Empirical testing of alternative price spread models in the South African maize market

Reduced‐form price spread models have been recently utilized by Wohlgenant and Mullen, and Thompson and Lyon to evaluate the economic factors affecting the marketing margins for agricultural products. Drawing on Gardner, Heien, Buse and Brandow, Waugh, Tomek and Robinson, and others they specify alternative retail‐farm price spread models and attempt to determine which best fit the data in the context of underlying theoretical rationale. This paper continues in the spirit of Wohlgenant and Mullen, and Thompson and Lyon by evaluating alternative specifications of the retail‐farm price spread for white maize in South Africa. However, several important differences do remain. Wohlgenant and Mullen analyzed the price spread for beef using annual data, while Thompson and Lyon modeled the price spread for oranges using weekly data. The time period under consideration can be expected to affect the choice of model because fixed markup rules that might be evident using a short‐run period of analysis (e.g., Thompson and Lyon) become untenable over the long run with underlying supply and demand shifts. In this paper, monthly data, which may be interpreted as an intermediate‐run period, are used along with dichotomous supply‐demand shifters. In addition, Brorsen et. al. have shown that price uncertainty affects the price spread in the marketing channels of agricultural commodities. Thus, the analysis in this paper extends the framework of Wohlgenant and Mullen, and Thompson and Lyon to include measures of price risk. Finally, like Brorsen et. al. this study pertains to the grain market, while Wohlgenant and Mullen, and Thompson and Lyon studied the marketing margin for non‐storable commodities.