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: st: Spatial regression with non-symmetric weights

From   Austin Nichols <>
Subject   Re: st: Spatial regression with non-symmetric weights
Date   Sat, 30 Jul 2011 09:01:55 -0400

Start with David Drukker's slides on spatial
autoregressive models with Stata:

and then
findit sppack
for e.g.

The working papers that document the commands are available at,, and .

On Sat, Jul 30, 2011 at 3:45 AM,  <> wrote:
> Dear Statalist,
> I am trying to run a spatial lag model (and spatial dependence tests in
> the base OLS regression) using a non-symmetric weights matrix (based on
> commuting flows). However, --spatwmat-- refuses to load the matrix
> because it is non-symmetric. I have tried to resolve this by deriving
> the eigenvalues of my weights matrix separately (using --geneigen-- on
> the basis of the discussion in
>, but Stata
> freezes when I run the command (I have a dataset of 376 observations, so
> the full matrix has 141376 cells). Further, I don't know if the
> eigenvalues derived from geneigen are suitable for use with spatreg,
> spatgsa, etc.
> Can anybody help?
> -       Is there some other package or set of commands that allows
> spatial analysis in Stata for non-symmetric weights matrices? (I have
> looked at the --spwmatrix-- command, which presumably works with
> non-symmetric matrices, but I haven't figured out how to use my
> commuting data in the context of this command...)
> -       Is the --geneigen-- route a valid way to overcome this?
> -       If so, any ideas as to what I should be looking at in order to
> make --geneigen-- run on my PC?
> Many thanks in advance!
> Vassilis

*   For searches and help try:

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