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

# Re: st: Symmetry Test

 From "Brian P. Poi" To statalist@hsphsun2.harvard.edu Subject Re: st: Symmetry Test Date Thu, 12 May 2011 20:34:03 -0400

```Brian P. Poi

On 05/12/2011 11:18 AM, Ana wrote:
```
```Dear Alex (and Statalisters),

I will try to make myself clear know.

My plan is to estimate a Seven Demand Equations using Quiads Model
(based on nlsurquaids.ado routine by Brian Poi (2008))
My main goal is to test one hypothesis based on Browning&  Chiappori
model (reference: Efficient Intra-Household Allocations: A general
Characterization and Empirical Tests, Econometrica, vol.66, no.6
(Nov.,1998), pp 1241-1278).

Based on the nslurquaids.ado routine (Brian Poi (2008)), without the
symmetry restriction and based on the originally demand for four goods
(and three share equations) generated by the author, I am trying to
implement this Browning and Chiappori test. If I am well suceeded, I
will extend this routine to "my" model (with more goods...).

Based on the following routine of the nlsurquaids.ado,

nlsur quaids @ w1 w2 w3 lnp1-lnp4 lnm, ifgnls nequations(3) param(a1
a2 a3 b1 b2 b3 g11 g12 g13 g21 g22 g23 g31 g32 g33 l1 l2 l3) nolog

the test I would like to carry out is something like:

(g31 - g13) = (g12 - g21)*(g23 - g32)&  (g21 - g12) = (g31 - g13)*(g23 - g32)

The problem is that when I execute the command:

testnl (([eq1]_b[/g31] - [eq1]_b[/g13]) = ...*[eq2]_b[/g23] -
[eq3]_b[/g32])) (([eq2]_b[/g21] - [eq1]_b[/g12]) = ...*[eq2]_b[/g23] -
[eq3]_b[/g32]))

```
```
```
With -testnl- and -nlsur-, you don't need to use equation names with your parameters:
```
. sysuse auto
. nlsur (mpg = {b0} + {b1}*gear) (head = {c0} + {c1}*weight)
. testnl (_b[/c1] / _b[/b1] = 1) (_b[/b0] / _b[/c0] = 4)

```
-nlsur- allows the same parameter to appear in multiple equations (as is the case with AIDS models, for example), so it doesn't make much sense to think of a parameter as belonging to any particular equation.
```
-- Brian Poi
-- brian@poiholdings.com
*
*   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/
```