Bayesian Local Projections

Abstract We propose a Bayesian approach to Local Projections that optimally addresses the empirical bias-variance trade-off intrinsic in choice between direct and iterative methods. (BLP) regularise LP regressions via informative priors, estimate impulse response functions capture properties of data more accurately than VARs. BLPs preserve flexibility LPs while retaining degree estimation uncertainty comparable VARs with standard macroeconomic priors. As regularised forecasts, are also valuable alternative BVARs for multivariate out-of-sample projections.