Trend Agnostic One-Step Estimation of DSGE Models

DSGE models are currently estimated with a two-step approach: the data is first transformed and then DSGE structural parameters are estimated. Two-step procedures have problems, ranging from component misspecification to incorrect assumptions about the correlation between cyclical and non-cyclical components. In this paper, I present a one-step method, where DSGE structural parameters are jointly estimated with filtering parameters. First, I illustrate the properties of the one-step procedures using simulated data. Then, I show that different data transformations imply different structural estimates and that two-step approaches lack a statistical-based criterion to select amongst them. The one-step approach allows to choose the most likely specification of the non-cyclical component for individual series and/or to construct robust estimates by Bayesian averaging. The role of the investment specific shock as source of GDP volatility is reconsidered.