Optimal Holdings of International Reserves: Self-Insurance against Sudden Stop

This paper addresses the issue of the optimal stock of international reserves in terms of a statistical model in which reserves affect both the probability of a Sudden Stop-as well as associated output costs-by reducing the balance-sheet effects of liability dollarization. Optimal reserves are derived under the assumption that central bankers conservatively choose reserves by balancing the expected cost of a Sudden Stop against the opportunity cost of holding reserves. Results are obtained without using calibration to match observed reserves levels, providing no a priori reason for our concept of optimal reserves to be in line with observed holdings. Remarkably, however, observed reserves on the eve of the global financial crisis were-on average-not distant from optimal reserves as derived in this model, indicating that reserve over-accumulation in Emerging Markets was not obvious. However, heterogeneity prevailed across regions: from a precautionary standpoint, Latin America was closest to model-based optimal levels, while reserves in Eastern Europe lay below optimal levels, and those in Asia lay above. Nonetheless, there are other motives for reserve accumulation: we find that differences between observed reserves and precautionary-motive optimal reserves are partly explained by the perceived presence of a lender of last resort, or characteristics such as being a large oil producer. However, to a first approximation, there is no clear evidence supporting the so-called neo-mercantilist motive for reserve accumulation.