Efficient estimation of Bayesian VARMAs with time‐varying coefficients

Abstract

Summary Empirical work in macroeconometrics has been mostly restricted to using vector autoregressions (VARs), even though there are strong theoretical reasons to consider general vector autoregressive moving averages (VARMAs). A number of articles in the last two decades have conjectured that this is because estimation of VARMAs is perceived to be challenging and proposed various ways to simplify it. Nevertheless, VARMAs continue to be largely dominated by VARs, particularly in terms of developing useful extensions. We address these computational challenges with a Bayesian approach. Specifically, we develop a Gibbs sampler for the basic VARMA, and demonstrate how it can be extended to models with time‐varying vector moving average (VMA) coefficients and stochastic volatility. We illustrate the methodology through a macroeconomic forecasting exercise. We show that in a class of models with stochastic volatility, VARMAs produce better density forecasts than VARs, particularly for short forecast horizons.

Time-varying influence of interest rates on stock returns: evidence from China

“…Moreover, the model adapts to a non-stationary time series. For MCMC sampling, this study follows the practices of Vihola et al (2020) and Chan and Eisenstat (2017), which can quickly and accurately obtain the edge tail qðrjyÞ: The regression coefficient is subsequently simulated with the given standard deviation of edge posterior, and the corresponding joint posterior relationship is obtained: qðr, ðb, cÞjyÞ ¼ qððb, cÞ r, yÞqðrjyÞ: j The mean and variance ofEððb, cÞ r, yÞ j parameters can also be obtained by calculating Varððb, cÞ r, yÞ j and ðb, cÞ with the standard Kalman smoothing.…”

Time-Varying Spillover of us Trade War on the Growth of Emerging Economies

“…Since the in-sample predictability of the TPU index, does not guarantee out-of-sample forecastability, in addition, we estimated various (Bayesian) constant and time-varying parameters VAR and VAR Moving Average (VARMA) models, with and without stochastic volatility as proposed by Chan and Eisenstat (2017). As reported in Table 2, based on an initial in-sample period of 1984:Q3-2000:Q4, the relative (to the random walk model) log predictive likelihoods at horizons of 1-, 2-, and 3-quarterahead confirms the ability of the TPU to forecast the density of the growth of the GDP of emerging economies over a recursively-estimated out-of-sample period (2001:Q1-2019:Q3), 10 with highest gains observed under the constant and time-varying VARMA models with stochastic volatility.…”