Official Intervention, Reserve Accumulation and Exchange Rate Volatility

Abstract

The Reserve Bank of India often claims that the official intervention in the foreign exchange market primarily aims at minimizing undue fluctuations in the exchange rate and this study is an attempt to explore the success story of such interventions. Although the stylized facts seem to indicate that maintaining adequate amount of official reserves help reduce the volatility, the marginal benefit of adding reserves in terms of containing excessive volatility declines as the level of reserve holding increases beyond certain threshold level. The evidence from a threshold vector autoregression model suggest that the response of exchange rate volatility is conditional upon the size of intervention and whether the level of reserve holdings is above or below certain threshold level. While the official purchase of foreign exchange reduces the variability of exchange rate, official sale seems to trigger volatility irrespective of whether reserve holding is above or below the threshold level. Further, a positive shock to absolute official sale (purchase) could reduce the pace of depreciation (appreciation) although it could not revert the direction of exchange rate. These evidences exemplify the fact that intervention cannot prevent exchange rate volatility from rising rather it can only moderate.

Appendices

Appendix A

Objectives of Official Intervention

Name of the countries	Objectives
China	To keep the exchange rate stable at an adaptive and an equilibrium level
Japan	To stabilize the value of the exchange rate
Switzerland	To align the exchange rate in line with monetary policy to influence the monetary conditions
Euro area	To maintain price stability
India	To avoid undue fluctuations in exchange rate
Taiwan	To avoid fluctuations in exchange rate arising out of seasonal and irregular events
Russia	To support financial stability and to service external debt for several years, even in times of distress
Hongkong	To fix the exchange rate
Korea	To ensure stability in the foreign exchange market
Saudi Arabia	To fix the exchange rate
Singapore	To fix the exchange rate
Brazil	To contain disorderly movements of the exchange rate
US	To counter disorderly market conditions
Thailand	To accumulate reserves which help in fulfilling the monetary and exchange rate policies
Israel	To meet external obligations, counter forex market crises and improve the country’s financial credibility
Mexico	To contain exchange rate volatility and thereby ensure financial stability
United Kingdom	To check undue fluctuations in the exchange rate
Czech Republic	To be used as an instrument of monetary policy
Poland	To curtail excess exchange rate volatility
Indonesia	To counter demand and supply mismatches in the forex market which help in achieving exchange rate stability as well as currency liquidity simultaneously
United Arab Emirates	To fix the exchange rate
Malaysia	To reduce undue exchange rate volatility, secure orderly conditions in the forex market
Vietnam	To maintain stability and control inflation
Canada	To promote orderly market conditions for the national currency
Philippines	To bring in monetary stability, check excessive exchange market speculation and to reduce sudden and sharp capital movements
Norway	To curb excess volatility of the exchange rate

Appendix B

The construct of weekly time varying conditional volatility of percentage change in ₹/$ exchange rate is obtained using the following EGARCH specification:

$${\widehat{\rm E}}_{t}=-0.007+0.101{\widehat{E}}_{t-1}+\sqrt{{\sigma }_{t}^{2}} {\nu }_{t}$$

$$ \left(0.61\right)\hspace{1em}\left(0.00\right)\hspace{1em}\hspace{1em}\hspace{0.33em}$$

$ \begin{gathered} \log \sigma _{t}^{2} = - 0.099 + 0.508\left| {\frac{{\varepsilon _{{t - 1}}^{2} }}{{\sqrt {\sigma _{{t - 1}}^{2} } }}} \right| - 0.378\left| {\frac{{\varepsilon _{{t - 2}}^{2} }}{{\sqrt {\sigma _{{t - 2}}^{2} } }}} \right| + 0.030\left( {\frac{{\varepsilon _{{t - 1}}^{2} }}{{\sqrt {\sigma _{{t - 1}}^{2} } }}} \right) + 0.991\left( {\log \sigma _{{t - 1}}^{2} } \right) \hfill \\ \quad \quad \quad \quad \left( {0.00} \right)\,\,\quad\left( {0.00} \right)\quad \,\,\,\,\quad \quad \;\left( {0.00} \right)\,\,\,\quad \quad \quad \quad \left( {0.00} \right)\quad \quad \quad \quad \,\,\quad\quad \left( {0.00} \right) \hfill \\ \end{gathered} $

where ${\widehat{\rm E}}_{t}$ is weekly percentage change in ₹/$ exchange rate. The p-values in parentheses indicate that all the coefficients, excepting constant in the mean equation, are found to be statistically significant. The estimated ${\chi }^{2}$ statistic is found to be 5.067 with a p-value of 0.16; suggesting that the null of no autocorrelation in the standardized residuals (${\nu }_{t}$) upto a lag order of 4 can be accepted. The F-statistic to test the null that there is no remaining ARCH effect upto 4th lag in the square of the standardized residuals is estimated to be 0.571 with a p-value of 0.68; suggesting that the EGARCH specification is appropriate to model exchange rate volatility.