EXCLUDE Instruction |
EXCLUDE(options)
# list of variables to be tested (in Regression Format)
EXCLUDE computes the test statistic for the restriction that all of the listed variables have zero coefficients in the preceding regression. You use EXCLUDE after you estimate the regression.
You cannot use EXCLUDE with NLLS, NLSYSTEM, MAXIMIZE, FIND, DLM or CVMODEL instructions, that is, any instruction which uses a PARMSET rather than a regressor list, nor after GARCH or SUR. TEST is a more general testing instruction which can be applied in these cases.
Wizard
You can use the Statistics—Regression Tests Wizard to do exclusion tests. Select the "Exclusion Restrictions" radio button.
Supplementary Card
The supplementary card lists the collection of variables from the previous regression which you want tested (as a block). List them in regression format.
Options
[PRINT]/NOPRINT
TITLE="title for output" [none]
NOPRINT suppresses the printing of the test information. This is useful if you only need the %CDSTAT or %SIGNIF variables, and don’t need to see the output. TITLE can be used to supply your own title to label the output.
ALL/[NOALL]
Use ALL to test whether all of the coefficients can be excluded. Omit the supplementary card if you use this. (This option was called WHOLE in version 6 and earlier.)
FORM=F/CHISQUARED
This determines the form of test statistic used. RATS can normally select the appropriate form by default based on the estimation technique used last. You can use the FORM option to manually select a distribution if you have made changes to the regression, such as altering the %XX matrix, that require a different distribution.
ENTRIES=standard ENTRIES option [use all]
Variables Defined
%CDSTAT |
the computed test statistic (REAL) |
%SIGNIF |
the marginal significance level (REAL) |
%NDFTEST |
(numerator) degrees of freedom for the test (INTEGER) |
Technical Information
These are done as Wald tests. The main test statistic is usually shown as an F, but will be shown as a chi-squared when EXCLUDE is applied to estimates from a DDV or LDV instruction, or from any instruction for which the ROBUSTERRORS option was used during estimation. You can also control the distribution yourself using the FORM option.
For F tests with one degree of freedom, EXCLUDE will report a two-tailed t test in addition to the F test. For chi-squared tests with more than one degree of freedom, EXCLUDE will report an F with an infinite number of denominator degrees of freedom (that is, the chi-squared statistic divided by the numerator degrees of freedom) in addition to the chi-square.
Examples
linreg gdph
# constant fm1{-4 to 8} gdph{1 to 8}
exclude(title="Geweke-Meese-Dent Causality Test")
# fm1{-4 to -1}
tests the leads (negative lags) on FM1 in the regression.
linreg(robusterrors) pcexp
# constant pcaid pcinc large dpcaid dpcinc
*
* Test dummies
*
exclude(title="Sample Split Test-Robust Standard Errors")
# large dpcaid dpcinc
Sample Output
Geweke-Meese-Dent Causality Test
Null Hypothesis : The Following Coefficients Are Zero
FM1 Lag(s) -4 to -1
F(4,534)= 2.19350 with Significance Level 0.06852115
Copyright © 2025 Thomas A. Doan