This file gives some basic information about the following seven Stata
commands:
har (in ado file har.ado)
kernel_F_all (in ado file kernel_F_all.ado)
kernel_t (in ado file kernel_t.ado)
OS_F_all (in ado file OS_F_all.ado)
OS_t (in ado file OS_t.ado)
har_tab (in ado file har_tab.ado)
hart (in ado file hart.ado)
har.ado is the main procedure, and the other are the sub procedures
(kernel_F_all.ado; kernel_t.ado; OS_F_all.ado; OS_t.ado; har_tab.ado) which are
used by the har command.
The program "har_tab.ado" tabulates the estimation results including coef,
standard error, t-statistics, df, p-value, and confidence region.
The program "kernel_F_all.ado" computes the data-driven testing-optimal b
(testing-optimal truncation lag) and F statistic (the Wald test version) for
testing the overall significance (the joint significance of all coefficients
except the intercept),
and the associated approximate degrees of freedom. See Sun (2014). This program
works only for the case with a kernel HAR variance estimator.
The program "kernel_t.ado" computes the data-driven optimal b and the t
statistic for testing a single restriction in Sun (2014), the associated
adjusted standard error, and the approximate degrees of freedom.
This program works only for the case with a kernel HAR variance estimator.
The program "OS_F_all.ado" computes the data-driven testing-optimal K
(testing-optimal number of series bases) and F statistic (the Wald test version)
for testing the overall significance (the joint significance of all coefficients
except the intercept). See Sun (2013). This program works only for the
case with an orthonormal series HAR variance estimator.
The program "OS_t.ado" computes the data-driven optimal K and the t statistic
for testing a single restriction in Sun (2013), and the associated adjusted
standard error. This program works only for the case with an orthonormal series
HAR variance estimator.
hart.ado performs Wald tests of simple and composite linear hypotheses about the
parameters in the most recently estimated har model.
The following four types of kernel or series function (three kernels and one
orthonormal series function can be used)
Sun (2014): Bartlett, Parzen, and Quadratic Spectral.
Sun (2013): Orthonormal Series.
out1.do-out5.do correspond to case1-case5 in the paper. htest1.do-htest3.do
correspond to case8-case10 in the paper.
Place all of the .ado files in a directory where Stata can find them.
Note:
kbomegaF.ado and kubomegaF.ado use mm_root(), which is an external mata command
that should be downloaded before using the programs here.
If an older version of -moremata- is already installed on your system, it may be
advisable to remove it typing
. ado uninstall moremata
Then type
. ssc install moremata, replace