NEW ROBUST INFERENCE FOR PREDICTIVE REGRESSIONS

We propose a robust inference method for predictive regression models under heterogeneously persistent volatility as well endogeneity, persistence, or heavy-tailedness of regressors. This approach relies on two methodologies, nonlinear instrumental variable estimation and correction, which are used to deal with the aforementioned characteristics regressors volatility, respectively. Our is simple implement applicable both in case continuous discrete time models. According our simulation study, proposed performs compared widely alternative procedures terms its finite sample properties various dependence persistence settings observed real-world financial economic markets.