Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down at the end of May, and its replacement, **statalist.org** is already up and running.

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

From |
Christopher Baum <kit.baum@bc.edu> |

To |
"statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |

Subject |
Re: RE: st: RE: mvreg with vce(robust)? |

Date |
Tue, 13 Jul 2010 06:52:46 -0400 |

<> On Jul 13, 2010, at 2:33 AM, Peter wrote: > > The issue is as follows: -mvreg- runs much faster on an 8-core server than a loop of -reg- while generating the same results in my case (balanced panel). I assume that is due to the superior parallelization of the Mata code used in -mvreg-. The only difference is that -mvreg- cannot compute heteroskedasticity robust standard errors. The reason that -mvreg- can run faster than a loop over -regress- is that -mvreg- forms the (X'X)^-1 matrix once and then postmultiplies it by X'Y, where Y is a matrix rather than a vector, yielding a corresponding matrix of b coefficients. It would be possible to 'roll your own' robust SEs, given the X matrix and the residuals from each equation estimated by -mvreg-, but in the interests of accuracy (you know that Stata computes these right, whereas your code might or might not do so) I would recommend looping over equations. Kit Baum | Boston College Economics & DIW Berlin | http://ideas.repec.org/e/pba1.html An Introduction to Stata Programming | http://www.stata-press.com/books/isp.html An Introduction to Modern Econometrics Using Stata | http://www.stata-press.com/books/imeus.html * * 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/

- Prev by Date:
**st: Re: Identify and Replace Values** - Next by Date:
**Re: st: Dummy variables** - Previous by thread:
**st: Dummy variables** - Next by thread:
**st: Least square regression involving time lags in the parameters** - Index(es):