Clive,
Try adding the -robust- option to ivreg2. I think the problem is
that you are requesting from ivreg2 standard errors that are robust
to autocorrelation but not heteroskedasticity. ivreg2 can provide
standard errors with and without heteroskedasticity-robustness, but
newey and newey2 standard errors are always the robust type.
Cheers,
Mark
Date sent: Thu, 13 Jan 2005 03:56:13 -0000 (GMT)
Subject: Re: st: Re: -newey2- vs. -ivreg2-
From: "Clive Nicholas" <[email protected]>
To: [email protected]
Send reply to: [email protected]
> Kit Baum replied to Mark Schaffer:
>
> > Indeed it can be confusing! Mark is absolutely right--you should ADD
> > one to bw(), which starts counting from zero, to match lags(). Sorry
> > for the added confusion!
>
> I don't know if Jonathan A. Schwabish, who sent the original post, is now
> satisfied, but I'm a bit puzzled. This is what I get when using Garrett
> and Mitchell's (2001) infamous dataset, applying Kit and Mark's "for N
> lags, add one to the bandwidth" formula:
>
> . newey2 growthpc trade fdi unem left spend captax labtax, lag(1)
>
> Regression with Newey-West standard errors Number of obs = 352
> maximum lag : 1 F( 7, 344) = 4.89
> Prob > F = 0.0000
> ----------------------------------------------------------------------------
> | Newey-West
> growthpc | Coef. Std. Err. t P>|t| [95% Conf. Interval]
> -----------+----------------------------------------------------------------
> trade | .0051762 .0063955 0.81 0.419 -.007403 .0177555
> fdi | .1112711 .1114347 1.00 0.319 -.107908 .3304503
> unem | -.0917037 .0489144 -1.87 0.062 -.1879127 .0045053
> left | -.0011842 .0035009 -0.34 0.735 -.0080701 .0057017
> spend | -.0295982 .0455134 -0.65 0.516 -.1191177 .0599213
> captax | -.0620164 .0155124 -4.00 0.000 -.0925275 -.0315053
> labtax | -.0282463 .0348024 -0.81 0.418 -.0966986 .0402061
> _cons | 6.832187 .9175741 7.45 0.000 5.027425 8.636949
> ----------------------------------------------------------------------------
>
> . ivreg2 growthpc trade fdi unem left spend captax labtax, bw(2) small
>
> Ordinary Least Squares (OLS) regression
> ---------------------------------------
> Autocorrelation-consistent statistics
> kernel=Bartlett; bandwidth=2
> time variable (t): year
> group variable (i): nation
> Number of obs = 352
> F( 7, 344) = 7.30
> Prob > F = 0.0000
> Total (centered) SS = 2042.961127 Centered R2 = 0.1708
> Total (uncentered) SS = 4004.310028 Uncentered R2 = 0.5769
> Residual SS = 1694.045908 Root MSE = 2.2
> ----------------------------------------------------------------------------
> growthpc | Coef. Std. Err. t P>|t| [95% Conf. Interval]
> -----------+----------------------------------------------------------------
> trade | .0051762 .0073358 0.71 0.481 -.0092525 .019605
> fdi | .1112711 .1040054 1.07 0.285 -.0932954 .3158377
> unem | -.0917037 .0581893 -1.58 0.116 -.2061553 .022748
> left | -.0011842 .0041281 -0.29 0.774 -.0093037 .0069353
> spend | -.0295982 .0382442 -0.77 0.440 -.10482 .0456236
> captax | -.0620164 .0160886 -3.85 0.000 -.0936608 -.030372
> labtax | -.0282463 .0339707 -0.83 0.406 -.0950628 .0385703
> _cons | 6.832187 .7036352 9.71 0.000 5.448218 8.216156
> ----------------------------------------------------------------------------
>
> The coefficients are the same, but the standard errors most certainly
> aren't. What's up here? Or, alternatively, what's up with me?
>
> CLIVE NICHOLAS |t: 0(044)7903 397793
> Politics |e: [email protected]
> Newcastle University |http://www.ncl.ac.uk/geps
>
> References:
>
> Garrett G and Mitchell D (2001) "Globalization, Government Spending and
> Taxation in the OECD", EUR J POLIT RES 39(1): 145-77.
>
> *
> * 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/
Prof. Mark E. Schaffer
Director
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS UK
44-131-451-3494 direct
44-131-451-3296 fax
44-131-451-3485 CERT administrator
http://www.sml.hw.ac.uk/cert
*
* 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/
