Clive W. J. Granger’s research while affiliated with University of California, San Diego and other places

No abstract available.

Even though the trend components of economic time series were among the first to be distinguished, even today the trend remains relatively little understood. As Phillips (2005) notes, no one understands trends, but everyone sees them in the data. Economists and econometricians can give plenty of examples of trends, such as straight lines, exponentials or polynomials in time, and also forms of random walks, but these are merely examples. Individuals or groups do have their own personal definitions, but these diverse approaches illustrate the lack of a generally accepted definition of a trend. They also suggest a richness of alternatives to consider, both individually and jointly. Here, we make a variety of observations about trends, and based on these, we offer working definitions of various kinds of trends. We emphasize that these are working definitions, as our purpose here is to invite discussion, not to settle matters once and for all. Our hope is that our discussion here may facilitate development of increasingly better methods for prediction, estimation and hypothesis testing for non-stationary time-series data, and ultimately may enable decision makers to make more informed decisions.

Phillips' (1958) original curve involves a nonlinear relationship between inflation and unemployment. We consider how his original results change due to updated theoretic and empirical studies, increased computer power, enlarged datasets, increases in data frequency and developed time series econometric models. In the linear models, there was weak causation from unemployment to inflation. Rather than using any of the many nonlinear models that are now available, we adopt a time-varying parameter linear model as their convenient proxy, which empirically supports Phillips' use of nonlinear model form and causation, but the strength of this result is much weaker in recent periods.

This volume explains recent theoretical developments in the econometric modelling of relationships between different statistical series. The statistical techniques explored analyse relationships between different variables, over time, such as the relationship between variables in a macroeconomy. Examples from Professor Terasvirta's empirical work are given. Professors Granger and Terasvirta are leading exponents of techniques of dynamic, multivariate analysis. They illustrate in this volume exploratory ways of using such techniques to provide models of nonlinear relationships between variables. This is an extension of previous work on linear relationships, and on univariate models. These developments will be of use to econometricians wishing to construct and use models of nonlinear, dynamic, multivariate relationships, such as an investment function, or a production function. Particular attention is paid to the case of a single dependent variable modelled by a few explanatory variables and the lagged dependent variable in nonlinear form. The book concentrates on stochastic series, since the existence of unexpected shocks strongly suggests that economic variables are stochastic. Granger and Terasvirta also discuss the division of these nonlinear relationships into parametric and nonparametric models.

This book contains an extensive up-to-date overview of nonlinear time series models and their application to modelling economic relationships. It considers nonlinear models in stationary and nonstationary frameworks, and both parametric and nonparametric models are discussed. The book contains examples of nonlinear models in economic theory and presents the most common nonlinear time series models. Importantly, it shows the reader how to apply these models in practice. For this purpose, the building of various nonlinear models with its three stages of model building: specification, estimation and evaluation, is discussed in detail and is illustrated by several examples involving both economic and non-economic data. Since estimation of nonlinear time series models is carried out using numerical algorithms, the book contains a chapter on estimating parametric nonlinear models and another on estimating nonparametric ones. Forecasting is a major reason for building time series models, linear or nonlinear. The book contains a discussion on forecasting with nonlinear models, both parametric and nonparametric, and considers numerical techniques necessary for computing multi-period forecasts from them. The main focus of the book is on models of the conditional mean, but models of the conditional variance, mainly those of autoregressive conditional heteroskedasticity, receive attention as well. A separate chapter is devoted to state space models. As a whole, the book is an indispensable tool for researchers interested in nonlinear time series and is also suitable for teaching courses in econometrics and time series analysis.

This paper describes how the notion of cointegration came about, and discusses some generalizations to indicate where the topic may go next. In particular, some issues in the analysis of possibly cointegrated quantile time series are discussed.

For the first three-quarters of the 20th century the main workhorse of applied econometrics was the basic regression.

Decisions in the fields of economics and management have to be made in the context of forecasts about the future state of the economy or market. As decisions are so important as a basis for these fields, a great deal of attention has been paid to the question of how best to forecast variables and occurrences of interest. There are several distinct types of forecasting situations, including event timing, event outcome, and time-series forecasts. Event timing is concerned with the question of when, if ever, some specific event will occur, such as the introduction of a new tax law, or of a new product by a competitor, or of a turning point in the business cycle. Forecasting of such events is usually attempted by the use of leading indicators, that is, other events that generally precede the one of interest. Event outcome forecasts try to forecast the outcome of some uncertain event that is fairly sure to occur, such as finding the winner of an election or the level of success of a planned marketing campaign. Forecasts are usually based on data specifically gathered for this purpose, such as a poll of likely voters or of potential consumers. There clearly should be a positive relationship between the amount spent on gathering the extra data and the quality of the forecast achieved.