On Wed, 31 Mar 2010, Torben Schmith wrote: > I want to do Monte-Carlo simulations with the Johansen cointegration > test. However, as far as I can see the $value and $test are not assigned > new values when this test is carried out. > > My question is if there is any way I can can get hold of the test output > (as variables) from the Johansen test? I can't find anything on this > issue in the Gretl-documentation. Hmm, the trouble is that the Johansen cointegration test produces an array of statistics (the lambda-max and trace tests for various hypotheses), so it can't be put into the framework of a pair of scalars ($test, $pvalue). I suppose the best thing would be to enable a matrix accessor after this test. In the meantime, however, it is possible to retrieve the test matrix using gretl's string-handling facilities. I illustrate this below (in somewhat rough and ready fashion), but note that for this to work you will need current CVS or snapshot, where we have some improvements in the sscanf function. The strategy is to write the test results out to file, then parse that file from within gretl. <script> open denmark scalar r = 2 outfile "jtest" --write coint2 r LRM LRY outfile --close matrix Jmat = zeros(r, 6) string jtest = readfile("jtest") string tmp scalar x1 x2 x3 x4 x5 x6 # get to the start of the test matrix jtest = strstr(jtest, "Rank") sscanf(jtest, "%128[^\n]", tmp) jtest = jtest + strlen(tmp) + 1 sscanf(jtest, "%128[^\n]", tmp) jtest = jtest + strlen(tmp) + 1 sscanf(tmp, "%lf %lf %lf [%lf] %lf [%lf]", \ x1, x2, x3, x4, x5, x6) Jmat = {x1, x2, x3, x4, x5, x6} sscanf(jtest, "%128[^\n]", tmp) tmp = jtest + strlen(tmp) + 1 sscanf(tmp, "%lf %lf %lf [%lf] %lf [%lf]", \ x1, x2, x3, x4, x5, x6) Jmat |= {x1, x2, x3, x4, x5, x6} print Jmat </script> Allin Cottrell _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users