Time Series with Long Memory

This volume provides in a convenient format for students and researchers the core papers in long memory time series analysis. Various methods and their theoretical properties are discussed, with empirical applications. The methods constitute a very flexible approach to analysing time series data arising in economics, finance, and other fields.

Long memory processes constitute a broad class of models for stationary and nonstationary time series data in economics, finance, and other fields. Their key feature is persistence, with high correlation between events that are remote in time. A single ''memory'' parameter economically indexes this persistence, as part of a rich parametric or nonparametric structure for the process. Unit root processes can be covered, along with processes that are stationary but with stronger persistence than autoregressive moving averages, these latter being included in a broader class which describes both short memory and negative memory. Long memory processes have in recent years attracted considerable interest from both theoretical and empirical researchers in time series and econometrics.This book of readings collects articles on a variety of topics in long memory time series including modelling and statistical inference for stationary processes, stochastic volatility models, nonstationary processes, and regression and fractional cointegration models. Some of the articles are highly theoretical, others contain a mix of theory and methods, and an effort has been made to include empirical applications of the main approaches covered. A review article introduces the other articles but also attempts a broader survey, traces the history of the subject, and includes a bibliography.