I'm interested in using Stata to forecast data based on a linear
regression, and then calculate standard errors for the resulting
forecasts. If the regression is y = b0 + b1*x + err, then I can
type "regress y x" to estimate the equation, then type "predict
yfore, xb" to get forecast values. I can also type "predict
yforse, stdp", and this, says the online help, calculates the
'standard error of the linear prediction'.
This sounds like what I want, but appears not to be. I am
doing this with Monte Carlo data where the standard deviation
of the err term is 5, and thus the forecast errors should all
be larger than 5. However, the results are all coming back
around 0.5 or 1.
I am guessing that what Stata is calculating here is the standard
error of the fitted value, which is b0hat + b1hat*x, and not
the standard error of the forecast error, which is b0hat-b0 +
(b1hat-b1)*x + err. b0 and b1 are not random, so the difference
between the two is that the forecast error contains err and the
linear prediction doesn't.
My question is whether I can get Stata to calculate the
forecast error. Is there a command to do that? If not, can
I calculate the forecast error by taking the standard error
of the fitted value and adding the standard devation of the
error term (known in this case because it's Monte Carlo)?
Thanks in advance,
Steve Schmidt
*
* 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/
