Statistical Inferences for Possibly Nonstationary and Nonlinear Time Series Models with a Hysteretic (Buffered) Structure

This project aims to develop statistical inferences for nonlinear time series data with a hysteretic (buffered) non-linearlity. A major objective is to derive unit root non-stationarity tests for such time series. Conversely tests for buffered non-linearity are also developed under the assumption of the presence of unit root. Extension of classical time series inferential procedures such as quantile estimation, smooth transitions in and out of the buffered region for this type of time series will also be considered. This project was originally under HKU but was transferred to this university last September.

Funding Source: RGC - General Research Fund (GRF)