Statalist


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

st: serial correlation test in fe model


From   Daniel Borowczyk Martins <stata.danielbm@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: serial correlation test in fe model
Date   Wed, 30 Sep 2009 13:39:34 +0100

Dear Stata Users,

I am trying to implement a test for serial correlation in the
idiosyncratic errors in a fixed effects model.

I am following Wooldridge's (2002: 275) first test of this type.

I will follow his notation so that:

- u are the idiosyncratic errors of the FE model;
- ü are the time-demeaned errors
- û are the residuals of the time-demeaned variables regression

His description of the test is the following:

1) regress û on L1.û  for T and T-1 (the last two periods)

2)) use the regression coefficient of L1.û (beta) and its standard
error (sigma) to obtain a sample estimate of delta= Corr(ü, L1.ü)

3) test h0: delta= -1/(T-1), where the t-statistic is asymptotically normal.

I face two problems.

First, what is the algebraic expression relating Corr(ü, L1.ü) to beta
and sigma?

Second, how can I formulate it using the command testnl?

Thanks in advance for your help.

Daniel

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index