A0335
Title: Measuring fiscal policy preferences based on the German Bundestag speeches and public discourse
Authors: Viktoriia Naboka-Krell - Justus Liebig Unversity of Giessen (Germany) [presenting]
Peter Winker - University of Giessen (Germany)
Peter Tillmann - Justus-Liebig University Giessen (Germany)
Albina Latifi - Justus Liebig University Giessen (Germany)
Abstract: Fiscal policy, i.e. changes in government spending and tax revenues, is an important determinant of business cycles. Research efforts, so far, heavily rely on human judgment in order to measure fiscal policy shocks. Therefore, reproducible quantitative computational text analysis to quantify fiscal policy preferences are applied. The analysis is based on the driver of fiscal policy, i.e. the fiscal policy debates in the Bundestag, which are also accompanied by public discourse. Thus, two novel data sets on parliamentary speeches and newspaper reporting covering the time period from 1960 to 2021 are used. First, advanced NLP (Natural Language Processing) techniques, such as topic modelling, are applied to uncover latent topics in the corpus and to study the evolution of topic importance over time. Second, an embedding-based approach, which allows the representation of words and documents in a shared vector space, is proposed to measure fiscal policy-related sentiment. For this reason, a dictionary containing terms related to expansive and restrictive policy measures is created. Finally, an index is proposed that captures the sentiment from speeches and news articles at a scale from restrictive to expansive. In future research, the novel indicator will be used in econometric models to examine the macroeconomic impact of fiscal policy.
