Preference asymmetry and international reserve accretion in India

Abstract

Reduced-form estimates of the Reserve Bank of India's (RBI) first-order condition indicate that its preferences have been asymmetric with respect to exchange-rate management, with the response to the rate of rupee appreciation being relatively larger than to the rate of rupee depreciation of the same magnitude. This behaviour is shown to account for a sizable fraction of reserve accretion in recent years.

Notes

¹ Moreover, the surge in international reserves holding among emerging market economies has attracted considerable attention recently since much of the foreign capital flowing into the United States has come from foreign central banks, which have allowed the U.S. current account deficit to be financed at prevailing asset prices and exchange rates (Summers, 2006).

² There is by now a large literature on monetary policy reaction functions, which focuses on how authorities adjust interest rates in response to goal variables. Our focus is, however, specifically on reaction functions designed to examine exchange rate management (Sarno and Taylor, 2001). Thus, intervention can be thought of as our policy instrument.

³ The traditional approach to the estimation of intervention equation has often been criticized because of the ad hoc specification typically employed by this literature. More recently, Almekinders and Eijffinger (1996) obtain an intervention reaction function by combining the exchange rate model with a particular loss function for the central bank. A similar strategy is followed here.

⁴ Ideally, we should use the deviation of real exchange rate from target. However, we use the nominal exchange rate, as monitoring it, as opposed to the real exchange rate has been the official policy (Jalan, 1999).

⁵ Whether or not official intervention is effective in influencing exchange rates is an issue of crucial policy importance, and have been subject of a vast academic literature (see Edison, 1993 for a survey). These issues are beyond the scope of this article.

⁶ We realize that changes in reserves represent a very crude proxy for intervention activity since reserves may change for a number of reasons and often not related to official intervention. Reserves increase, for example, with valuation changes on existing reserves. Our results are subject to these data limitations.

⁷ We note that determining the ‘optimal’ level of reserves is not straightforward: in fact, it might also depend on institutional factors (Frenkel and Jovanovic, 1981). For the purpose on hand we assume this to be a constant as the objective of the article is not what constitutes optimal reserve demand.

⁸ The gold stock, SDRs and Reserve Tranche Position held with the IMF are not included, as they constitute a very negligible proportion of reserves and are not used as an intervention asset.