Advances in the computational methods for models with occasionally binding constraints

This course will present the most recent methods to study models with occasionally binding constraints. The focus will be on models with financial constraints which bind in some state of nature (e.g. recessions) or when variables reach a threshold ( e.g. a capital ratio trigger).

The lectures will be divided in three parts. The first covers piecewise linear methods in the spirit of Kulish and Pagan (2014) and Guerrieri and Iacoviello (2015), the construction of the likelihood function and its use in Bayesian estimation.

The second will consider a regime switching approach where binding state of natures may be recurrent as described e.g. in Liu et ta (2011) or Boccola (2016). The solution and the estimation of the parameters of such models will be performed using the package RISE toolbox.

The last part deals with projection techniques and policy function iterations. Although these methods can be quite complicated, particularly when implemented with all the advanced bells and whistles, their key building blocks are very simple. They can easily deal with nonlinearities and occasionally binding constraints.

A discussion of numerical integration methods will also be provided.