Supplement to "Inference Based on Structural Vector Autoregressions Identified With Sign and Zero Restrictions: Theory and Applications"

This supplement is organized as follows. Section I shows that using one-sided numerical derivatives can decrease computational time without compromising numerical accuracy. Section II tests for the variance of the importance sampler weights to be finite. Section III provides a step-by-step pseudo-code for using Algorithms 1 and 3 in the paper.


Supplemental Authors: 
Arias, Jonas - Federal Reserve Board
Rubio-Ramirez, Juan - Emory University
Waggoner, Daniel F. - Federal Reserve Bank of Atlanta
Online Appendix