Identification of the Characteristics of the Regional Strategic Development Based on the Indicators’ Statistical Analysis

Abstract

Forecasting of the economic indicators’ dynamics is an important task that ensures the economic security of the Russian Federation regions. Statistical analysis reveals key linkages between the indicators, even if their nature is unknown. We aimed to develop and verify a method for identifying regional factors without taking into account federal trends towards the economic conditions’ improvement or deterioration. We used regression analysis for assessing the changes in the corresponding indicators’ values for the previous periods. We assumed that the nature of the indicators’ impact for the previous years does not depend in a statistically meaningful way on a region and analysed year. For the short-term (one-year) forecast, we used the multiple linear regression method. Assessment of the quality of forecasting the indicators’ changes was based on the adjusted determination coefficient. We showed that separation of federal trends increases the regional indicators’ predictability. Further, we performed the long-term forecast using the Monte Carlo method. We predicted the indicators’ values based on the obtained regression formula adding random variables corresponding to the regression’s standard error. We presented the result of the calculations as percentile estimates of the indicators’ values. Finally, we verified this method, using a retrospective forecast that has shown a good agreement with the real data. The study’s results can be used as a basis for constructing a system of statistical forecasting of the development dynamics in the Russian regions. One of this method’s limitations, particularly, is a tendency to changing the indicators’ predictability for different years, which leads to an inaccuracy in assessing the possible deviation of the indicators’ values. The presented method only predicts regional indicators normalized by condition of the state economy as whole. Future research will be focused on identifying the nonlinear relationships between the indicators.