Bayesian Econometric Methods for Central Bank Policy Analysis: Time-Varying Parameter VAR, Stochastic Volatility, and Prior Specification for Monetary Policy Transmission

Abstract

Bayesian econometric methods have become central to modern central bank policy analysis, providing tractable methods for high-dimensional inference, principled treatment of parameter uncertainty, and flexible modelling of time-varying dynamics. This paper presents a comprehensive empirical analysis using time-varying parameter vector autoregression (TVP-VAR) with stochastic volatility, applied to UK monetary policy transmission over 1990-2024. The analysis characterizes the evolution of monetary policy transmission through interest rate, exchange rate, and asset price channels; identifies structural breaks associated with major policy regime changes; and evaluates the sensitivity of policy analysis to prior specification. Empirical results indicate significant time variation in transmission mechanisms, with amplitude of interest rate pass-through to inflation declining by approximately 40 per cent between 1990-1999 and 2010-2024. Bayesian model comparison identifies stochastic volatility as an essential feature of the model, and prior sensitivity analysis reveals that policy inferences are robust to reasonable variations in prior specification but sensitive to the choice of identifying assumptions for structural shocks. The paper contributes empirical evidence on UK monetary transmission dynamics, methodological guidance on prior specification and identification, and structured discussion of Bayesian methods in central bank policy analysis contexts.