Bayesian Analysis of Latent Threshold Dynamic Models

Abstract

Chapter 1 overviews the idea of the latent threshold approach and outlines the dissertation. Chapter 2 introduces the new approach to dynamic sparsity using latent threshold modeling and also discusses Bayesian analysis and computation for model fitting. Chapter 3 describes latent threshold multivariate models for a wide range of applications in the real data analysis that follows. Chapter 4 provides US and Japanese macroeconomic data analysis using latent threshold VAR models. Chapter 5 analyzes time series of foreign currency exchange rates (FX) using latent threshold dynamic factor models. Chapter 6 provides a study of electroencephalographic iv (EEG) time series using latent threshold factor process models. Chapter 7 develops a new framework of dynamic network modeling for multivariate time series using the latent threshold approach. Finally, Chapter 8 concludes the dissertation with open questions and future works.

Identifying conventional and unconventional monetary policy shocks: a latent threshold approach

“…Data-driven inferences on thresholds and time-varying sparsity patterns induce a dynamic variable selection; i.e., the LTM structure explores the best set of variables at each time point. This approach to reducing the dimension of parameters has the great benefit of improving forecast performance and facilitating model interpretation, as shown by Nakajima and West (2013). Here, we apply and extend the LTM approach to the time-varying parameters of simultaneous relations in order to identify monetary policy shocks and to take account of the ZLB in conventional and unconventional policy regimes.…”

“…The time-varying parameters are usually modeled as following a random walk process, which aims at allowing for the possibility of permanent shifts and reducing the number of parameters, as discussed by Primiceri (2005). We instead introduce the stationary process, because a highly-persistent AR(1) process with close to one can generate a parameter trajectory of seemingly permanent shifts in a finite sample, and because the stationarity of the latent process yields more appropriate estimates of the simultaneous relations in the latent threshold modeling introduced below (Nakajima and West 2013).…”

“…To overcome this difficulty, we exploit the latent threshold model (LTM) developed by Nakajima and West (2013). The LTM can be effectively applied in cases when a parameter process such as the time-varying parameters in a VAR may be close to zero and practically insignificant in some periods.…”

“…The purpose underlying the LTM approach is to induce parsimony in modeling, which simplifies high-dimensional time-varying parameters in time series analysis such as the TVP-VAR models. By inducing temporal sparsity in time-varying parameters, the LTM approach helps to reduce estimation uncertainty and improve forecasting performance (see Nakajima and West 2013). This paper utilizes the approach for regime-switching constraints on time-varying parameters associated with the simultaneous effects of structural shocks to identify conventional and unconventional monetary policy shocks.…”

“…3 Specifically, we incorporate a regime change between conventional and unconventional policies using the "latent threshold model (LTM)" recently proposed by Nakajima and West (2013).…”