Perform Wald test of coefficients using symmetric quantiles. The symmetric quantile test restrictions are of the form:
.
Syntax
eq_name.qrsymm(options) [arg] [@coefs coeflist]
where arg is a optional list containing the quantile values (specified using numbers, scalar objects, or vectors) for which you wish to compute estimates, and optionally the @coefs keyword followed by a coeflist of the subset of coefficients to display.
• If arg is not specified, EViews will perform one of two tests, depending on the original equation specification:
If the original specification is a median regression (
), EViews will test using estimates obtained at the specified outer quantiles as specified by the “n=” option. If “n=” is not specified, the default is to display results for the outer quartiles {0.25, 0.75}.
For specifications estimated with
, EViews will include the original quantile in the set of quantiles to test. You may specify “n=e” to perform a test using only estimates obtained at the symmetric pair {
,
}.
• If arg is specified, EViews will perform the test using only the specified quantiles and their complements. The original equation quantile will not be tested unless it is entered explicitly.
• If a coeflist is not provided, results for all coefficients will be displayed. For models that contain an intercept, the coeflist may consist of the @incptonly keyword, indicating that only results for the intercept will be displayed.
You may specify a maximum of 1000 total coefficients (number of coefficients in the equation specification times the number of quantiles) and a maximum of 500 quantiles in the test.
All estimation will be performed using the settings from the original equation. Note that the original equation must include an intercept for you to perform this test
Options
n=arg (default=4) | Number of quantiles for testing. |
quantout=name | Save vector containing test quantile values. |
coefout=name | Save matrix containing test coefficient estimates. Each column of the matrix corresponds to a different quantile matching the corresponding quantile in “quantout=”. To match the covariance matrix given in “covout=” you should take the @vec of the coefficient matrix. |
covout=name | Save symmetric matrix containing covariance matrix for the vector set of coefficient estimates. |
prompt | Force the dialog to appear from within a program. |
p | Print output from the test. |
Examples
equation eq1.qreg log(y) c log(x)
eq1.qrsymm
estimates a quantile (median) regression of LOG(Y) on a constant and LOG(X), and performs a symmetry test using the outer quartiles.
We may restrict the hypothesis to just consider the intercept,
eq1.qrsymm @coefs @incptonly
and we may specify alternative quantiles to test
eq1.qrsymm(quantout=qo) .2 .4 .7
Note that the latter command will test using the symmetric quantiles {0.2, 0.3, 0.4, 0.6, 0.7, 0.8}, and at the median. Note that the median is automatically estimated, even though it is not specified explicitly, since it is always required for testing.
Alternatively, the commands
equation eq1.qreg(quant=.4) log(y) c log(x)
eq1.qrsymm(n=0)
will perform the test using the symmetric quantiles {0.4, 0.6} and the median.
To performs the test using all of the deciles, you may enter
vector(4) v1
v1.fill .1,.2,.3,.4
eq1.qrsymm v1
Cross-references
See
“Symmetric Quantiles Test” for a discussion of the symmetric quantiles test.