Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 1993 → A supply forecasting model for Zimbabwe’s corn sector : a time series and structural analysis

Texas A&M University (1993)

A supply forecasting model for Zimbabwe’s corn sector : a time series and structural analysis

Makaudze, Ephias

Titre : A supply forecasting model for Zimbabwe’s corn sector : a time series and structural analysis

Auteur : Makaudze, Ephias

Université de soutenance : Texas A&M University

Grade : Master of Science (MS) 1993

Résumé
The Zimbabwean government utilizes the corn supply forecasts to establish producer prices for the following growing season, estimate corn storage and handling costs, project corn import needs and associated costs, and to assess the Grain Marketing Board’s financial resource needs. Thus, the corn supply forecasts are important information used by the government for contingency planning, decision-making, policy-formulation and implementation. As such, the need for accurate forecasts is obvious. The objectives of the study are : (a) determine how changes in the government-established producer price affects the quantity of corn supplied to the Grain Marketing Board by the large-scale corn-producing sector and (b) whether including rainfall or rainfall probabilities into econometric models would result in an improvement of corn supply forecasts compared to current forecasts by the government. In order to accomplish the first objective a supply elasticity model was specified and estimated using ordinary least squares. This model is intended to provide ’de insight to the government regarding the influence of the government-established corn price and other related variables on corn supplied to the Grain Marketing Board by the large-scale producers. Thus, the estimated model would be useful to the government when establishing corn prices in March/April for production in the following growing season (October - February). To achieve the second objective, preliminary analysis was carried out to verify whether there is statistical evidence to support the hypothesis that rainfall cause" corn production and supply, and also corn prices and sales. Specifically the preliminary analysis involved using the Granger causality tests, stationarity tests and innovation accounting (impulse responses and forecast error decomposition). Having verified and quantified the causal effects of rainfall on corn production and supply, the next task was to investigate whether including rainfall and/or drought probabilities into forecasting econometric models would help provide improved out-of-sample forecasts compared to the government’s forecasts. The forecasting accuracy of the models (short-run) was evaluated using standard statistical measures such as, the mean square error (MSE), mean absolute percentage error (MAPEI), improved mean absolute percentage error (IMAPE) and Theil’s U-statistic, and thereupon select the best model. The results indicated that by incorporating rainfall and/or rainfall probabilities into econometric forecasting models, there was substantial improvement in corn supply forecasts. It follows that the the government would likely find it beneficial to incorporate the rainfall variable into their forecasting effort.

Présentation

Version intégrale (2,78 Mb)

Page publiée le 10 novembre 2012, mise à jour le 23 février 2018