Research

We focus on the fundamental role of security analysts as information intermediaries using recent advances in the realized variance literature. We construct a signal-to-noise volatility ratio to examine the heterogeneity in the efficiency contributions of analysts' recommendations while controlling for the noise contained in price data. We find that only analysts' revisions with greater efficiency contributions generate significant stock price reactions in the directions expected by the analysts. Furthermore, these revisions increase the degree of informed trading in the options market and reduce the uncertainty related to the covered firms.

Using a recently developed measure of financial market risk perceptions, we show that risk perceptions affect firm-level corporate financing behavior. Firms tend to adjust their capital structures to cater to investors’ appetite for risk. When perceived risks are low, firms tend to choose more leveraged capital structures to take advantage of higher valuations associated with higher risk exposure. When perceived risks are high, firms tend to deleverage to avoid undervaluation associated with higher risk exposure. Furthermore, in periods of low risk perceptions, bond issue announcement returns tend to be higher, whereas long-run returns tend to decline with leverage.

This paper develops a novel functional predictive regression framework linking option-implied distributions to stock market returns, motivated by the fundamental link between risk-neutral and physical densities. Using S&P 500 option panels, our model exhibits significant forecasting power, achieving robust out-of-sample R-squared exceeding 4% for monthly return predictions, outperforming traditional predictors. Superior performance arises from leveraging the full spectrum of the risk-neutral density via functional principal components. Our analysis reveals that forecasting success stems from nuanced variations in risk-neutral densities, reflecting various economic states of risk perception and trading frictions, and demonstrates potential economic gains through a market-timing strategy.

We propose a novel methodology to integrate firm characteristics into the estimation of systematic diffusive and jump risks using high-frequency return panels, ultimately to identify mispricing that is orthogonal to the decomposed risks. Applying our framework to U.S. equities, we obtain characteristic-based diffusive and jump factor betas and discover mispricing. We attribute diffusive risk, jump risk, and mispricing to distinct characteristic clusters, revealing differences in their sources and time variation. The resulting intraday arbitrage portfolio delivers large risk-adjusted returns. We show that our characteristics-based diffusive and jump factors command significant risk premia, with signed jump risks playing economically distinct roles.

The efficiency impact of passive investment on stock prices is conditional on existing active mutual fund ownership. Using annual Russell 1000/2000 index reconstitutions as an instrument for passive ownership, we show that instrumented increases in passive holdings produce economically large improvements in price efficiency---lower intraday pricing errors, weaker return autocorrelation, and faster information incorporation---in firms with substantial active ownership, but have negligible effects where active ownership is limited. This complementarity is specific to active ownership: it is not replicated when conditioning on analyst coverage, trading activity, or institutional breadth, and only the active ownership interaction retains significance when all proxies are nested simultaneously. Consistent with an information-intermediation channel, analyst coverage increases, forecast dispersion declines, and post-earnings-announcement drift attenuates exclusively in high-active-ownership firms. These findings reconcile conflicting evidence on passive investing and price informativeness.

The maturity structure of crash insurance pricing forecasts equity returns and contains information beyond the variance risk premium level that is consistent with time-varying pricing of crash-risk persistence. A framework with time-varying persistence pricing generates three predictions that find support in the data: predictability is confined to the tail corridor, the slope subsumes the level in joint estimation, and statistical power concentrates in stressed markets. The slope's stronger forecasting performance relative to the tail VRP level is primarily attributable to genuine term structure content. A channel decomposition shows that the level’s predictive content loads primarily on investor sentiment proxies, while the slope is partially linked to intermediary constraints at longer horizons, consistent with dealer funding conditions shaping the pricing of long-dated tail protection. Even after conditioning on observable channels, the slope retains significant forecasting power and produces a quintile spread of 19.9% per year.

[Abstract text — please add]