Journal of Business Cycle Measurement and Analysis

Frequency :
3 times a year
ISSN :
1729-3626 (online)
ISSN :
1729-3618 (print)
DOI :
10.1787/17293626
Hide / Show Abstract

The Journal of Business Cycle Measurement and Analysis is jointly published by the OECD and the Centre for International Research on Economic Tendency Surveys (CIRET) to promote the exchange of knowledge and information on theoretical and operational aspects of economic cycle research, involving both measurement and analysis (see www.ciret.org/bcma). Now published as a part of the OECD Journal.

Article
 

Out-of-sample Performance of Leading Indicators for the German Business Cycle

Single vs. Combined Forecasts You do not have access to this content

 
Click to Access: 
    http://oecd.metastore.ingenta.com/content/3305011ec003.pdf
  • PDF
  • http://www.keepeek.com/Digital-Asset-Management/oecd/economics/out-of-sample-performance-of-leading-indicators-for-the-german-business-cycle_jbcma-2005-5km7v183qs0v
  • READ
Author(s):
Christian Dreger, Christian Schumacher
Publication Date
08 June 2005
Pages
3
Bibliographic information
No.:
3,
Volume:
2005,
Issue:
1
Pages
71–87
DOI
10.1787/jbcma-2005-5km7v183qs0v

Hide / Show Abstract

In this paper the forecasting performance of popular leading indicators for the German business cycle is investigated. Survey based indicators (ifo business climate, ZEW index of economic sentiment) and composite leading indicators (Handelsblatt, Frankfurter Allgemeine Zeitung, Commerzbank) are considered. The analysis points to a significant relationship of the indicators to the business cycle within the sample period, as measured by the direction of causality. But, their out-of-sample forecasts do not improve the autoregressive benchmark. This result may be caused by structural breaks in the out-of-sample period. As combinations of forecasts tend to be more robust against such shifts, pooled forecasts are constructed using different methods of aggregation, including linear combinations of forecasts and common factor models. In contrast to the single indicator approach, the combined indicator forecasts are able to beat the benchmark at each forecasting horizon. Therefore, the analysis points to the usefulness of pooling information in order to get more reliable forecasts.