An overshooting model of exchange rate determination and forecasting: a threshold regression approach

Abstract

Purpose This study examines the impact of structural shocks and policy interventions on the India/US exchange rate post the 1991 economic reforms in India. The study aims to improve forecasting accuracy by incorporating macroeconomic and microeconomic factors into the analysis using the threshold regression model (TRM), a nonlinear approach to estimation. Design/methodology/approach Extending Dornbusch’s (1976) overshooting model, the study incorporates micro factors, such as investor behaviour, beliefs and preferences, alongside traditional macroeconomic variables. Additionally, it introduces a capital control variable to assess monetary policy interventions. Using quarterly data from 1996Q2 to 2019Q3, TRM identifies two distinct economic regimes, providing a comprehensive understanding of India’s exchange rate dynamics. Findings The study reveals that macro and micro factors have varying effects on the exchange rate across regimes, reflecting India’s different economic conditions and policies. Furthermore, the TRM-based model achieves superior out-of-sample forecasting accuracy compared to the random walk model across all forecast horizons. Originality/value Unlike prior studies, where not all variables were deemed significant, our analysis demonstrates that all factors significantly influence the exchange rate. The innovative use of TRM deepens understanding of exchange rate behaviour, particularly in response to structural shocks and policy shifts. By identifying distinct economic regimes, the model offers insights into targeted policy measures tailored to India’s economic conditions, a previously unexplored perspective.