Dennis Kristensen (University College London)

Non-parametric estimation of time-varying parameters in nonlinear models

Abstract:  We propose nonparametric estimators of time-varying parameters in a general class of nonlinear time series models. Under weak regularity conditions, we show the proposed estimators are consistent and follow a normal distribution in large samples.  A key concept in our analysis is local stationarity, for which we provide primitive conditions to hold in the case of Markov processes.  To demonstrate the usefulness of our general results, we provide primitive conditions for our theory to apply in a number of examples, including ARCH models and Poisson autoregressions with time-varying parameters.