International Congress on Eurasian Economies
26-28 June 2024 – Bishkek, KYRGYZSTAN
Paper properties
Paper ID : 2907
Status : Paper accepted
Language : English
Topic : Finance and Financial Crises
Presenter: Research assistant Sedat Çerez
Session : 3A Finance
Evaluating Sentiment Changes in BIST Dividend Index Excluded Companies: A DistilBERT Analysis Approach
Evaluating Sentiment Changes in BIST Dividend Index Excluded Companies: A DistilBERT Analysis Approach
- Dr. Abdullah Kürşat Merter (Gebze Teknik University, Türkiye)
- Dr. Yavuz Selim Balcıoğlu (Gebze Technical University, Türkiye)
- Research assistant Sedat Çerez (Gebze Teknik University, Türkiye)
- Prof. Dr. Gökhan Özer (Kyrgyz-Turkish Manas University, Kyrgyzstan)
Abstract
The purpose of this study is to investigate whether the annual reports of companies that were excluded from the BIST Dividend Index in 2022 have shifted from a positive to a negative sentiment. To accomplish this, a sophisticated sentiment analysis was conducted using a DistilBERT model, a variant of the BERT model that is pre-trained on financial datasets. The study involved a meticulous data preparation process where texts were extracted from 12 annual activity reports of these companies, six from 2021 and another six from 2022. After the text extraction process, the data underwent thorough cleaning to remove unnecessary information and standardize the format so that it would be suitable for analysis. The 2021 reports exhibited an average sentiment score of +0.35, suggesting a predominantly positive sentiment. In contrast, the 2022 reports showed a negative shift, with an average sentiment score of -0.20. This shift from positive to negative sentiment is particularly noteworthy as it mirrors the companies' removal from the BIST Dividend Index. The study also explored the correlation between the sentiment of the reports and the market performance of the companies. A positive sentiment in 2021 was aligned with an 11.8% increase in market performance, whereas the negative sentiment in 2022 correlated with a 5.8% decline. These findings were presented with a clear distinction between the optimistic performance in 2021 and the downturn in 2022.
JEL codes: G17, G34, M41
