|
PresentationsOn the study of the structure of market processesFinancial University under the Government of the Russian Federation, Department of AD, and PR FT, Russia, 125993, Moscow, Leningrad prosp., 49, 8 (926) 499-47-36, kontsevaya07@list.ru The specificity of the dynamics of market indicators is the correlation of their changes with seasonal factors, i.e. the structure of market series includes, in addition to long-term trends, a cyclic (periodic) component. There are several approaches to the allocation of seasonal fluctuations in the structure of market data – based on the decomposition into a harmonic series, based on the use of additive and multiplicative models with a cyclic component. The paper compares these approaches and assesses their shortcomings and advantages of methods on the example of the study of the dynamics of labor market indicators. The constructed models are evaluated and ranked in terms of quality and accuracy, the accuracy of predictive properties is estimated. The choice of the best model is proved, the results of forecasting on its basis are shown. A combined method of modeling combining both adaptive and regression models is proposed. The statistical base for the study was the dynamics of indicators of the youth labor market on the example of a single region. The last few years in Russia there are radical transformations in the labor market. The area of employment has traditionally been difficult, and in the light of the large-scale reduction in both the demand for specialists and the jobs themselves, today the problem of employment becomes extremely important for maintaining stability in society. There are many reasons for the existing tension in the labour market, and in order to optimize the mechanism of interaction between the subjects in this market, in order to correctly assess the prospects for development and future levels of unemployment and to effectively organize a set of necessary measures to control and manage market processes, it is necessary to take a systematic approach to the study and modeling of the labour market. Comparison of the models allows us to make a choice in favor of the Holt-winters model [1], since this model demonstrates the damping of the oscillation amplitude synchronously with the initial process, while the use of dummy variables may be justified in the case of a constant amplitude of seasonal changes. Another criterion for selection is the accuracy of the retrospective forecast – for example, in the study, the error in predicting the assessment of the indicator for the third quarter of 2018 (the last actual observation) was 0.46%. Thus, an adaptive Holt-winters model is recommended for practical use, provided that its parameters are optimized. In this paper, as a criterion of optimization, we used an approach that implements the minimization of deviation squares.
|
