Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/233340 
Autor:innen: 
Erscheinungsjahr: 
2019
Schriftenreihe/Nr.: 
Discussion paper No. 125
Verlag: 
Aboa Centre for Economics (ACE), Turku
Zusammenfassung: 
We use the gradient boosting estimation technique and the ROC curveto non-parametrically measure and exploit the maximal predictive powerof leading indicators for the future state of the business cycle. We de-velop novel procedures for finding the best performing transformationsof individual indicators, for combining them to form an optimal reces-sion prediction model and for assessing which predictors are contribut-ing in the model. Among our empirical findings with US data are thatthe predictive impact of various indicators is non-monotone and thatrecession predictions based on our nonparametric procedures clearlyoutperform the ones based on a conventional probit model.
Schlagwörter: 
gradient boosting
leading indicators
non-parametric esti-mation
optimal binary prediction
recession prediction
JEL: 
C22
C25
C53
E37
Dokumentart: 
Working Paper

Datei(en):
Datei
Größe





Publikationen in EconStor sind urheberrechtlich geschützt.