rhobis
commented
2 months ago The log t-test by Phillips and Sul is a left one-tailed test with threshold -1.65 (the null hyphotesys is rejected if t<-1.65, and as per step 2 their algorithm the club is formed if t>-1.65, so basically a group is formed when the null is not rejected, so p-value > 5%). For this reason, in function estimateMod() we use Pr(T<t) to calculate the pvalue, and a p-value >0.05 is a sign that the rule for creating a club is satisfied.
Anyway, the pvalue does not affect the results, as the t-value is used in the algorithm to form clubs.
References: https://github.com/rhobis/ConvergenceClubs/blob/master/R/estimateMod.R Sichera, R.; Pizzuto, P., 2019. ConvergenceClubs: A Package for Performing the Phillips and Sul’s Club Convergence Clustering Procedure. The R Journal Phillips, P. C.; Sul, D., 2007. Transition modeling and econometric convergence tests. Economet- rica 75 (6), 1771-1855. Phillips, P. C.; Sul, D., 2009. Economic transition and growth. Journal of Applied Econometrics 24 (7), 1153-1185
clubs <- findClubs(a, dataCols=2:39, unit_names = 1, refCol=2,HACmethod="AQSB",cstar_method = "fixed")
beta: 0.4517 std.err: 0.1121 tvalue: 4.0281 pvalue: 1 cstar: 0
======================================================================== club 2
KNR, ALP, KSGD, MLP, TCR, KKD
beta: 2.957 std.err: 0.2374 tvalue: 12.4577 pvalue: 1 cstar: 0
======================================================================== divergent
IDK
here even though High t values are there, I am getting P values 1