NONLINEAR COINTEGRATING POWER FUNCTION REGRESSION WITH ENDOGENEITY

This paper develops an asymptotic theory for nonlinear cointegrating power function regression. The framework extends earlier work on the deterministic trend case and allows both endogeneity heteroskedasticity, which makes models inferential methods relevant to many empirical economic financial applications, including predictive A new test linear cointegration against departures is developed based a simple linearized pseudo-model that very convenient practical implementation has standard normal limit in strictly exogenous regressor case. Accompanying of regression, establishes some results weak convergence stochastic integrals go beyond usual semimartingale structure considerably extend existing theory, complementing other recent findings integral asymptotics. also provides general extremum estimation encompasses stochastically nonstationary time series should be wide applicability.