Econometrics

Research Area and Mission

We focus on three themes:

1. Development of dynamic econometric models with nonlinearities in the model specification and plausible restrictions in the parameter space.
2. Advanced computational techniques that can handle such nonlinearities and restrictions in an efficient and reliable way.
3. Discrete choice analysis at an individual level. 

The aim of this programme is to provide econometric tools that lead to efficient forecasting of the level and risk of economic activities in macroeconomics and finance and lead to improvements in individual economic decision making in the field of marketing.

Programme leader(s)

Prof.dr. D.J.C. van Dijk

Description of Research Areas

The strategy and policy of this research group are characterized by the style of leadership and plans for future research.

In the previous evaluation of this research programme, the research area was specified as: ‘Econometric Analysis of Dynamic Models’. It had six subtopics. This was considered too large a number by the evaluation committee. A recommendation was made to restrict our focus to three subtopics. In the present programme, we now focus on three themes where cross-fertilization plays an important role:

1. Development of dynamic econometric models with nonlinearities in the model specification and plausible restrictions in the parameter space.
2. Advanced computational techniques that can handle such nonlinearities and restrictions in an efficient and reliable way.
3. Discrete choice analysis at an individual level.

The motivation for this shift relates to three important changes that occurred over the past thirty years. One such change is the deregulation of many economic systems (in Eastern Europe in particular). The second change is the technological advancement in financial products, systems and services. As a consequence of these two changes, one observes more shocks to economic processes and more nontrivial dynamic behaviour in important economic variables. For instance, the processing of information moves at a much greater speed than in the past. This has had a substantial effect on the level of volatility of financial processes. The third important change is the enormous increase in the availability of large data sets: high frequency data in the financial sector and scanner data in the marketing sector.

Nontrivial dynamic behaviour and the availability of large data sets has led to our investigation of nonlinear structures in dynamic models like threshold processes and to the investigation of time varying volatility. Equally important were plausible restrictions in the parameter space such as reduced rank restrictions due to structural information, co integration and common factors. In this context, macro-economic and financial processes are studied.

The increased availability of large scanner panel data sets in marketing led to the study of dynamic behaviour of individuals and the analysis of behaviour of micro-economic variables like prices and quantities of individual products.

The knowledge obtained during the construction, estimation and forecasting of macroeconomic and financial variables is usefully applied in the field of marketing. An interesting observation is that research results in dynamic macroeconomic modelling can be used for the analysis of individual decision-making processes in a dynamic context. This cross-fertilization of macro- and microeconometrics has produced a considerable number of new insights and has led to an increased output of high-quality papers that have appeared in top journals.

Alongside this development, the ‘Computational Revolution’ took place in several fields of scientific activity including psychology, physics, chemistry and biology. Nonlinear, partially observed processes where simulation techniques are used are also becoming increasingly important in Economics. In the present programme, a major research topic is the development of simulation-based inferential techniques that are used in a Bayesian and/or likelihood analysis of econometric models in macroeconomics, finance and marketing. The architecture of constructing effective Markov Chain Monte Carlo methods for several alternative model structures is actively pursued with considerable success. Another topic of importance is advanced computational techniques for non-parametric statistics, and the development of such techniques for visualization models. The latter area is becoming increasingly important in many scientific fields. In this present research programme, such techniques perform well in areas such as prediction of macroeconomic time series and in the context of market share attraction models.

Key Publications

The top five publications were selected due to the fact that they present solutions to an economic or econometric problem that is widely applicable in practice. 

1. Dijk, D.J.C. van, P.H.B.F. Franses, and R. Paap, (2002). A nonlinear long memory model, with an application to US unemployment, Journal of Econometrics, 110, 135-165.
2. Hoogerheide, L.F., J.F. Kaashoek, and H.K. van Dijk, (2007). On the shape of posterior densities and credible sets in instrumental variable regression models with reduced rank: an application of flexible sampling methods using neural networks, Journal of Econometrics, 139, 154-180.
3. Sensier, M. and D.J.C. van Dijk, (2004), Testing for volatility changes in US macroeconomic time series, Review of Economics and Statistics 86, 833-839.
4. Kleibergen, F.R. and R. Paap, (2002), Priors, Posteriors and Bayes factors for a Bayesian Analysis of Cointegration, Journal of Econometrics, 111, 223-249.
5. Harvey, A.C., T.M. Trimbur, and H.K. van Dijk, (2007). Trends and cycles in economic time series: a Bayesian approach, Journal of Econometrics, 140, 618-649.