Staff working papers

Differentiable, Filter Free Bayesian Estimation of DSGE Models Using Mixture Density Networks

Staff Working Paper 2025-3 Chris Naubert
I develop a method for Bayesian estimation of globally solved, non-linear macroeconomic models. The method uses a mixture density network to approximate the initial state distribution. The mixture density network results in more reliable posterior inference compared with the case when the initial states are set to their steady-state values.

See More