نرخ ارز در هند و اقدام بانک مرکزی : تجزیه و تحلیل EGARCH

We analyze the impact of conventional monetary policy measures such as interest rates, intervention, and other quantitative measures, on exchange rate level and volatility, and compare these to the impact of Central Bank communication using dummy variables in the best of a family of GARCH models estimated with daily and monthly Indian data. Since India has a managed float, we also test if the measures affect the level of the exchange rate. We find variations in the Euro/Dollar rate strongly affect the Rupee/Dollar level and volatility. The interest rate differential has strong perverse effects, tending to increase variance and depreciate the Indian currency. News decreases volatility as it adds to scarce information. Domestic policy variables affect both level and volatility, and persist at the monthly frequency, but sometimes work at cross-purposes. Communication channels have potential but were not used effectively.

Exponential growth has occurred in research on monetary policy, but the rich and challenging experiences in emerging markets are still under-explored. In this paper we estimate the best model in the family of autoregressive conditional heterosckedasticity (ARCH) and generalized ARCH (GARCH) models of exchange rate volatility, for the period following maturing of Indian money and foreign exchange (FX) markets. We insert policy dummies together with controls to study the impact on exchange rate volatility of conventional monetary policy measures such as interest rates, intervention and other quantitative measures, and of Central Bank (Reserve Bank of India, RBI) communication.1 Since India has a managed float, it is worthwhile to test if the measures also affect the level of the exchange rate, and if their effect persists beyond the very short-run. Since a range of policy instruments continue to be used on the path towards freer markets it is a good period to analyze the effectiveness of various instruments. An assessment of their relative impact contributes to understanding transition and the way forward. 1.1. Indian scenario In the past decade, there was rapid development in markets, in institutions and in instruments of monetary policy. A liquidity adjustment facility (LAF) was introduced. Injections and absorptions of liquidity largely kept the overnight inter-bank loan rate (the call money rate) in a band between two policy rates (Ghosh & Bhattacharya, 2009). The stated aim of Indian exchange rate policy is to reduce volatility, while the level is said to be market determined around fundamentals.2 In the period under analysis there was a movement away from a fixed exchange rate. Partial capital account convertibility followed full current account convertibility of the balance of payments. Equity flows were liberalized before debt flows.3 Surges in inflows have created problems for monetary management. Despite a current account deficit, reserves crossed the $300 billion mark in 2008.4 Development of foreign exchange markets has been rapid. The average daily turnover in Indian FX markets, which was about US $3.0 billion in 2001, grew to US $34 billion in 2007, the fastest rate of growth among world markets, BIS (2007). Growth in derivatives especially was strong, increasing to more than double the spot transactions (Goyal, 2011). Monetary policy follows a multiple indicator approach, giving weight to both inflation and growth. Though the RBI is not formally independent, a series of measures granted greater independence after the liberalizing reforms of the early nineties.5 A populous low per capita income democracy, where inflation is a politically sensitive issue, requires rapid monetary response to contain inflationary expectations. But at the same time developmental issues cannot be ignored. The RBI is not at the point of the impossible trinity, where monetary policy becomes ineffective, since the exchange rate is not fixed, and the capital account is not fully open. But it is a challenge to address the needs of the domestic cycle while managing external shocks. An important question is the impact of policy rates on the exchange rates. If this impact is low then rate change can be targeted to the domestic cycle. Alternative policy instruments are useful since segmented domestic financial markets make it difficult to close interest differentials. Moreover, differences in domestic and international policy cycles require positive differentials, as in the exit from the global financial crisis when emerging markets faced inflation while mature markets still battled deflation. As larger FX market turnover and rapid market deepening makes standard intervention less effective, communication could offer an additional instrument to policy. 1.2. Literature survey There is evidence of the effect of Central Bank (CB) communication, largely for developed countries. Blinder, Ehrmann, Fratzscher, Haan, and Jansen (2008) offer a survey. They argue that communication makes monetary policy more effective either by creating news, or reducing noise when markets are not perfect or there is learning.6 The literature suggests CB transparency can work in a number of ways depending upon how effectively the CB is able to maintain credibility and how public expectations are formed. It is not only CB communication that matters, but also its timing. Communication becomes intense before any monetary policy meeting to prepare the market for the forthcoming decisions. Fratzscher (2005) finds oral interventions to be more successful in moving exchange rates in the desired direction as compared to actual interventions. Literature on the effect of CB communication on exchange rates, to which our paper contributes, is still in a nascent stage for emerging markets. Since uncertainties are pervasive in such markets, communication should have a larger impact there. Goyal, Ayyappan Nair, and Samantaraya (2009) demonstrate this theoretically, and present some evidence for India, in a study of strategic interaction between monetary policy and FX markets. Fišer and Horváth (2010) show Czech National Bank communication tends to decrease exchange rate volatility using a GARCH framework. Égert (2007) finds appropriate CB communication enhances the effect of actual intervention and of interest rate news in the emerging markets of the European Union. Égert (2009) finds short-lived effects of CB communication and news on the South African Rand. There is relatively more work on the effect of intervention. Faced with large inflows a number of emerging market CBs intervene and believe it works to damp appreciation and volatility although in deep mature markets it is thought to be ineffective. Galati and Disyatat (2005) find Czech authorities intervene mainly in response to koruna appreciation, but could not find evidence of influence on short-term exchange rate volatility. In the Guimarães and Karacadag (2005) study of foreign market interventions in Mexico and Turkey, FX sales have a small impact on the exchange rate level and raise short-term volatility in Mexico. In Turkey, however, intervention does not appear to affect the exchange rate level, but reduces its short-term volatility. Domac and Mendoza (2002) estimate interventions to be highly successful in reducing exchange rate volatility in both Mexico and Turkey. They stress that sensible interventions could help contain the adverse impact of temporary exchange rate shocks on