Economic Time Series Decomposition Using Seasonal ARIMA in Macro-Level Planning Models

The macro-level indicators, including Gross Domestic Product (GDP), inflation, and employment, cannot be explained without economic time series decomposition. SARIMA models provide an efficient statistical methodology that can be used to trace such conditional seasonal economic trends. Nevertheless, the current procedures tend to ignore the complex seasonal correlations and cannot differentiate between transient and enduring economic shifts, which result in the inadvisory policy decisions. The current paper suggests a model of GDP growth pattern decomposition by the use of SARIMA to improve the accuracy of macroeconomic planning. The suggested approach is able to provide more accurate trend, seasonal, and residual separation, which is a major weakness of the previous models. The SARIMA-based decomposition can help the policymakers to develop data-based strategies of fiscal and monetary policies by exposing the economic dynamic that may have been supported by seasonal changes under the surface. SARIMA in the experimental results has a forecast accuracy of 6.5, an error rate of 11.5, the data processing time is 95 seconds and moderate computational complexity (5/10). These values prove that SARIMA is better than the available models in terms of accuracy and the efficiency of operation.

Перевод пока недоступен