Follow us
Publications Commentary Research People Events News Resources and Videos About IFS
Home Publications bootwildct


Bansi Malde and Molly Scott

This command implements the wild-cluster bootstrap-t procedure, with a specified null hypothesis, as described in Cameron et al (2007). This procedure is shown to improve inference in cases with few clusters.

It is a post-estimation command, which works for linear models with clustered standard errors and with simple hypotheses only. The command should work with versions of stata above 10.1. Note that varlist should include ALL the right hand side variables included in the linear model for which one estimates these t-statistics.

The zip folder contains an ado file and a help file.