Assessing the Dynamics of Interaction in Multiple Time-Varying Variables Via the Symmetric Lag Vector AutoRegressive (VAR) Model | ||||
| التجارة والتمويل | ||||
| Volume 44, Issue 4, December 2024, Page 162-181 PDF (1.07 MB) | ||||
| DOI: 10.21608/caf.2024.392536 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
| Sohair F Higazi1; Hani A, Khedr2; Sarah F Aboud3 | ||||
| 1Professor of Applied Statistics, Faculty of Commerce, Tanta University | ||||
| 2Lecturer, Statistics Department, Faculty of Commerce, Damanhour University | ||||
| 3Lecturer, Statistics Department, High Institute of Computers, Information and Management Technology, Tanta. | ||||
| Abstract | ||||
| This paper aims to investigate the effects and interactions of time series data when using the bivariate and the multivariate symmetric VAR model; and also, to evaluate the impact of relationship on shock and on short and long-term Impulse response analysis. Data used in the analysis were World Bank data for Egypt for the years from 1990 to 2020. Two symmetric VAR models were used to analyze the relationship between inflation (INF) and some key macroeconomic indicators that include the trade balance deficit (TBD), gross domestic product (GDP), government expenditure (GEX), and foreign investments (FV). The first model used was the bivariate VAR model, which focuses on inflation (INF) and trade balance deficit (TBD), and then, applying a multivariate VAR model containing all other variables. The primary objective was to assess the accuracy and explanatory power of these models in forecasting inflation based on Akaike information criterion (AIC), Schwarz criterion (SC), and also on Hannan-Quinn information criterion (HQ). Statistical results show that the multivariate VAR model significantly improves forecasting accuracy; the bivariate model achieved lower values for AIC and SC indicating a simpler structure, but the multivariate model demonstrates more robust performance, particularly in capturing the long-term effects of additional variables. Both models revealed that inflation is primarily driven by its own shocks, with only a minor contribution from the trade balance deficit. However, the multivariate model provided a broader explanation of inflation’s variance, as financial variables became increasingly influential over time. Impulse response analysis indicated that both models exhibited similar short-term reactions to shocks, though the effects diminished faster in the multivariate model. The study concludes that the bivariate VAR model is more appropriate for analyzing the effects of a variable on itself and short-term shock impacts, while the multivariate VAR model is more suited for variance decomposition and understanding complex economic dynamics. The recommendations emphasize the importance of advanced statistical models like the multivariate VAR to enhance forecasting accuracy and long-term Impulse Response Analysis economic decision-making. And the univariate VAR model for short-term Impulse Response analysis. | ||||
| Keywords | ||||
| Autoregressive Models; Variance Decomposition; Box-Jenkins Models; Asymmetric Lags Inflation; Economic Growth; Impulse Shock response; Granger Causality; Optimal Lag | ||||
|
Statistics Article View: 104 PDF Download: 66 |
||||
