Efficient Bayesian generalized linear models with time-varying coefficients: The walker package in R

TL;DR

The paper introduces the walker R package that enables Bayesian generalized linear models with time-varying coefficients, facilitating modeling of dynamic effects such as policy interventions using efficient Hamiltonian Monte Carlo algorithms.

Contribution

The paper presents a novel R package extending Bayesian GLMs to include time-varying effects with efficient inference via Hamiltonian Monte Carlo and state space modeling.

Findings

Efficient Bayesian inference for models with time-varying coefficients.

Application to policy intervention effects demonstrates model flexibility.

State space representation improves sampling efficiency.

Abstract

The R package walker extends standard Bayesian general linear models to the case where the effects of the explanatory variables can vary in time. This allows, for example, to model the effects of interventions such as changes in tax policy which gradually increases their effect over time. The Markov chain Monte Carlo algorithms powering the Bayesian inference are based on Hamiltonian Monte Carlo provided by Stan software, using a state space representation of the model to marginalise over the regression coefficients for efficient low-dimensional sampling.