inflation and on financial stability. In most Indian studies of an earlier period RBI intervention decreases volatility (Edison et al., 2007, Pattanaik and Sahoo, 2003 and Goyal et al., 2009).7 Since the latter use a simultaneous equation technique, and include FX microstructure variables, their additional results are RBI intervention affects the level of the exchange rate and increases daily FX market turnover. In informal conversations, FX dealers often suggest that RBI intervention can increase FX market activity. Dealers with private information, who anticipate RBI action and its effect on the exchange rate, would use this to buy or sell, making money at the expense of less informed market participants. Any shock/new information to markets would increase expected returns and therefore volatility in high frequency data capturing actual trades. This is the creating news function of CB action. But studies show that, the effect can be in either direction in longer horizons (Blinder et al. (2008)). In the long run no news remains unprocessed. CB action can enhance scarce news and decrease the volatility of returns. 1.3. Study and results The literature implies the effectiveness of CB communication and intervention should be studied along with other macroeconomic variables and CB instruments. So the basic question we address is ‘What is the impact of various types of intervention (verbal and actual) and monetary policy measures on Indian exchange rate volatility and level?’ Policy measures are classified as follows: - Interest rates: Reverse repo rate, and repo rate. - Quantitative variables: Intervention, liquidity absorption or injection, and cash reserves. - Communication variables: Review, and speeches. Since we are dealing with high frequency data to analyze real time policy decisions, changes in domestic and international macroeconomic fundamentals such as relative money supply and output affect the exchange rate through the control variables such as news, US Federal Open Market Committee (FOMC) Meetings, Euro/US dollar (EURO/USD) exchange rate. The interest rate differential, which affects capital flows through the carry trade, is explicitly included as a control. Estimation of the monthly equilibrium exchange rate based on the monetary model, assuming purchasing power parity (PPP) and uncovered interest parity (UIP), (Appendix A) showed insignificant deviation of the rupee from equilibrium values, so that it was not necessary to include a deviation from equilibrium variable in the controls to capture threshold effects (as in Égert (2009)). The monetary model was estimated at a monthly level, using Indian and US variables: relative broad money supply, relative gross domestic product (GDP), and relative short-term Treasury bill rates. Since our aim is to study the effect of a number of policy variables on the exchange rate, time series modeling is most appropriate. Event study analysis can investigate the impact of only 2–3 variables at a time and results are highly subjective. EGARCH estimations for daily and monthly time periods show that the monthly affect, which allows for policy feedback and simultaneity, differs from the short-run daily effect, and the effect of policy actions persists. Control variables have major effects. During the period, policy actions were inadequate to neutralize the impact of USD movements, which dominated Indian rupee (INR) movements. The interest rate differential had strong perverse effects, tending to increase variance and depreciate the INR. Arbitrage opportunities induced capital inflows, which Indian capital account controls, including limits on bank open positions, were unable to restrain. But the negative effects of high interest rates on growth tend to reduce inflows, raise risk premiums and depreciate the exchange rate. US monetary policy announcements increase variance and depreciate the monthly exchange rate, so US policy strongly affects Indian markets. Although news generally increases volatility in markets, it has the reverse effect here. The creating news function of CB communication could be reducing noise in emerging markets, where information is scarce. Domestic policy variables affect both exchange rate level and volatility, and persist at the monthly frequency, but sometimes work at cross-purposes. Despite statements that policy only aims to reduce volatility it does affect the level of the exchange rate – so better coordination could improve impact. FX market intervention has a significant impact, again illustrating the importance of policy coordination. Intervention aims to reduce volatility, but taking account of the effect of policy variables on the level also, volatility actually increases. A rise in the liquidity adjustment facility (LAF) interest rates decreases daily variance. The reverse repo or absorption rate continues to decrease variance at the monthly frequency, and depreciates the exchange rate. Blunt instruments such as cash reserve requirements are also significant, but have limitations. Among communication variables, speeches are persistently significant, appreciating the daily exchange rate and decreasing its variance. But monetary policy announcements increase daily and monthly variance and depreciate the exchange rate at the monthly frequency. The results contrary to CB objectives of reducing volatility imply an ineffective use of the communication channel. Better communication design would allow more consistent effects, and realize the potential of this channel. The structure of the paper is as follows: Data and methodology are presented in Section 2. Section 3 analyses the empirical results. Section 4 concludes. Data sources and estimation details are in appendices.

In our tests of policy actions on exchange rate mean and volatility, using policy dummies in an EGARCH framework, communication outperforms more traditional policy variables. The results support the Blinder et al. (2008) position that in a climate of uncertainty CB actions matter. As a consequence of steady deepening of FX and money markets, while quantitative interventions continue to be important, communication can serve as a focal point, coordinating the actions of market participants (Sarno and Taylor (2001)). In particular, these variables allow the CB to achieve its objectives even when interest rates alone have perverse effects because of differentials being affected by risk premia or segmented domestic markets or asynchronous domestic cycles make it difficult to close the differentials. Given the stated RBI objective was to reduce volatility, blunt quantitative actions such as cash reserves have perverse effects. A positive interest differential increases volatility and does not strengthen the long-run exchange rate. News tends to calm markets, suggesting that in emerging markets news may be at less than optimal levels. This greater uncertainty, combined with a credible CB, gives impact to CB communication, but this potential is underutilized in our period of analysis. Communication and FX market intervention not only affect exchange rate volatility but also mean levels despite policy statements that there is no target exchange rate. It follows that policy makers would gain by investigating and evaluating the impact of alternative instruments, one by one and together, and by coordinating policy action on levels and volatility. The communication channel needs to be further studied, developed, and used more intensively.