We derive a new criterion for the optimal and efficient selection of the number of lags in the context of testing linear Granger causality environment in the sense of theory and empirics respectively, which may be better ...