[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: How do I specify linear restrictions and AR(1) in xtivreg or xtivreg2
> -----Original Message-----
> From: email@example.com
> [mailto:firstname.lastname@example.org] On Behalf Of
> David Granlund
> Sent: 21 April 2006 13:30
> To: email@example.com
> Subject: st: How do I specify linear restrictions and AR(1)
> in xtivreg or xtivreg2
> Is it possible to specify linear constraints and/or to allow
> the disturbance term to be first-order autoregressive when
> estimating panel-data models with some endogenous
> right-hand-side covariates? I would like to specify
> restrictions that some parameters are product of other
> parameters, for example that B3 = B1 B2.
This is a nonlinear constraint, no?
Nick Cox's answer to this question two years ago probably still stands:
> In short, in general, and as a guess:
> Stata doesn't lend itself to any kind of problem involving
> non-linear constraints unless you are willing to do all
> the programming yourself.
‑xtivreg2‑ will, however, accommodate linear constraints if the estimation uses the ‑cue‑ or continuously-updated GMM option. The CUE estimator is implemented by a call to ‑ml‑, and you can pass options to ‑ml‑ with the ‑xtivreg2‑ option ‑cueopt‑. If you do this, be sure you have the latest version of ‑ivreg2‑ installed, 2.1.15; earlier versions had a bug in the implementation of the CUE estimator.
As for AR(1) errors, ‑xtivreg2‑ give you two possibilites, neither of which is exactly what you want. First, you can use the ‑cluster‑ option, which allows for arbitrary within-panel autocorrelation. Second, you can use the ‑bw‑ option for "Newey-West" and related kernel-based robust covariance estimation.
> Thank you in advance for your help!
> David Granlund
> * 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/
* For searches and help try: