Mathematical programming models as tools for centralized planning in decentralized decision situations:A critical examination of the reliability of agricultural sector models as a basis for policy decisions

Summary We consider the solution of decision problems of political agencies that are interested in the agricultural production activity, its spatial or temporal distribution etc. on a regional, national or international level. These agencies are faced with decision problems concerning the realization of economic activities that are basically determined by the decisions of individual farms that are free to make their own decisions within the limits of economic and technological constraints. This raises the question of to what extent an agency's computations of possible future economic activities will conform to reality if its information about the farms' future decision situation is not complete. Examples show that the computed results might be so unreliable that their relevance for the solution of an agency's decision problem might be very limited. In an economy of a non-command decision structure this applies to the use of ‘positive’ and ‘normative’ models. The paper discusses the relevance of reliability examinations for the evaluation of computed agricultural production activity. Furthermore, approaches for the identification and improvement of the reliability of computed results are discussed. Although the article concentrates on the use of programming models for the simulation of possible activities of the farms, it refers also to other types of models that might be used for this purpose.