Essays in Empirical Finance: An Analysis of Brazilian Hedge Funds

Advisor: Nathalie Gimenes

Co-advisor: Marcelo Medeiros

This dissertation consists of three essays on dynamic trading by Brazilian macro hedge funds. In the first essay, we investigate whether aggregate hedge fund time-varying risk exposures to a set of risk factors contain predictive information about future inflation. We estimate time-varying exposures using state-space models and structural break models applied to daily Brazilian macro hedge fund returns. We find that macro hedge fund exposure to the exchange rate has explanatory power over cumulative 12-month ahead inflation.We also find that incorporating exchange rate exposure data into high-dimensional inflation forecasting models yields meaningful performance improvements. In the second essay, we investigate hedge fund replication strategies that condition on cross-sectional dispersion in estimated macro hedge fund exposures. Simple replication strategies yield modest and insignificant performance, but conditioning positions on a measure of dispersion substantially improves risk-adjusted returns. Dispersion-conditional strategies achieve improved Sharpe ratios and lower maximum drawdowns, suggesting that cross-sectional agreement is a strong signal for market timing. In the third essay, we explore the cross-section of Brazilian hedge fund returns using a novel measure of risk factor timing intensity - the tracking error between returns implied by estimated time-varying versus assumed-constant factor exposures. Fama-MacBeth regressions reveal that timing intensity significantly predicts monthly excess returns, robust to controlling for fund characteristics and exposure to conventional risk factors. This finding extends the active share literature to hedge funds and provide actionable insights for fund selection.