With thanks to Kit Baum, a new version of ranktest by Kleibergen-Schaffer-Windmeijer, version 2.0.03, is now available on SSC Archives.

Tests of rank have various practical applications; in econometrics probably the most common is the test of the requirement in a linear IV/GMM model that the matrix E(z_i x_i') is full rank, where z_i is the vector of instruments and x_i is the vector of endogenous regressors.

The updates to ranktest are extensive and include the addition of GMM-based J-type rank tests as proposed by Windmeijer (2018); see https://ideas.repec.org/p/bri/uobdis/18-696.html.

ranktest is a required component for ivreg2 and related packages. Please note, however, that the new features of ranktest require Stata 13 or higher. If called under version control or by an earlier version of Stata, ranktest will call a new program also included in the package, ranktest11. ranktest11 is essentially the previous version of ranktest version 1.4.01. Because ivreg2 runs under version control, the results reported by the current version of ivreg2 will be unaffected by this update. Similar remarks apply to related packages.