Stochastic forecasting of the grain industry development parameters in the Voronezh region

Abstract

Purpose: the article is devoted to the use of stochastic methodological tools for long-term forecasting of regional level grain production. Discussion: those tools is a probabilistic models set of grain production basic indicators. A multiple experiment on these models allows obtaining statistical characteristics of the grain industry development parameters, on the basis of which alternative scenarios are designed. Within the proposed methodology framework, the regional municipal districts, which have similar patterns of grain yield long dynamics, are grouped on the basis of cluster analysis. A yield deviations map was compiled for each of the selected groups. Deviation maps are used to simulate pseudo-random mechanisms for fluctuating crop yields in the future, including persistent recurring situations (patterns) of group and general trends. Forecasting the harvested areas dynamics of grain crops is proposed to be implemented on the basis of two components: trend models of time series and probabilistic models of oscillations, including the previously observed deviations characteristics. A lot of alternative forecasts of the grain industry development parameters are formed on the basis of yields and harvest areas forecasts using the Monte Carlo method. Statistical processing of the obtained data set allows to interpret the forecast parameters as a whole, noting different various development scenarios. Results: the proposed probabilistic simulation model of grain industry development parameters prediction was tested on the Voronezh Region data for 1976-2016.