An empirical analysis of stochastic behavior of Sri Lanka exchange rate changes.

Abstract

The main purpose of this study was to explore the main characteristics of stochastic behaviour of Sri Lankan exchange rate aginst to US dollar,(LKR/US$). This study used daily spot exchange rate time series collected from Central Bank, Sri Lanka web site. The study covers the time period from 2008 to 2010, which represents 722 trading days. The sample period was divided into two. One period was from January 1, 2008 to May 19, 2009 and the other period from May 30, 2009 to December 31, 2010. Graphical techniques, Kernel density function, autocorrelation function, and GARCH models were used to analyse the behaviour of the exchange rate in this study. The results show that basic statistical properties of Sri Lankan exchange rate series was a nonlinear, asymetric shape ,nonstationary series with stochastic trend, I(1). The change in the logarithm of the daily exchange rate (Exchange rate return) series has fatter tails, serial dependence, volatility clustering and ARCH effects in both sample periods. During the period I, the exchange rate was depreciating, distribution was positively skewed , larger volatility (SD=3.4) , non normal and nonstationary. During the period II, exchange rate was appreciating, high persistent and skewed negatively. The changes of log exchange rate behave as normal with an autoregressive conditional heteroscedasticity process for innovations. The characteristics of the exchange rate changes indicates the presence of heterogeneity among market participants as well as changing parameters over time. Standard deviation of this distribution dominates the mean value. variance was also time varying. The results of this study has important implications for exchange rate determination, balance of payments, risk modeling and management, forecasting, market efficiency, statistical inference in empirical work and for the economy, as whole.