Dear Stata users -
I have been working with synthetic cohort for some time now, and
would need to use estimators that correct for heteroskedasticity that
is caused by such aggregation. One of the models I am considering is
the Errors-in-Variables Estimator (EVE, consistent when the number of
cohorts is taken to infinity), which is essentially a Weighted Least
Squares (WLS) estimator with a correction: one adjusts by subtracting
the variance-covariance matrix from the two components of the estimator.
You have my apologies for the vagueness of this description - I
cannot think of a way to include the equation defining the estimator
in an e-mail that will come through for users who might be employing
a text-based client...I don't know if this listhost accepts
attachments. The estimator may be found in Paul Devereux's paper
(2003) "Improved Errors-in-Variables Estimators for Grouped Data",
page 14, Equation 50.
Is there any existing way to handle EVE or Generalized EVE models in
Stata? (I am using version 9.)
Also, the paper proved the equivalence of Jackkife Instrumental
Variables Estimator (JIVE) - if there is a way to handle this, I will
be able to modify the unbiased EVE as necessary.
Any help in this matter would be greatly appreciated. Also - if there
is no existing way to handle this - what estimator would provide a
good starting point for source code that I can modify?
Thank you very much for your time.
Cordially,
Manasi
PS: I might have sent this very same message to the listhost from a
different e-mail address (I was not registered for it under that e-
mail address). It did not go through, probably because I was not
registered. I have waited a decent amount of time before re-sending
this question, and in the unlikely scenario that you do receive it
twice, allow me to apologize for the excess mail and consequent
annoyance.
*
* 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/
