semiBRM-package: Semiparametric Binary Response Models in R

In henrykye/semiBRM: Semiparametric Binary Response Models in R

Description

Implementation of semiparametric binary response models theorized in \insertCiteklein1993efficient;textualsemiBRM.

Details

This package offers an implementation of semiparametric binary response models, theorized in a seminal work in the semiparametric econometrics literature, \insertCiteklein1993efficient;textualsemiBRM.

Compared to other related packages that help run non-/semi-parametric analysis, this reflects the econometrician's perspective to the best, taking conditions for asymptotic properties seriously. For instance, the default choice of bandwidth size in the Nadaraya-Watson estimator meets the conditions for square-root N consistency of the coefficient estimator.

In turn, this package will be useful in conducing econometric analysis on binary response models. For example, this package offers computation of marginal effects, which are often the most important quantity of interest in binary response models in the economics context.

This is built upon Rcpp \insertCiteeddelbuettel2011rcppsemiBRM along with OpenMP, speeding up computation of nonparametric conditional expectation over data points. In author's opinion, the only significant disadvantage of this semiparametric approach to binary choice models over 'standard' ones such as Probit and Logistic, particularly in the cross-sectional setup, is high computation costs. In this package, this limitation is meaningfully overcome by taking advantage of multithreading via OpenMP in Rcpp.

The econometric theory underlying this package mostly comes from \insertCiteklein1993efficient;textualsemiBRM. However, a few important parts are based on lectures of Prof. Klein at Rutgers University who wrote that paper. Since \insertCiteklein1993efficient;textualsemiBRM was published, he made several improvements in asymptotic theories. For example, the 'current version' of asymptotics no longer requires the use of 'higher-order' kernel in the Nadaraya-Watson estimator, which was asked in the original paper.

Author(s)

Hyungil Kye (Henry) <[e-mail]>

References

\insertAllCited

henrykye/semiBRM documentation built on Dec. 20, 2021, 3:49 p.m.