## Programs*
### 1. Testing for mis-specification in probit
and tobit: Conditional Moment Tests
Probit and tobit
regressions are based on the assumption that the residual in the
latent variable equation is normally distributed and has constant
variance. If these assumptions are not fulfilled, the parameters in
the latent variable equation will be inconsistent, and marginal
effects may be misleading. In applied work it may therefore be
important to investigate if the underlying assumptions are supported
by the data. One way of doing this is to use a conditional moment
(CM) test (Newey, 1985; Tauchen, 1985). Click here for a zipped file containing the
following files:
cmtest.ado: a STATA
ado file that can be used to compute conditional moment tests for
tobit and probit - put this in C:\ado;
manual1.pdf: a first
draft of a manual;
simcmt1.do: a STATA
do-file that simulates some data and illustrates how the conditional
moment tests are implemented;
run_port.do: a STATA
do-file that uses real data to estimate the model reported in the
manual;
martins.dta: the STATA
datafile used in run_port.do
###
### 2. Imposing common factor restrictions for AR(1)
model "ex post" using a minimum distance
procedure
Blundell and Bond ("GMM estimation
with persistent panel data: An application to production functions")
consider a panel data model of the following type:
y(it) = b*x(it) +
e(it)
where
e(it) = rho*e(i,t-1) +
v(it)
This can be re-written
as
y(it) = b*x(it) -
rho*b*x(i,t-1) + rho*y(i,t-1) + v(it)
or
y(it) =
a1*x(it) +a2*x(i,t-1) + a3*y(i,t-1) + v(it)
Suppose you estimate
the latter equation using some suitable linear estimator. My STATA
program md_ar1, can then be used to impose, and test the validity
of, the common factor restrictions ex post, based on a minimum
distance procedure. Click here to
download a zipped file containing a STATA ado file
called md_ar1 (put this in C:\ado), a file containing basic
instructions on how to use the program, and the dataset used by
Blundell & Bond to play with.
###
* Note: ALL PROGRAMS ARE
PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE
RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAMS IS WITH YOU.
SHOULD THE PROGRAM(S) PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL
NECESSARY SERVICING, REPAIR OR CORRECTION. IN NO EVENT WILL I OR ANY OTHER
PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAMS, BE LIABLE TO
YOU FOR DAMAGES.
### |