A Model of Analysing Long Memory with And Without Structural Breaks in Specific Reference to Indian Exchange Rate with Major Currencies

Abstract

This paper investigates presence of long memory with and without structural breaks in exchange rate data. We propose to compare between standard GARCH and FIGARCH to find out suitable model for capturing volatility dynamics in case of INR versus USD, GBP, EURO and YEN over the sample period.  We run two tests; GPH and Hurst exponent to detect presence of long memory before applying GARCH based models. Structural breaks were identified endogenously using Bai Perron test (BP) and optimum numbers of breakpoints were confirmed from log likelihood of BP test. We empirically examined daily return data from January 2000 through July 2018 and found presence of mild long memory in case of INR/USD and INR/YEN and strong memory in other two. Our result shows the evidence of parameter instability in case of INR/USD and INR/YEN while applying FIGARCH (1, d,1). We however found that volatility persistence significantly decreases in all the series except for INR/YEN after incorporating structural breaks in FIGARCH model. In case of INR/YEN, the value of long memory estimate has rather increased after adjusting for breaks. This study will have implications for investors and portfolio managers.