A Time-Varying Bayesian Compressed Vector Autoregression for Macroeconomic Forecasting

Abstract

This paper presents macroeconomic forecasting by using a time-varying Bayesian compressed vector autoregression approach. We apply a random compression by using projection matrix to randomly select predictive variables in vector autoregression (VAR), and then perform true out-of-sample forecast where the forecast values are averaged across all estimated models, containing different in both explanatory variables and number of those variables by using Bayesian model averaging (BMA). In addition to this, we allow the parameters in Bayesian compressed VAR to be time-varying by implementing dynamic model averaging (DMA) algorithm that is applicable with VAR using forgetting factor to control the degree of time-varying in the estimating parameters. We validate the performance of the proposed method via real macroeconomic data including up to 53 variables. The empirical results demonstrate that the predictive performance of time-varying Bayesian compressed VAR can beat traditional VAR types which are considered to have a potentiality to deal with large size variables. INDEX TERMS Bayesian econometrics, macroeconomics, Bayesian model averaging, mathematical models, time-varying parameters, dynamic model averaging, compression, Kalman filter

The Positive Influences of Renewable Energy Consumption on Financial Development and Economic Growth

“…VAR model has some significant advantages. For instance, multiple variables can be evaluated at the same time with the help of VAR methodology (Ao, 2010; Aunsri & Taveeapiradeecharoen, 2020). In addition, the impact of an exogenous shock of one variable in the system on the others can be investigated by using this approach (Luo et al, 2019).…”

Estimation of market efficiency process within time-varying autoregressive models by extended Kalman filtering approach

“…We may mention that time-varying autoregressive models are often utilized in econometric literature along with Bayesian estimation methods. For instance, the readers are referred to a recent research devoted to Bayesian compressed vector autoregression for financial time-series analysis and forecasting in [16,17] and many other works. In this paper, we focus on the tests for evolving market efficiency based on time-varying AR models, only.…”

Multi-step Forecasting with Large Vector Autoregressions

“…This results in a gap between the tremendous number of model parameters and the small number of available observations [45], which further induces difficulty in model generalization. As an addition to the traditional econometric approaches, alternative methods [46]- [48] have been proposed to achieve robust parameter estimation, where vector auto-regressive models [45], [48], [49] suffer from an outof-sample prediction performance in the mid-long horizon. VOLUME 9, 2021 The intrinsic problem related to such application is that directional guidance and timing become more important issues than tracking error [50], [51]; therefore, alternative techniques that can leverage alternative utility functions of the forecast values should be highlighted [52].…”