-fcast graph- is not doing what Eleftherios Demetriou
<[email protected]> wants.
The problem is that -fcast graph- assumes that there are variables
containing the bounds of the confidence interval (CI), unless the -noci-
option is specified.
Eleftherios should either specify the -noci- option for -fcast graph- or
should request a bootstrapped CI from -fcast compute-
We will look into making -fcast graph- be more accommodating when the CI
variables are not in the dataset.
Here are examples of two solutions for Eleftherios.
The first, and simpler solution, simply adds the -noci- option to
-fcast graph-
. clear all
. webuse lutkepohl
(Quarterly SA West German macro data, Bil DM, from Lutkepohl 1993 Table E.1)
. quietly var dlincome dlconsumption in 1/80, exog(dlinvestment)
. fcast compute f_, step(9)
asymptotic standard error not available with exogenous variables
. fcast graph f_dlincome , observed noci name(noci)
This method will produce a graph of the form Eleftherios requested.
The second method has -fcast compute- produce a bootstrapped estimate
of the CI.
. drop f_*
. fcast compute f_, step(9) bs reps(250)
asymptotic standard error not available with exogenous variables
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(output omitted)
. fcast graph f_dlincome , observed name(withci)
The method will produce a graph including a bootstrapped estimate of
the CI.
--David
[email protected]
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
