
Fast and Flexible Bayesian Inference in Timevarying Parameter Regression Models
In this paper, we write the timevarying parameter regression model invo...
read it

General Bayesian timevarying parameter VARs for predicting government bond yields
Timevarying parameter (TVP) regressions commonly assume that timevaria...
read it

Machine learning based iterative learning control for nonrepetitive timevarying systems
The repetitive tracking task for timevarying systems (TVSs) with nonre...
read it

Variational DualTree Framework for LargeScale Transition Matrix Approximation
In recent years, nonparametric methods utilizing random walks on graphs...
read it

Efficient Sampling from TimeVarying LogConcave Distributions
We propose a computationally efficient random walk on a convex body whic...
read it

Branching random walk solutions to the Wigner equation
The stochastic solutions to the Wigner equation, which explain the nonlo...
read it
The Macroeconomy as a Random Forest
Over the last decades, an impressive amount of nonlinearities have been proposed to reconcile reducedform macroeconomic models with the data. Many of them boil down to have linear regression coefficients evolving through time: threshold/switching/smoothtransition regression; structural breaks and random walk timevarying parameters. While all of these schemes are reasonably plausible in isolation, I argue that those are much more in agreement with the data if they are combined. To this end, I propose Macroeconomic Random Forests, which adapts the canonical Machine Learning (ML) algorithm to the problem of flexibly modeling evolving parameters in a linear macro equation. The approach exhibits clear forecasting gains over a wide range of alternatives and successfully predicts the drastic 2008 rise in unemployment. The obtained generalized timevarying parameters (GTVPs) are shown to behave differently compared to random walk coefficients by adapting nicely to the problem at hand, whether it is regimeswitching behavior or longrun structural change. By dividing the typical ML interpretation burden into looking at each TVP separately, I find that the resulting forecasts are, in fact, quite interpretable. An application to the US Phillips curve reveals it is probably not flattening the way you think.
READ FULL TEXT
Comments
There are no comments yet.