Comments on Chapter 3

Adrián Armas, Eduardo Levy Yeyati, and Alain Ize
Published Date:
July 2006
  • ShareShare
Show Summary Details
Roberto Chang

One consequence of the sequence of emerging markets crises that started with the Mexican devaluation of 1994 is a widespread recognition of a need to revise exchange rate theory to take into account the role of financial institutions and frictions. Topics such as financial dollarization, balance sheet effects, contractionary devaluations and the like are no longer exotic side trips but have become central to the formulation of monetary and exchange rate policy. There is a lot of related research going on.

But while it is clear that we have travelled a long way, I am not sure where we are getting to. Our models have improved by allowing for currency mismatches, endogenous risk premia and so on, but this progress has been obtained only under ad hoc assumptions on the currency denomination of assets and liabilities. As a consequence, any implications, especially for policy, are suspect and indeed may not be robust, as stressed by Olivier Jeanne, Marcos Chamón and others.

The obvious response is that research efforts should be aimed at understanding the basic determinants of financial phenomena such as dollarization, from first principles if possible. The chapters in this section are good contributions to this agenda.

As requested by the editors, my comments focus on Chapter 3 by Alain Ize and Eduardo Levy Yeyati, which surveys the state of the art on the growing literature on financial dollarization, and attempts to identify useful directions for further research. The chapter reviews some theories of financial dollarization, related empirical work, why financial dollarization is a ‘problem’, and policy issues related to de-dollarizing.

The chapter starts with the crucial question of how to explain observed degrees of financial dollarization, and basically finds three kinds of answers in the literature:

Risk sharing: If financial markets are incomplete, financial instruments denominated in different currencies have implications for allocating risks to borrowers and lenders. Then the degree of financial dollarization is typically determined by a CAPM model or similar.

Market failures: Financial dollarization may be seen as an endogenous response of an economy to the time inconsistency of government policy, asymmetric information, or imperfect bankruptcy procedures.

Relative country size: Financial dollarization may express an economy’s attempt to exploit economies of scale in financial intermediation.

Distinguishing between these hypotheses is clearly important, in particular, for the formulation of appropriate policy. For instance, if financial dollarization reflects risk sharing there is little presumption that government intervention can improve upon the market outcome. On the other hand, public policy may be able to help if dollarization reflects an underlying market failure.

Accordingly, Chapter 3 turns to empirical studies of financial dollarization. Unfortunately, the authors find that existing work is sketchy at best, and more of a collection of interesting stylized facts than a body of systematic evidence. This state of affairs is due, to a large extent, to the lack of adequate data for exploring new models of financial dollarization. The insufficiency of the data may itself be endogenous, as only recently we have started looking for empirical measures of currency mismatches and the like.

A more significant concern, I believe, is that the interpretation of even the few available studies is unclear. Significant identification problems remain and prevent us from drawing conclusions for discriminating between the alternative theoretical views of financial dollarization. This turns out to be crucial for policy analysis.

Let me illustrate this point with a specific example. Consider the model of Chang and Velasco (2006), which is cited in Chapter 3. That model features workers that are risk averse and borrow from the world market in dollars, and firms that borrow from workers to finance production of a home good. Because markets are incomplete, workers and firms write debt contracts in pesos and dollars to share risk, and the dollarization ratio is determined as in the consumption CAPM model.

The twist is that, in the model, nominal wages are rigid, and the central bank chooses a policy regime (fixed or flexible exchange rates) after private contracts are written. Therefore, the central bank decision depends, among other things, on the degree of financial dollarization. But debt and wage contracts, in turn, will reflect market expectations of future price and exchange rate variability.

The key result is that there may be two self-fulfilling equilibria. There is always an equilibrium with flexible exchange rates. But there may be also an equilibrium with fixed exchange rates, which displays substantial currency mismatches. The fixed rates equilibrium is Pareto dominated, and hence phenomena associated with it (currency mismatches, financial dollarization and fear of floating) reflect a coordination failure.

What are the lessons for our current discussion? In the model just described, portfolio choice and dollarization ratios satisfy a strict version of the consumption CAPM. Hence, in a world of many economies similar to this one, an empirical researcher would indeed find very good evidence that dollarization is explained by minimum variance portfolios and other variables suggested by the CAPM approach. But also, and by construction, the model exhibits multiple equilibria and implies that government policy may be Pareto improving.

In terms of the categories mentioned earlier, regressions of financial dollarization on minimum variance portfolios may give us little or no information about the relative merits of the risk sharing versus market failure hypotheses. Of course, here the reason is that those two hypotheses are not inconsistent with each other: CAPM-like theories have implications for individual portfolios, while market failures may be systemic.

Since existing evidence fails to validate a particular theory of financial dollarization, it also fails to identify welfare-improving policy. For example, in the model of Chang and Velasco (2006), multiple equilibria exist because the central bank has a time inconsistency problem. How the central bank can improve its ability to precommit to solve time inconsistency is unclear. On the other hand, Chang and Velasco show that the bad equilibrium can be eliminated by appropriate direct regulation of financial dollarization. This is noteworthy, as the discussion by Ize and Levy Yeyati would imply that direct intervention on portfolios is the wrong choice if financial dollarization is well explained by CAPM considerations.

My conclusion is not that we need to implement direct controls on financial portfolios but that, more generally, we must deal carefully with serious identification issues. CAPM-motivated regressions of financial dollarization are useful to uncover some stylized facts, but we have to dig more deeply before obtaining conclusions useful for policy.

Let me finish with two brief remarks. We seem far from having a widely accepted theory of financial dollarization, and there is no obvious presumption that government policy should be actively aimed at ‘de-dollarizing’ our economies. That financial dollarization has been so persistent can be used to argue, in fact, that it has not been such a crucial restriction on macroeconomic policy: Peru, having brought annual inflation from 7,000 per cent to almost zero is the obvious example.

This being said, a tight case for de-dollarization remains but has not been mentioned: the ability of the central bank to act as a lender of last resort. One key difference between an economy that has financial instruments primarily denominated in domestic currency and a highly dollarized economy is that the central bank can always, if needed, take over and honour financial obligations in the first economy, because the central bank prints domestic currency at negligible cost. In contrast, the central bank cannot honour obligations in foreign currency unless it maintains a large and costly stock of foreign reserves. This argument was explored in some detail in Chang and Velasco (2000), but perhaps should be added to the list of considerations for further study.


    Chang, R. and A.Velasco (2000) ‘Financial Fragility and the Exchange Rate Regime’, Journal of Economic Theory, Vol. 92, pp. 134 (May).

    Chang, R. and A.Velasco (2006) ‘Monetary Policy and the Currency Denomination of Debt: A Tale of Two Equilibria’, Journal of International Economics, forthcoming.

    Other Resources Citing This Publication