Research

In this paper, we examine the finite sample properties of test statistics designed to identify distinct underlying components of high-frequency financial data, specifically the Brownian component and infinite vs. finite activity jumps. We conduct a comprehensive set of Monte Carlo simulations to evaluate the tests under various types of microstructure noise, price staleness, and different levels of jump activity. We apply these tests to a dataset comprising 100 individual S&P 500 constituents from diverse business sectors and the SPY (S&P 500 ETF) to empirically assess the relative magnitude of these components. Our findings strongly support the presence of both Brownian and jump components. Furthermore, we investigate the time-varying nature of rejection rates and we find that periods with more jump days are usually associated with an increase in infinite jumps and a decrease in finite jumps. This suggests a dynamic interplay between jump components over time.

This paper examines the impact of intraday periodicity on forecasting realized volatility using a heterogeneous autoregressive model (HAR) framework. We show that periodicity inflates the variance of the realized volatility and biases jump estimators. This combined effect adversely affects forecasting. To account for this, we propose a periodicity-adjusted HAR model, HARP, where predictors are constructed from the periodicity-filtered data. We demonstrate empirically (using 30 stocks from various business sectors and the SPY for the period 2000–2020) and via Monte Carlo simulations that the HARP models produce significantly better forecasts across all forecasting horizons. We also show that adjusting for periodicity when estimating the variance risk premium improves return predictability.

We propose a novel approach to decompose realized jump measures by type of activity (finite/infinite) and sign, and also provide noise-robust versions of the ABD jump test (Andersen et al., 2007b) and realized semivariance measures. We find that infinite (finite) jumps improve the forecasts at shorter (longer) horizons; but the contribution of signed jumps is limited. As expected, noise-robust measures deliver substantial forecast improvements at higher sampling frequencies, although standard volatility measures at the 300-s frequency generate the smallest MSPEs. Since no single model dominates across sampling frequency and forecasting horizon, we show that model averaged volatility forecasts – using time-varying weights and models from the model confidence set – generally outperform forecasts from both the benchmark and single best extended HAR model. Finally, forecasts using volatility and jump measures based on transaction sampling are inferior to the forecasts from clock-based sampling.

This paper introduces a novel class of volatility forecasting models that incorporate market realized (co)variances and semi(co)variances within the framework of a heterogeneous autoregressive (HAR) model. Our empirical analysis shows statistically and economically significant forecasting gains. For our most parsimonious market-HAR specification, stock volatility forecasting is improved by 9.80% points. Using a mixed sampling frequency market-HAR variant with low (high) sampling frequency for the stock (market) improves forecasting by a further 6.90% points. Our paper also develops noise-robust estimators to facilitate the use of realized semi(co)variances at high sampling frequencies.

This paper develops a Bayesian Wald-type test for detecting differences in network connections from variance decompositions over time. The test statistic asymptotically follows a χ2 distribution under the null hypothesis and performs well in Monte Carlo simulations. We apply our methodology to high-frequency options data on sectoral ETFs from January 2013 to June 2023 to identify monetary policy surprises. Our results show that total connectedness among sectoral ETF implied volatilities increases significantly following FOMC announcements, with the differences persisting for up to an hour on surprise days. Our empirical findings shed light on the impact of intra-daily monetary policy shocks on financial markets.

We decompose aggregate market variance into average downside and upside variances and correlations. Average downside implied correlation drives cross-sectional volatility risk pricing. Stocks with high downside correlation risk exposure earn 7.30\% lower annual returns than low-exposure stocks, consistent with providing insurance against diversification risk. This factor emerges as the only pervasive component across all variance elements during volatile periods. Our findings show that downside correlation risk represents a more fundamental driver of diversification risk premiums than previously identified factors. Mean-variance investors achieve substantial economic gains by incorporating the average downside correlation factor into their portfolio optimization framework.

We document asset pricing implications of the new zero days-to-expiration (0DTE) options, which today account for half of the total S&P 500 option volume. We show that: (i) most of the intra-day equity premium is attributable to market returns between -5% and 0%; (ii) investors demand a high compensation to bear variance risk over the day, which is mainly due to compensation for upside risk; (iii) the variance risk premium predicts intra-day market returns, with a negative relation that is driven by the upside risk premium; and (iv) 0DTE options violate stochastic dominance restrictions, where exploiting this relative mispricing is highly profitable. Our findings contrast with evidence from longer horizons and are consistent with a nonmonotonic pricing kernel that is especially high for positive market returns.

We propose option realized semivariances and signed jumps which can be seen as new “observable quantities” to summarize the asymmetric information contained in the sign of high-frequency options returns. We show that these measures successfully capture the direction of the discontinuities related to the underlying asset and risk factor, resulting in additional incremental information neither contained in the aggregate option realized measures nor in other end-of-day stock and options data. In specific, using options data on SPDR S&P 500 ETF (SPY) and 15 individual equities, we find that the negative (positive) semivariance and signed jump of out-of-the-money call (put) options play a prominent role in predicting future variance, variance risk-premia, and excess monthly returns.

We combine high-frequency stock returns with risk-neutralization to extract the daily common component of tail risks perceived by investors in the cross-section of firms. Our tail risk measure significantly predicts the equity premium and variance risk premium at short-horizons. Furthermore, a long-short portfolio built by sorting stocks on their recent exposure to tail risk generates abnormal returns with respect to standard factor models. Incorporating investors’ preferences via risk neutralization is fundamental to our findings: the predictive power of the physical tail risk is weaker and generally subsumed by its risk-neutral counterpart.

This paper proposes a robust framework for disentangling undiversifiable common jumps within the realized covariance matrix. Simultaneous jumps detected in our empirical study are strongly related to major financial and economic news, and their occurrence raises correlation and persistence among assets. Our application to 20 Dow Jones stocks, shows that common jumps and directional common jumps substantially improve the in- and out-of-sample forecasts of the realized covariances at the day-, week- and month-horizon. Applying these new specifications to minimum variance portfolios results in superior positions from reduced turnover. The implication is that investors willingly sacrifice up to 100 annual basis points in switching to those strategies.

We propose a dilution bias correction approach to deal with the errors-in-variables problem observed in realized volatility (RV) measures. The absolute difference between daily and monthly RV is shown to be proportional to the relative magnitude of the measurement error. Therefore, in implementing the latter metric, and in allowing the daily autoregressive parameter to vary as a function of the error term, the result is more responsive forecasts with greater persistence (faster mean-reversion) when the measurement error is low (high). Empirical results indicate that our models outperform some of the most popular univariate and multivariate HAR and GARCH models across various forecasting horizons.