Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/312430 
Erscheinungsjahr: 
2024
Schriftenreihe/Nr.: 
Ruhr Economic Papers No. 1124
Verlag: 
RWI - Leibniz-Institut für Wirtschaftsforschung, Essen
Zusammenfassung: 
We present an Uncertainty Perception Indicator (UPI) for Germany based on the dynamic topic modelling technique RollingLDA. In contrast to conventional LDA, where all data is processed in one go, the recursive structure of RollingLDA ensures that data is made available for modeling as soon as it is actually published, which prevents information leakage. Employing this approach facilitates the close-to-realtime identification of both the magnitude of an uncertainty shock as well as its specific characteristics. Thereby, more precise predictions about the likely impact are possible, as different sources of uncertainty have different repercussions in the macroeconomy. Employing a Bayesian VAR approach, we analyze the effects of various uncertainty shocks on fixed investment and other macroeconomic variables. Our results document the asymmetric nature of uncertainty shocks, as their consequences are highly dependent on the respective sources of uncertainty. We find that international shocks only have weak effects on the German macroeconomy, while domestic policy shocks prove to be highly significant. Uncertainty hurts most when it originates close to home. These results markedly differ from earlier studies that, in the case of Germany, tend to maintain the opposite. Interestingly, the results for the entire UPI (the sum of all individual UPI time-series) are broadly insignificant. In contrast, some single uncertainty topics show quite strong effects. We attribute this to information losses that occur when the entirety of uncertainty-related reporting is employed. Different UPI topics tend to offset one other. The RollingLDA technique helps disentangling the information hidden in the analysis corpus.
Schlagwörter: 
Uncertainty
topic modeling
business cycle
fixed investment
geoeconomics
JEL: 
C32
C82
D80
E20
Persistent Identifier der Erstveröffentlichung: 
ISBN: 
978-3-96973-306-6
Dokumentart: 
Working Paper

Datei(en):
Datei
Größe
651.95 kB





Publikationen in EconStor sind urheberrechtlich geschützt.