Title: Economic policy uncertainty and COVID-19 pandemic media coverage: The machine learning approach
Authors: Wojciech Charemza - Vistula University (Poland)
Svetlana Makarova - University College London (United Kingdom) [presenting]
Krzysztof Rybinski - Vistula University Warsaw (Poland)
Abstract: New uncertainty measures are proposed, which directly incorporate the effects of media coverage of the pandemic. We hypothesise that the excessive media coverage of the COVID-19 pandemic in 2020, crowded out other news; in particular, these related to economic policy uncertainty. As a consequence, the economic policy uncertainty (EPU) measures constructed from the frequencies of the appearance of the uncertainty-related terms in the media become biased downward, even though the pandemic reporting actually increased uncertainty. To show this, we construct health-augmented (HEPU) uncertainty indices for five countries: Belarus, Kazakhstan, Poland, Russia and Ukraine. These countries conducted, in the first half of 2020, different anti-pandemic policies and their media exhibited different approaches towards the pandemic coverage. We apply Word2vec to define cosine-similar words to distinguish health-related, economic, policy and uncertainty terms. For topics identification, we use the unsupervised learning technique based on the Latent Dirichlet Allocation method. The results confirm the impact the reporting of the pandemic had on the development of the crowding-out effect. We also evaluate the resulting bias in the EPU indices. The findings vary across the analysed countries, which is explained to the different styles of media coverage of the pandemic.