Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re: Antwort: re: Antwort: re: st: xtreg xtregar

From   Christopher Baum <[email protected]>
To   <[email protected]>
Subject   re: Antwort: re: Antwort: re: st: xtreg xtregar
Date   Tue, 4 Jan 2011 15:53:06 -0500

For the simplest model (an AR(1) in y_t), the Nickell bias is of the order - (1+\rho) / (T-1), where \rho is the coefficient on the LDV and T is the number of timeseries observations. 

If the true \rho is 0.5, and T=10 (a long panel by DPD standards!) the bias is -0.167, or 33% of the true coefficient, so that its estimate is seriously attenuated.
The inclusion of additional regressors does not remove this bias; indeed, if the regressors are correlated with the lagged dependent variable to some degree, their coefficients may be seriously biased as well.


Kit Baum   |   Boston College Economics & DIW Berlin   |
                              An Introduction to Stata Programming  |
   An Introduction to Modern Econometrics Using Stata  |

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index