ANALYSIS OF GDP, INFLATION, AND UNEMPLOYMENT IN KAZAKHSTAN BASED ON ARIMA AND VAR MODELS

Abstract

This article is devoted to the analysis of changes in the macroeconomic indicators of the Republic of Kazakhstan – Gross Domestic Product (GDP), inflation, and unemployment rate – using the econometric models ARIMA and VAR. Amid economic instability and the country’s dependence on external factors, the issue of accurately forecasting and interpreting these indicators is particularly relevant. The article aims to determine how effectively classical time series methods can model and reveal the state of key macroeconomic variables in Kazakhstan’s economy. By comparing the ARIMA and VAR models based on Kazakhstan’s macroeconomic data for the period from 1994 to 2023, the study demonstrates that classical time series methods are effective tools for short-term forecasting and for understanding the main dynamic interrelationships between variables. The computational and analytical part of the study was performed using the R Studio software environment. The results of the econometric modeling enabled the identification of interconnections and dynamic dependencies among these indicators. Furthermore, the application of these modeling methods can serve as a scientific and practical foundation for forecasting the future development of the national economy. The obtained results can be used by policymakers as an effective tool for short-term forecasting using classical time series models. However, during periods of structural shocks, such as the COVID-19 pandemic – the limitations of these models become especially apparent. Therefore, in such periods, it is advisable to use time series models cautiously and to complement them with alternative approaches. The theoretical significance of the study lies in its ability to facilitate a deeper understanding of the dynamic relationships between economic indicators using econometric models. From a practical perspective, the findings may be valuable for national economic policy planning and the timely implementation of anti-crisis measures. For future research, it is recommended to further improve forecasting accuracy by applying structural VAR models or machine learning methods.