People concerned with planning and analysis of the agricultural sector are faced with two general problems which in turn gives rise for two general types of models. One problem involves positive or predictive models which attempt to predict the "real world" as it actually exists based on response functions. Models directed towards these types of predictions are usually based on time series observations and use statistical estimation techniques such as regression equations. The second problem involves normative models, which ask the question: what conditions could prevail if certain conditions and goals were to be met? Frequently, these conditions have never prevailed in the past and time series observations do not exist. Problems of this type cannot be handled by time-series regression models but more nearly involve some type of operations research methods. Specific techniques in the set of possibilities include mathematical programming and systems simulation. Mathematical programming models lend themselves to great detail on spatial characteristics of agriculture that cannot be accomplished with time-series regression models.
Schatzer, Raymond J.; English, Burton C.; and Heady, Earl O., "A Tatonnement Model for Determining Future Market Prices and Quantities for some U.S. Crops" (1983). CARD Working Papers. 1.