TSTATS prints a table of names, coefficients, standard errors, t-statistics, and (if it is not suppressed using REGOPT) a variance-covariance matrix. TSTATS is useful when you compute your own estimate of a variance-covariance matrix, since it is tedious to take the square roots of the diagonal elements and generate a readable printout of the results.
TSTATS (NAMES=(<list of names>)) <coefficient vector> <variance-covariance matrix> ;
Usage
The simplest form is TSTATS followed by the name of a vector containing the estimated values of a set of coefficients and the name of a symmetric matrix which contains the estimated covariance matrix of those coefficients. If the vector of coefficients is N long, the matrix must be of order N. The NAMES option allows you to label the coefficients in the table conveniently.
A table of regression coefficients, etc. is printed unless it has been SUPRESed. @RNMS, @COEF, @SES, and @VCOV are stored.
NAMES= <list of coefficient names>. The default is just to number the coefficients 1,2, etc.
Suppose, for example, that we have manually created PDL variables for use in a nonlinear regression, and then unscrambled the regression coefficients and their covariance matrix (by using the PDL transformation matrix). The estimated lag coefficients are called BETA and their covariance matrix estimate is VARB. A table of estimates with the corresponding t-statistics is printed by implementing the following command:
TSTATS(NAMES=(BETA1-BETA7)) BETA VARB ;