Forecasting of Onion Price through GARCH and EGARCH Time Series Models in Nasik District of Maharashtra

Abstract

The experiment was conducted from March, 2023 to March, 2024 at Dr. RPCAU, Pusa, Bihar, India to study the performance of GARCH and EGARCH models for forecasting onion prices. The study was based on secondary data on onion prices taken from Agmarket, Indiastat, and other websites. Data on modal prices was collected from January 2011 to March 2024. The study employed Generalized Autoregressive Conditional Heteroskedasticity (GARCH) and Exponential GARCH (EGARCH) models to forecast onion prices in the Nasik district of Maharashtra, addressing price fluctuations that affected farmers and consumers. Weekly onion price data from January 2011 to March 2024 was reviewed to identify volatility and trends. After differencing, the ADF test verified stationarity, and the ARCH-LM test confirmed the existence of volatility clustering. Asymmetric volatility was captured by the EGARCH model, while symmetric volatility was captured by the GARCH model. Model performance was assessed using AIC, BIC, and error measures such as MSE, MAE, and MAPE. The GARCH model performed better than the EGARCH model in predicting onion prices, according to the results, which showed lower error metrics with MSE, MAE, and MAPE values of 68,771.29, 208.59, and 15.91%, respectively. EGARCH, however, provided important insights into asymmetric price volatility, wherein price swings were more affected by positive shocks than by negative ones. This study highlighted GARCH’s efficiency for short-term price forecasting and EGARCH’s utility in understanding complex market behavior. The findings supported robust statistical approaches to managing price volatility, aiding farmers and policymakers in mitigating market risks.