CEU Electronic Theses and Dissertations, 2021
| Author | Kondrat, Lili |
|---|---|
| Title | News Sentiments and Expectations: Text Analysis of Hungarian Economic News |
| Summary | The paper examines whether a Sentiment Index based on Hungarian economic news can be helpful to predict economic expectations, currently measured in the form of the Consumer Confidence Index (CCI). The sentiment and outlook of the actors in an economy is of key importance for policymakers, knowing more precisely what the consumers or businesses expect can help in policy choices and could lead to better economic outcomes. The currently available survey-based measure of economic expectations is available monthly, therefore methods that could provide more frequent and timely measurements of sentiment could provide huge benefits. The paper builds a substantial dataset of Hungarian economic news articles from 4 leading news portals and applies text analysis to create a monthly Sentiment Index. The paper then examines whether the Sentiment Index is useful for forecasting the Consumer Confidence Index. The hypothesis is tested by ADL(2,2) model and Granger Causality, and the results suggest that the Sentiment Index is a useful predictor for the CCI, but the results depend on the precision of the sentiment analysis. The findings suggest that policymakers could construct the Sentiment Index on a weekly or even a daily basis and use it to finetune their forecasts about economic expectations. More precise forecasts would enable better policy choices, more suited for the current conditions, therefore the Sentiment Index is a useful tool for nowcasting. |
| Supervisor | Sebok, Miklos; Matyas, Laszlo |
| Department | Economics MA |
| Full text | https://www.etd.ceu.edu/2021/kondrat_lili.pdf |
Visit the CEU Library.
© 2007-2021, Central European University