Modeling exchange rate volatility with GARCH models: A comparison based on a volatility breaks

Mubeen Abdur Rehman and Dr. Ashfaq Salamat

Abstract

Volatility has been defined as a worthy indicator of uncertainty, which has implications on various factors such as international trade, investment decisions, and valuation for a currency. This paper investigates the structure of volatility in the exchange rate data by considering a structural break. Monthly data of Pakistani Rupee's exchange rate is considered for 21 years starting from January 2000 to November 2020. The State Bank of Pakistan supplied the nominal exchange rate data. It is found that the threshold GARCH (TGARCH) model is more suitable to estimate the volatility of the exchange rate for comprehensive data of 21 years. Results show that if data is bifurcated based on structural break, then the low and high volatility can be estimated more accurately with exponential GARCH (EGARCH) and square GARCH (SGARCH), respectively. Before the structural break, the exchange rate volatility is lower than that after the structural break due to
clustering volatility. Also, the research showed that the volatility clustering effect is found in the volatility of exchange rate data as low volatility is
followed by low and high volatility is followed by high volatility for a prolonged period.
Keywords: Volatility, exchange rate, structural break, GARCH, TGARCH, SGARCH, EGARCH.