Are Nonlinear Methods Necessary at the Zero Lower Bound?

26 Pages Posted: 7 Aug 2016 Last revised: 13 Nov 2016

Multiple version iconThere are 2 versions of this paper

Date Written: November 11, 2016


This paper examines the importance of using nonlinear methods to account for the zero lower bound (ZLB) on the Fed's policy rate. We estimate three models with a particle filter: (1) a nonlinear model with a ZLB constraint; (2) a constrained linear model that imposes the constraint in the filter but not the solution; and (3) an unconstrained linear model that never imposes the constraint. The linear models have a lower likelihood than the nonlinear model when the Fed is constrained and predict large monetary policy shocks during the ZLB period. We also compare the predictions from our nonlinear model to the quasi-linear solution with OccBin. OccBin captures the ZLB much better than the linear solutions but it still generates less endogenous volatility than the nonlinear model and it is not as conducive to estimation. Finally, we extend the baseline model to include a banking sector. We find larger differences between the predictions from the nonlinear model and both the linear and quasi-linear models.

Keywords: Bayesian Estimation, Model Comparison, Zero Lower Bound, Particle Filter

JEL Classification: C11, E43, E58

Suggested Citation

Richter, Alexander W. and Throckmorton, Nathaniel A., Are Nonlinear Methods Necessary at the Zero Lower Bound? (November 11, 2016). Available at SSRN: or

Alexander W. Richter (Contact Author)

Federal Reserve Bank of Dallas ( email )

2200 North Pearl Street
PO Box 655906
Dallas, TX 75265-5906
United States
214-922-5360 (Phone)


Nathaniel A. Throckmorton

William & Mary ( email )

Economics Department
P.O. Box 8795
Williamsburg, VA 23187
United States


Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
PlumX Metrics