- Comments and Discussion
Since the 1980s, the world economy has witnessed an intriguing development: despite the proliferation of more flexible exchange rate regimes, ratios of international reserves to GDP have increased substantially, and practically all of that increase has taken place in developing countries, mostly in East Asia. This puzzling phenomenon has stirred a lively debate among economists and financial observers, and several interpretations have been offered. These focus on the observation that the deeper financial integration of developing countries into the global economy has increased their exposure to volatile short-term flows of capital (dubbed "hot money"), which are subject to costly and frequent sudden stops and reversals.1 In these circumstances, hoarding international reserves can be viewed as a precautionary adjustment, reflecting the desire for self-insurance against exposure to future sudden stops, currency crises, and capital flight.2
In this paper, Olivier Jeanne appraises the explanatory power of self-insurance models of international reserves. He provides a careful formulation of a utility-based welfare analysis of optimal hoarding of international reserves, and a calibration that allows one to evaluate the degree to which recent hoarding trends are consistent with the model's predictions. The framework is based on an elegant model, providing strong predictions about optimal usable reserves in the context of self-insurance against sudden stops and currency crises. Applying the model to the data produces some intriguing results: for the typical emerging market country, the model [End Page 56] can plausibly explain a reserves-GDP ratio on the order of 10 percent, close to the long-run historical average, and it can justify even higher levels if one assumes that reserves have a significant role in terms of crisis prevention. The levels of reserves recently observed in many countries, in particular in Latin America, are within the range of the model's predictions. For emerging market countries in the aggregate, however, the insurance model fails to account for the recent reserves buildup, because the risk of a capital account crisis in the Asian countries where most of the buildup has taken place seems much too small to justify such levels of self-insurance.
Jeanne's paper is an important and timely contribution to the debate about the relative costs and benefits of hoarding reserves. Yet his results raise the question of what factors might account for the sizable gaps between the model's predictions and the actual hoarding of reserves observed during recent years. After briefly summarizing his methodology and key results, I will discuss several alternative interpretations of reserves hoarding during the 2000s that go beyond the self-insurance paradigm, and I will offer several extensions that may improve the explanatory power of Jeanne's model.
The Model and Its Calibration.
Jeanne's framework explains hoarding of international reserves in the context of a small open economy populated by a representative consumer and exposed to crises triggered by a loss of access to external credit, possibly associated with a fall in output. The model has three periods. The focus of the analysis is on the optimal hoarding of net reserves in period 0, when the representative consumer anticipates the possibility of a sudden stop and a costly foreign currency crisis in period 1. The consumer's welfare is the sum of the expected utility of consumption in period 1 plus discounted expected (terminal) wealth in period 2.
Within this framework, Jeanne provides a carefully worked-out model in which international reserves affect welfare in two distinct ways. The first is through crisis prevention: an increase in reserves may reduce the probability of a crisis and thus directly reduce the expected output cost. The second is through crisis mitigation: reserves may reduce the cost of smoothing consumption during a crisis. In this model optimal international reserves are shown to be determined by minimizing a loss function, where the total loss equals the sum of the opportunity cost of reserves and the expected welfare cost of a crisis.
The model is calibrated in two steps. First, using multivariate probit regressions on data for a group of emerging market countries from 1980 to [End Page 57] 2000, Jeanne estimates...