Chow predictive failure test
[resulch]=predfailin(resulols,n1)
* resulols = a results tlist from a first stage estimation
* n1 = # of observations of the sub-period
* resulch = a results tlist with:
- resulch('meth') = predfail
- resulch('rols') = the tlist results from first stage estimation
- resulch('fstat') = value of the test statistic
- resulch('f_pvalue') = its p-value
- resulch('dfnum') = # dof of the numerator
- resulch('dfden') = # dof of the denominator
- resulch('cut') = date (if there is a ts in the regression) or index of the observation (if there is no ts in the regression) of the break
load(GROCERDIR+'/data/bdhenderic.dat') ; bounds('1964q3','1989q2') // column HE in table 9 of Hendry and Krozlig paper r=ols('delts(lm1-lp)','lagts(lm1-lp-ly)','delts(lp)','rnet','delts(lagts(lm1-lp-ly))','cte') predfailin(r,50) // Example taken from predfailin_d. The example provides the predictive failure test with a break at observation # 50 // for Hendry and Ericsson equation # 6, whose results have been stored in tlist r. | ![]() | ![]() |