robust2sls 0.2.3

Bug Fixes

• update documentation
• re-create binaries with newer R version to address CRAN NOTEs

Minor Fixes

robust2sls 0.2.2

Minor changes

• move package doRNG from Imports to Suggests
  because the package has been orphaned as of 2022-12-19. Only affects the function
  mc_grid(), which is anyway for advanced users only

Bug Fixes

• fix bug in unit testing

IIS Functionality

robust2sls 0.2.1

Major changes

• addition of Impulse Indicator Saturation (IIS) as initial estimator for outlier detection (from ivgets package)

Minor changes

• code coverage has been extended to 100%
• update formulae for FODR under normality (estimation not required)
• all 2SLS estimation commands originally from AER package now use the ivreg package
• fixed \eqn or \deqn or with empty second argument, as suggested by CRAN maintainers

Outlier Testing

robust2sls 0.2.0

Major changes

Introduction of a suite of new functions implementing several tests for the
presence of outliers in the sample.

• proportion test
• count test
• global test using Simes (1986) procedure
• scaling sum test
• scaling sup test

A vignette is included that illustrates the usage of these testing functions.

Minor changes

• change gauge_avar() to accept NULL as input for split argument
• implement asymptotic covariance of the FODR for different cut-offs
• utility function multi_cutoff() that allows to apply the same outlier detection algorithm with different cut-off / gamma values

First release of robust2sls package

The robust2sls package

The goal of robust2sls is to provide easy-to-use tools for outlier-robust inference and outlier testing in two-stage least squares (2SLS) models.

For a detailed introduction to the model framework, the different trimmed 2SLS algorithms, and examples, see the vignette Introduction to the robust2sls Package.

utils::vignette("overview", package = "robust2sls")

Warning

This release has been tested extensively but there could still be errors in the code, especially for the functionality relating to bootstrap standard errors. The next version will improve this.