ISAR Publisher

International Scientific and Academic Research Publisher

Submit Manuscript

Volatility, Seasonality, and Tomato Price Forecasting in Kupang City: Implications for Regional Price Stabilisation


Author: Louis Marlina*, Irnad, Reflish
Master of Agribusiness Study Program, Faculty of Agriculture, University of Bengkulu
Published Date: 2024-08-31
Keywords: forecasting accuracy, exponential smoothing, tomato price, seasonal index, Kupang City, moving average.
Abstract:

Tomatoes are a strategic horticultural commodity whose price instability has a direct impact on farmers' incomes, consumer purchasing power, and regional inflation, especially in cities in eastern Indonesia that depend on supply from outside the region. This study aims to characterize the dynamics, volatility, and seasonal patterns of tomato prices in Kupang City, compare the accuracy of forecasting exponential smoothing models, moving averages, and linear trends, and compile price forecasts for 2024. The data used are monthly retail prices for the period January 2019 to December 2023 (n = 60) sourced from the Central Statistics Agency of East Nusa Tenggara Province. All accuracy measures were calculated strictly on the basis of one-step forward forecasting using mean absolute deviation, mean squared error, root mean squared error, and mean absolute percentage error. The results showed an overall coefficient of variation of 26.41 percent, with the highest price of IDR 24,792 per kilogram occurring in January 2023 and the lowest price of IDR 7,083 per kilogram in October 2019. The most dominant feature of this series is the very strong annual seasonal cycle, where the seasonal index moves from 68.9 in September to 128.2 in April with a range of fifty-nine index points. The Dickey-Fuller augmented test rejects the unit root hypothesis so that data differentiation is not necessary. The slope coefficient of the linear trend is positive but not statistically significant and only explains 0.56 percent of price diversity. Among the three methods defined, exponential smoothing with a smoothing constant of 0.95 is the most accurate model with a mean absolute percentage error of 16.79 percent, followed by a two-period moving average of 19.63 percent and a linear trend of 23.71 percent. The inclusion of seasonal indices into the exponential smoothing model substantially reduced the error to 12.12 percent. The recommended forecast for 2024 ranges from IDR 9,698 per kilogram in September to IDR 18,041 per kilogram in April. These findings show that seasonality, not trends, should be the basis for analysis and policy for tomato price management in Kupang City.