B0835
Title: Log-concavity and quasi-maximum likelihood estimation
Authors: Yining Chen - London School of Economics and Political Science (United Kingdom) [presenting]
Abstract: We propose a general quasi-maximum likelihood approach in time series analysis. Our quasi-likelihood function is constructed by assuming that the innovations follow a log-concave distribution. This framework can be easily adapted to many well-known time series models, including autoregressive moving-average (ARMA) models, threshold autoregressive (TAR) models, generalised autoregressive conditional heteroscedasticity (GARCH) models, and ARMA-GARCH models. Furthermore, we show that the estimator under this new framework is consistent in all the above-mentioned settings, even when the innovations are not log-concave. We demonstrate its promising finite sample performance via a thorough simulation study and apply it to model the daily return of FTSE 100 index.
