Scripted and unscripted disclosure: Tone in Brazilian earnings calls and short-window market reactions

Abstract

Earnings calls provide a setting in which investors observe both scripted managerial communication and more interactive analyst–management exchanges. This paper examines whether linguistic tone in Brazilian earnings calls is associated with short-window market reactions, and whether the question-and-answer (Q&A) segment is more informative than prepared remarks. Using 1640 earnings-call transcripts from 77 Brazilian listed firms over 4Q 2016–4Q 2025, we measure tone separately for the prepared introduction and Q&A segments using Loughran–McDonald, VADER, and FinBERT, and relate these measures to cumulative abnormal returns and cumulative abnormal trading volume around call dates. The main finding is that negative tone in earnings calls is systematically associated with lower short-window abnormal returns, with negative tone in the Q&A segment emerging as the most robust textual correlate. Although formal tests indicate that the economic magnitude of negative tone in the Q&A segment is statistically indistinguishable from that in the prepared remarks, the Q&A association is estimated with substantially greater precision and cross-specification stability. By contrast, the association for tone in the prepared introduction is statistically unreliable and less consistently estimated across specifications, and the evidence for abnormal trading volume is more fragile. Additional evidence from call timing suggests that Q&A negativity is most consistently associated with abnormal returns when calls occur during trading hours, when investors can react as the discussion unfolds. The results are interpreted as reduced-form evidence that unscripted analyst–management exchanges contain information that investors regard as relevant in an emerging-market setting. The paper does not claim to identify a causal effect of tone, nor to establish the general superiority of any single sentiment model. Instead, it contributes to the disclosure literature by providing new evidence on earnings-call communication in Brazil and by showing that the internal segmentation of the call matters for understanding investor responses to qualitative corporate disclosure.