Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: =?utf-8?Q?st:_test_of_nonlinear_combination=E2=80=8F=E2=80=8F=E2=80=8F_in?==?utf-8?Q?_MATRIX_form?=


From   tanaka yoshi <santouka@live.jp>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: =?utf-8?Q?st:_test_of_nonlinear_combination=E2=80=8F=E2=80=8F=E2=80=8F_in?==?utf-8?Q?_MATRIX_form?=
Date   Thu, 27 Dec 2007 22:07:31 +0900

Dear Statalist 

I want to test nonlinear combination of estimators. i know the command for it is "nlcom," but i am in trouble because "nlcom" is not available in my case because the expressions which follow "nlcom" in the syntax are too long.

my method is as follows:

First,  I estimated the following system equations by three stage least squares.

A(t)=aX(t)+cA(t-1)////////////////////////eq. (1) 

       A(t):      7*1 dependent variables in period t
       a:         7*24 parameters
       X(t):     24*1 independent variables in period t
       c:         7*7 parameters
       A(t-1):  7*1 independent variables in period (t-1)

eq.(1) indicates the SHORT run effects, and i can test "a" and "c." this is easy.

Second,  I want to obtain the LONG run effects.
assuming A(t)=A(t-1), eq.(1) becomes

A(t)=inv(I-c)aX(t)///////////////////////// eq.(2)

Finally,  I want to test each element of "inv(I-c)a" using the estimation results of eq.(1).
each element of "inv(I-c)a" is nonlinear combination of estimators obtaind in eq.(1).
so generally, it is normal to use "nlcom," but the expression of each element in "inv(I-c)a" is too long to be available in "nlcom."

could anyone tell me how to test each element of inv(I-c)a???

yoshi
_________________________________________________________________
Hotmailがお届けする、幸せになるためのメールマガジン「ビジネス幸福論」実施中
http://go.windowslive.jp/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index