|
Activity Number:
|
63
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #304809 |
|
Title:
|
Obtaining Early Signals About US Recessions: An Application of a New and Efficient Multivariate Real-Time Filter
|
|
Author(s):
|
Marc Wildi*+
|
|
Companies:
|
Institute of Data Analysis and Process Design
|
|
Address:
|
Rosenstrasse 4, Winterthur, International, ch-8401, Switzerland
|
|
Keywords:
|
Multivariate filter ; Real-time filter ; Efficiency
|
|
Abstract:
|
This paper proposes a new approach to the detection of US recessions that seeks to replicate the recession dates declared by the Business-Cycle Dating Committee (BCDC) of the National Bureau of Economic Research (NBER). Releases of recession dates by the BCDC are typically delayed by 6 to 18 months. We propose a Multivariate Direct Filter Approach (MDFA) and apply it to a `standard' small set of economic variables. Our new filter is both fast and reliable. It anticipates the last seven recessions from 1969 to 2009 by an average of 3.5 months after including data publication lags and the last recession is signaled with only a two month delay in February 2008. Revisions in the filter due to data-updating are small because the filter assigns most of the weight to data that is either unrevised or that is only slightly revised.
|
