Modeling of Wholesale Prices for Selected Vegetables Using Time Series Models in Kenya

Abstract

Price forecasting is more sensitive with vegetable crops due to their highly nature of perishability and seasonality and is often used to make better-informed decisions and to manage price risk. Further, to improve domestic market potential for smallholder producers, who are the biggest suppliers in the market and in line with the government‟s Agriculture Sector Development Strategy (ASDS). Three autoregressive models are used to predict and model the wholesale prices for selected vegetables in Kenya shillings per kilogram. The models are; Autoregressive Moving Average (ARMA), Vector Autoregressive (VAR), Generalized Autoregressive Condition Heterostadicity (GARCH) and the mixed model of ARMA and GARCH. This time series data for tomato, potato, cabbages, kales and onions for markets in Nairobi, Mombasa, Kisumu, Eldoret and Nakuru wholesale markets are considered as the classical national average. The result indicates the models are valid in predicting. Based on the model selection criterion the best forecasting models in ARIMA are; Potato ARIMA (1,1,0), Cabbages ARIMA (2,1,2), tomato ARIMA (3,0,1), onions ARIMA (1,0,0), Kales ARIMA (1,1,0) . Further, the mixed model of ARMA (1, 1) and GARCH (1, 1) model is also identified best model in forecasting.