vcovPC {pcse} | R Documentation |
The package pcse contains a function to estimate panel-corrected standard errors. Data may contain balanced or unbalanced panels. This function extracts the resulting variance covariance matrix.
vcovPC(x, ...) ## Default S3 method: vcovPC(x, groupN, groupT, pairwise=FALSE, ...)
x |
A lm object containing the initial run of OLS. |
groupN |
A vector containing the cross-sectional group identifier for each observation. |
groupT |
A vector containing the time identifier for each observation. |
pairwise |
An optional logical flag indicating whether the X's used to estimate the "middle" matrix should be chosen in a pairwise fashion or casewise fashion. If pairwise, the correlation between observations $i$ and $j$ is based on the time periods common to $i$ and $j$. If casewise, the correlation between observations i and j is based on the largest rectangular subset of the data, i.e., $T_i$ = $T_j$ = $T^*$ for all $i$ and $j$ if casewise is selected. |
... |
Further arguments passed to methods. |
Delia Bailey <delia.bailey@gmail.com> and Jonathan N. Katz <jkatz@caltech.edu> Maintainer: Delia Bailey <delia.bailey@gmail.com>
## see demo file.