This paper introduces a new test statistic for the null hypothesis of short memory against long memory alternatives. The novelty of our statistic is that it is based on only high order sample autocovariances and by construction eliminates the effects of nuisance parameters typically induced by...