Option-implied risk-neutral densities are widely used for constructing forward-looking risk measures. Meanwhile, investor risk aversion introduces a multiplicative pricing kernel between the risk-neutral and true conditional densities of the underlying asset’s return. This paper proposes a simple local estimator of the pricing kernel based on inverse density weighting, and characterizes its asymptotic bias and variance. The estimator can be used to correct biased density forecasts, and performs well in a simulation study.

We develop a dynamic framework to detect the occurrence of permanent and transitory breaks in the illiquidity process. We propose various tests that can be applied separately to individual events and can be aggregated across different events over time for a given firm or across different firms. In an empirical study, we use this methodology to study the impact of stock splits on the illiquidity dynamics of the Dow Jones index constituents and the effects of reverse splits using stocks from the S&P 500, S&P 400 and S&P 600 indices.

We show that three prominent consumption-based asset pricing models—the Bansal–Yaron, Campbell–Cochrane and Cecchetti–Lam–Mark models—cannot explain the dynamic properties of stock market returns. We show this by estimating these models with GMM, deriving ex-ante expected returns from them and then testing whether the difference between realised and expected returns is a martingale difference sequence, which it is not. Mincer–Zarnowitz regressions show that the models’ out-of-sample expected returns are systematically biased.

A popular self-normalization (SN) approach in time series analysis uses the variance of a partial sum as a self-normalizer. This is known to be sensitive to irregularities such as persistent autocorrelation, heteroskedasticity, unit roots and outliers. We propose a novel SN approach based on the adjusted-range of a partial sum, which is robust to these aforementioned irregularities.

We investigate how to improve efficiency using regression adjustments with covariates in covariate-adaptive randomizations (CARs) with imperfect subject compliance. Our regression-adjusted estimators, which are based on the doubly robust moment for local average treatment effects, are consistent and asymptotically normal even with heterogeneous probabilities of assignment and misspecified regression adjustments.