Abstract

Whether book-to-market characteristics influence long-memory dynamics remains an open question, particularly in emerging equity markets and during periods of systemic stress. This study investigates long-memory dynamics in Brazilian stock returns and volatility and examines whether persistence differs across book-to-market (BTM) portfolios before, during, and after the COVID-19 pandemic. Using daily data (roughly 2,500 observations per stock) for ten highly liquid IBOVESPA stocks from January 2016 to December 2025, we estimate the fractional integration parameter (d) using the semiparametric Geweke-Porter-Hudak (GPH) estimator for the full sample, three subperiods, and rolling windows. The sample comprises ten selected stocks and does not represent the entire Brazilian equity market. The subperiod and rolling-window estimates describe variation in persistence across regimes and over time. The results show little evidence of long memory in returns, which remain broadly consistent with weak-form market efficiency despite a temporary increase in persistence during the pandemic. In contrast, volatility exhibits strong and pervasive long-memory behavior, with persistence intensifying substantially during the COVID-19 crisis before declining only gradually afterward. No systematic differences are found between Low- and High-BTM portfolios in either returns or volatility, suggesting that persistence is driven primarily by firm-specific sources of uncertainty rather than by book-to-market classification. Overall, the evidence indicates that long memory in the Brazilian equity market is more evident in volatility than in returns.

This work is licensed under a Creative Commons Attribution 4.0 License.