Home | Business News | Browse by Publication | I | International Advances in Economic Research

Modeling the European cycle with factor structure and regime switching.

Article Excerpt
Abstract

This paper compares the stylized facts of the European growth cycle stemming from the Gross Domestic Product (GDP) of the European Monetary Union with an unobserved common factor derived from a dynamic factor model with regime switching. The aim of this paper is to provide empirical evidence about the most adequate indicator for short-term monitoring of the cyclical state of the European economy. (JEL C32, E32, F15)

**********

Introduction

For the moment, the European Monetary Union (EMU) is the final step in creating a fully integrated European economic space. Although there are a few members of the European Union (EU) who maintain their currencies, the need for policy coordination is so strong that, in practice, all European economies are compelled to evolve in a similar way. Decisions at the European level are highly influenced by the cyclical position of the European economy. However, in spite of that, there still lacks a formal definition of what the European cycle is and how to measure it. At least one basic question remains unanswered: should the analysis be based on an observable aggregate like the EU-15 GDP or the euro area GDP (EMU-GDP)? Or is it better to follow Burns and Mitchell's approach [Burns and Mitchell, 1946; Stock and Watson, 1989, 1991] of monitoring several different variables to estimate the underlying, unobservable common cycle?

Both possibilities look valid. Focusing on an observable aggregate is easier and the EU-15 or the EMU-GDP would measure the activity level in the economic space that is of interest. On the other hand, one may argue that the relevant signal is the European cycle and that its best estimator need not be derived from a single series of level. One may also argue that the aggregate is dominated by the two or three biggest economies in the area and may not reflect the situation in the whole Union. These are arguments in favor of using the common factor approach.

This paper intends to shed some light on the most adequate indicator for short-term monitoring of the European economy. This paper also aims to show empirical evidence about two specific issues. First, is it possible to monitor the European cycle by concentrating on a single aggregate like EMU-GDP? The author derives a common factor underlying the observed evolution of national economies and analyzes whether EMU-GDP and the common factor display the same basic features. Second, in the affirmative case, the paper investigates whether there are substantial differences between the two indicators in tracking the observed short-term evolution of national GDPs. A related relevant issue is whether co-movements across countries may be fully explained by aggregate shocks at the European level, a result that would rule out the existence of clubs within the euro area or the EU.

This paper is organized as follows. The following section introduces a framework for analyzing the European cycle from national, disaggregated series that incorporates factor structure and regime switching. Next, the data is presented and the estimated models are reported. Then, the paper compares the stylized facts of the European cycle stemming from the common factor approach and EMU-GDP and evaluates their usefulness in tracking short-term performance of national economies. The analysis shows that there are not remarkable differences. However, in general, the common factor seems to be preferable. Empirical evidence of structural change is found in cross-country relations as the evolution of national economies depends more and more on a general indicator of the European economy and less on group-specific shocks. The final section concludes the paper.

The Model

To estimate a common factor underlying national GDPs that could be identified with the European cycle, a statistical characterization of the multivariate process of output growth that combines factor structure and regime switching (FSRS) is considered. The model is in the spirit of Diebold and Rudebusch [1996]. It may also be viewed as an extension of Stock and Watson [1991] that introduces regime switching in the common factor. The basic algorithm for estimation, filtering, and smoothing was presented in Kim [1994]. The model...