A1525
Title: Financial attention and disclosure tone: Mixed frequency analysis
Authors: Alev Atak - METU (Turkey) [presenting]
Abstract: The aim is to examine how demand-side financial attention and supply-side disclosure tone jointly relate to market volatility in a retail-dominated emerging market. A monthly Financial Attention Index (FAI) is constructed from Turkish-language Wikipedia pageviews via PCA; disclosure tone is measured by a FinBERT-based, probability-and market-cap-weighted NetTone Polarity Index (PI) from Borsa Istanbul annual reports (2016--2024). Mixed Data Sampling (MIDAS) regressions integrate indicators at native frequencies. Granger tests show that lagged FAI predicts volatility (not vice versa), and an event-driven difference-in-differences (DiD) around the February 2023 earthquake indicates a temporary strengthening of the attention--volatility link. By contrast, the tone--volatility association is negative in baseline models, whereas it becomes statistically insignificant after controlling for year fixed effects, consistent with tone primarily reflecting annual macroeconomic states rather than an independent within-year behavioral effect. Two temporal-direction tests -- Granger causality and an event-driven DiD design -- address concerns about simultaneity and clarify temporal precedence. These findings support the Wikipedia-based attention indices as scalable, real-time surveillance tools in retail-heavy emerging markets, where the FAI captures high-frequency behavioral dynamics, while disclosure tone reflects slower-moving macroeconomic conditions.
