Wald-style Chi-square test and F test of slope coefficients being zero jointly, including robust versions of the tests.
pwaldtest(x, ...)
# S3 method for class 'plm'
pwaldtest(
x,
test = c("Chisq", "F"),
vcov = NULL,
df2adj = (test == "F" && !is.null(vcov) && missing(.df2)),
.df1,
.df2,
...
)
# S3 method for class 'pvcm'
pwaldtest(x, ...)
# S3 method for class 'pgmm'
pwaldtest(x, param = c("coef", "time", "all"), vcov = NULL, ...)an estimated model of which the coefficients should be
tested (usually of class "plm"/"pvcm"/"pgmm")`,
further arguments (currently none).
a character, indicating the test to be performed, may
be either "Chisq" or "F" for the Wald-style
Chi-square test or F test, respectively,
NULL by default; a matrix giving a
variance–covariance matrix or a function which computes such;
if supplied (non NULL), the test is carried out using
the variance–covariance matrix indicated resulting in a robust
test,
logical, only relevant for test = "F",
indicating whether the adjustment for clustered standard errors
for the second degrees of freedom parameter should be performed
(see Details, also for further requirements regarding
the variance–covariance matrix in vcov for the
adjustment to be performed),
a numeric, used if one wants to overwrite the first degrees of freedom parameter in the performed test (usually not used),
a numeric, used if one wants to overwrite the second degrees of freedom parameter for the F test (usually not used),
(for pgmm method only): select the parameters to be tested:
"coef", "time", or `"all"“.
An object of class "htest", except for pvcm's within model for which
a data.frame with results of the Wald chi-square tests and F tests per
regression is returned.
pwaldtest can be used stand–alone with a plm object, a pvcm object,
and a pgmm object (for pvcm objects only the 'random' type is valid and no
further arguments are processed; for pgmm objects only arguments param
and vcov are valid). It is also used in
summary.plm() to produce the F statistic and the Chi-square
statistic for the joint test of coefficients and in summary.pgmm().
pwaldtest performs the test if the slope coefficients of a panel
regression are jointly zero. It does not perform general purpose
Wald-style tests (for those, see lmtest::waldtest() (from package
lmtest) or car::linearHypothesis() (from package
car)).
If a user specified variance-covariance matrix/function is given in
argument vcov, the robust version of the tests are carried out.
In that case, if the F test is requested (test = "F") and no
overwriting of the second degrees of freedom parameter is given (by
supplying argument (.df2)), the adjustment of the second degrees
of freedom parameter is performed by default. The second degrees of
freedom parameter is adjusted to be the number of unique elements
of the cluster variable - 1, e. g., the number of individuals minus 1.
For the degrees of freedom adjustment of the F test in general,
see e. g. Cameron and Miller (2015)
, section VII;
(Andreß et al. 2013)
, pp. 126, footnote 4.
The degrees of freedom adjustment requires the vcov object supplied
or created by a supplied function to carry an attribute called
"cluster" with a known clustering described as a character (for now
this could be either "group" or "time"). The vcovXX functions
of the package plm provide such an attribute for their
returned variance–covariance matrices. No adjustment is done for
unknown descriptions given in the attribute "cluster" or when the
attribute "cluster" is not present. Robust vcov objects/functions
from package clubSandwich work as inputs to pwaldtest's
F test because a they are translated internally to match the needs
described above.
Wooldridge JM (2010). Econometric Analysis of Cross–Section and Panel Data, 2nd edition. MIT Press.
Andreß H, Golsch K, Schmidt-Catran A (2013). Applied Panel Data Analysis for Economic and Social Surveys. Springer. doi:10.1007/978-3-642-32914-2 .
Cameron AC, Miller DL (2015). “A Practitioner's Guide to Cluster-Robust Inference.” Journal of Human Resources, 50(2), 317-372. https://ideas.repec.org/a/uwp/jhriss/v50y2015i2p317-372.html.
vcovHC() for an example of the vcovXX functions, a robust
estimation for the variance–covariance matrix; summary.plm()
data("Grunfeld", package = "plm")
mod_fe <- plm(inv ~ value + capital, data = Grunfeld, model = "within")
mod_re <- plm(inv ~ value + capital, data = Grunfeld, model = "random")
pwaldtest(mod_fe, test = "F")
#>
#> F test for joint significance
#>
#> data: inv ~ value + capital
#> F = 309.01, df1 = 2, df2 = 188, p-value < 2.2e-16
#> alternative hypothesis: at least one coefficient is not null
#>
pwaldtest(mod_re, test = "Chisq")
#>
#> Wald test for joint significance
#>
#> data: inv ~ value + capital
#> Chisq = 657.67, df = 2, p-value < 2.2e-16
#> alternative hypothesis: at least one coefficient is not null
#>
# with robust vcov (matrix, function)
pwaldtest(mod_fe, vcov = vcovHC(mod_fe))
#>
#> Wald test for joint significance (robust), vcov: vcovHC(mod_fe)
#>
#> data: inv ~ value + capital
#> Chisq = 63.549, df = 2, p-value = 1.587e-14
#> alternative hypothesis: at least one coefficient is not null
#>
pwaldtest(mod_fe, vcov = function(x) vcovHC(x, type = "HC3"))
#>
#> Wald test for joint significance (robust), vcov: function(x) vcovHC(x,
#> type = "HC3")
#>
#> data: inv ~ value + capital
#> Chisq = 45.836, df = 2, p-value = 1.114e-10
#> alternative hypothesis: at least one coefficient is not null
#>
pwaldtest(mod_fe, vcov = vcovHC(mod_fe), df2adj = FALSE) # w/o df2 adjustment
#>
#> Wald test for joint significance (robust), vcov: vcovHC(mod_fe)
#>
#> data: inv ~ value + capital
#> Chisq = 63.549, df = 2, p-value = 1.587e-14
#> alternative hypothesis: at least one coefficient is not null
#>
# example without attribute "cluster" in the vcov
vcov_mat <- vcovHC(mod_fe)
attr(vcov_mat, "cluster") <- NULL # remove attribute
pwaldtest(mod_fe, vcov = vcov_mat) # no df2 adjustment performed
#>
#> Wald test for joint significance (robust), vcov: vcov_mat
#>
#> data: inv ~ value + capital
#> Chisq = 63.549, df = 2, p-value = 1.587e-14
#> alternative hypothesis: at least one coefficient is not null
#>