Estimation and inference in the presence of fractional \(d=1/2\) and weakly nonstationary processes (Q2039810)

The article builts upon the issue of regression modeling with temporally dependent data. Specifically the article contributes the asymptotic analysis of regression estimators for a class of processes which lies between the stationary and more strongly dependent nonstationary processes (weakly nonstationary processes). The developments enable the establishment of the asymptotic (mixed) normality of kernel estimates of the regression function when the indepenent variables are weakly nonstationary. This is used further to develop a test for parametric specifications of the regression function. The test is based on a comparison of the fit provided by parametric and nonparametric estimates of the regression function, and has the nice property that its limiting distribution is invariant to the persistence of the explanatory variables. In turn this enables practitioners to apply the test using a fixed set of critical values, while remaining agnostic about the dependence properties of the regressors. Finally, the finite-sample performance of the test is evaluated through numerical examples.