Tea Price Dynamics in the Kolkata Market: An Asymmetric ARMAXEGARCH Approach with Seasonal Regressors
Published by: Admin
Authors: Bhola Nath, Rohit Kumar and Yogita Sharma
Abstract
Forecasting prices of agricultural commodity is challenging due to long-term trends, seasonal fluctuations, and market volatility. The present study investigates the behaviour of monthly tea prices in the Kolkata market, which is also known as India’s principal tea trading hub. To fulfill the objectives of the study, we have used tea prices from January 1960 to April 2026. Several forecasting models, including Seasonal Naïve (SNaïve), Exponential Smoothing (ETS), and Seasonal ARIMA (SARIMA), were evaluated. SARIMA achieved the best performance among the linear alternatives (RMSE: 0.2132 and MAPE: 6.0545). However, residual diagnostics uncovered prominent volatility clustering and conditional heteroskedasticity, demonstrating that linear models fail to capture the market’s true risk structure (Lag 4, Lag 8 and Lag 12 with p=0.001). To address these limitations, an integrated ARMAX-EGARCH (1,1) model with monthly seasonal dummy variables and a Skewed Student-t distribution was developed. The proposed framework simultaneously captures serial dependence, recurring seasonal patterns, asymmetric market responses, and heavy-tailed price movements. The empirical results show structural seasonal changes, enduring volatility memory, and significant heavy-tailed risk factors. The model reveals a significant asymmetric leverage impact, indicating that negative price shocks generate over double the volatility of positive shocks. The findings demonstrate that incorporating asymmetric volatility dynamics significantly improves the understanding and forecasting of tea price behavior. The proposed approach provides a reliable risk-adjusted forecasting tool that can support decision-making by producers, traders, agribusiness firms, and policymakers operating in volatile agricultural markets.