Projections of US farm numbers by Markov processes

The structure of agriculture for the remaining years of the twentieth century is an issue crucial to farmers and ranchers. Much interest has been focused in recent years on the structure of agriculture and specifically on the size distribution and number of farms and ranches. In recent studies, Markov chain processes have been applied to help predict the numbers and sizes of farms, and to analyze the movement of firms between size classes;Each of the previous attempts to apply the Markov process to the problem of estimating and projecting farm numbers has encountered some major difficulty in deriving a solution. Several of the model formulation and computational difficulties, such as the addition of a feasibility restriction, external constraints, cycling difficulties and computer algorithms, have been solved in the first two papers of this study;This study synthesized the results , problems and advances of previous reports to predict farm numbers by size class. Values of sales classes were chosen to represent size classes of farms, thus avoiding definitional questions of farms, ranches and nonland based farming operations;A series of USDA indexes were chosen for explanatory variables in the Markov regression problem. Approximately 30 combinations of nine different indexes were applied to the Markov model;Model results indicate that the historical decline in farm numbers will continue to a level of about two million farms by 1990. The farm number projections are closely tied to growth of remaining firms. The primary factors causing or allowing firms to grow in size include increases in the prices that farmers must pay for inputs and increased productivity of those inputs, especially their own labor;The results of this study indicate a continuing trend toward fewer, larger farms. Although much of the increase in size is caused by increases in prices received rather than real growth, there will be real growth as well.