Granger (non-)causality tests from a VAR
res=vargranger(rvar,causing,caused,noprint)
rvar = a 'var' results tlist
causing = a (j x 1) vector, the indexes of the causing variables
caused = a (k x 1) vector, the indexes of the caused variables
noprint = 'noprint' if the user does not to print the results
res = a results tlist with:
* res('meth') = 'var Granger causality'
* res('rvar res') = the input 'var' results tlist
* res('causing') = the index vector of the supposed causing variables
* res('caused') = the index vector of the supposed caused variables
* res('chistat') = the vector of Wald statistics for non-causlity
* res('chi_pvalue') = their p-values
* res('chi_df') = the degrees of freedom of the chi2
* res('f') = the Fisher statistics for non-causlity
* res('dfnum') = the numerator degrees of freedom
* res('dfden') = the denominator degrees of freedom
global GROCERDIR; load(GROCERDIR+'/data/lutk1.dat') bounds('1960q4','1978q4') results=VAR(2,'endo=delts(log(rfa_inv));delts(log(rfa_inc));delts(log(rfa_cons))') r=vargranger(results,[1 2],3) // provides Ganger (non-)causality test from variables delts(log(rfa_inv)) and delts(log(rfa_inc)) to variable delts(log(rfa_cons)) r=vargranger(results,'delts(log(rfa_inv))',['delts(log(rfa_inc))' 'delts(log(rfa_cons))']) // same test but with the name of the variables instead of their order in the VAR | ![]() | ![]() |