition, exchange runs cannot occur. If the first system has N equilibria, the second has N-1.

Regardless of what anyone thinks about the theoretical viability of a fractional reserve banking system and/or its stability under a dollarization regime, the evidence shows that the banking systems of Panama, Ecuador, and El Salvador—without a lender of last resort—have been as stable as those of Chile, Peru, and Colombia. Moreover, they have been far more stable than Argentina’s with the peso and have contributed much more significantly to economic development through the expansion of credit to the private sector.

Having declared two sovereign debt defaults (in 2008 and 2020) and faced several internal and external shocks, Ecuador experienced its worst bank run in 2015, when total deposits fell by nearly 10%. This was largely due to the drop in the price of crude (which is Ecuador’s main export) and the policies of Rafael Correa. Notably, net reserves at the central bank turned negative after 2015 but did not provoke another deposit run or a banking crisis. And dollarization certainly did not collapse.

In Argentina, the worst deposit runs occurred in 1995, 2001, and 2019. In the first case, the run was caused by the aftershocks of the so called "Tequila crisis" and fears of devaluation. It was longer and more pronounced for peso deposits than for dollar deposits. In 2001, the run on peso deposits was also more intense and prolonged than on dollar deposits, while in 2019, the opposite occurred, massive dollar withdrawals took place after the PASO. In this case it was due to fears that the government that would like win the presidential election would confiscate or freeze those deposits.

This evidence by no means would allow anyone to conclude that a bank run is inevitable under dollarization. As long as banks are well capitalized and there are enough dollars at the central bank to buy back all the pesos in circulation at a market exchange rate, there is no reason to assume a run would occur.

However, according to some analysts, if the government decided to dollarize, people would rush to the banks to withdraw their pesos, would convert them into dollars and then stash them under their mattress or transfer them abroad, provoking the collapse of the financial system and triggering a deep economic crisis. Undoubtedly, the probability of such scenario is not zero. But it also not zero under any other monetary regime, even with capital and FX controls. Argentina would face the same risk under any regime with a convertible peso, free capital movements and a bimonetary banking system (as it exists today).

To assess the potential risk of a bank run under dollarization (or any regime with a convertible peso), we need to consider two key factors.

First, in Argentina today M3 is mostly transactional. Assuming that all bank deposits would “flee” overnight is a very strong assumption. Where would they go? The economy is a circular flow. Economic agents must pay taxes, salaries, and expenses on a daily basis through bank transfers. Even if they decided to transfer all their funds abroad, they would need to bring them back into the economy to continue with their normal economic activity. Moreover, given the level of insecurity, operating with cash seems far riskier than keeping liquidity in the bank.

In other words, if deposits “leave” today, they will have to return tomorrow or the day after. Furthermore, we must not forget that the key role played by interest rates in bringing the money market into equilibrium. There is a rate at which flows would stabilize. And that level is not infinite. This what the experience of dollarized countries tells us.

Second, the flip side of peso demonetization is de facto dollarization—i.e., the spontaneous adoption of the dollar as a currency by the private sector (against the wishes and policies of the government). According to the latest official statistics, the private sector’s net liquid international position as of December 31, 2024, amounted to US$286.6 billion (40% of GDP at the official dollar rate and 48% at the blue dollar rate). This figure does not include undeclared liquidity, which according to some sources could be as high as US$100 billion.

Why and how would the Argentine private sector accumulate more dollars than it already has? The obvious answer is that it would do so if it could acquire them cheaply—that is, if the government offered to sell them below the market exchange rate. Under any scenario, acquiring dollars would require liquidating meager peso savings (but not transactional liquidity) and/or obtaining peso financing at low interest rates (which does not exist).

When considering the likelihood of a deposit run under dollarization, it is worth noting that as of January 2025, the private sector held dollar deposits in the banking system in an amount close to US$31 billion. These deposits were backed by nearly US$6.9 billion in cash in banks’ vaults, US$14 billion in short-term loans to the private sector, and US$13 billion in reserves deposited (“encajes”) at the BCRA. Given that net international reserves were negative as of such date, we can conclude that these deposits were backed by less than 20% in liquid assets. Yet, they didn’t “go” anywhere. In fact, dollar deposits have almost doubled since December 2023.

As long as the government can credibly maintain a primary fiscal surplus and the banking system has no incentives to originate poor-quality credit assets (aka gaucho banking), the risk of a deposit run will be relatively low. In fact, given the relatively large share of government securities in bank’s balance sheets, the quality of the assets that back deposits would continue to increase.

To conclude, a properly designed and implemented official dollarization would not only not increase the risk of a bank run, but also could actually reduce if banks remain properly supervised and capitalized and the government maintains fiscal discipline. For starters, the devaluation premium would disappear and interest rates would fall, which would improve banks' asset quality. Increased confidence would contribute to increased monetization, which, in turn, would increase profitability and strengthen the soundness of the banking system.