Refine
Document Type
- Working Paper (5)
- Article (1)
Language
- English (6)
Has Fulltext
- yes (6)
Is part of the Bibliography
- no (6)
Keywords
We introduce Implied Volatility Duration (IVD) as a new measure for the timing of uncertainty resolution, with a high IVD corresponding to late resolution. Portfolio sorts on a large cross-section of stocks indicate that investors demand on average about seven percent return per year as a compensation for a late resolution of uncertainty. In a general equilibrium model, we show that `late' stocks can only have higher expected returns than `early' stocks if the investor exhibits a preference for early resolution of uncertainty. Our empirical analysis thus provides a purely market-based assessment of the timing preferences of the marginal investor.
We show that time-varying volatility of volatility is a significant risk factor which affects the cross-section and the time-series of index and VIX option returns, beyond volatility risk itself. Volatility and volatility-of-volatility measures, identified model-free from the option price data as the VIX and VVIX indices, respectively, are only weakly related to each other. Delta-hedged index and VIX option returns are negative on average, and are more negative for strategies which are more exposed to volatility and volatility-of-volatility risks. Volatility and volatility of volatility significantly and negatively predict future delta-hedged option payoffs. The evidence is consistent with a no-arbitrage model featuring time-varying market volatility and volatility-of-volatility factors, both of which have negative market price of risk.
Predictability and the cross-section of expected returns: a challenge for asset pricing models
(2020)
Many modern macro finance models imply that excess returns on arbitrary assets are predictable via the price-dividend ratio and the variance risk premium of the aggregate stock market. We propose a simple empirical test for the ability of such a model to explain the cross-section of expected returns by sorting stocks based on the sensitivity of expected returns to these quantities. Models with only one uncertainty-related state variable, like the habit model or the long-run risks model, cannot pass this test. However, even extensions with more state variables mostly fail. We derive criteria models have to satisfy to produce expected return patterns in line with the data and discuss various examples.
Non-standard errors
(2021)
In statistics, samples are drawn from a population in a data-generating process (DGP). Standard errors measure the uncertainty in sample estimates of population parameters. In science, evidence is generated to test hypotheses in an evidence-generating process (EGP). We claim that EGP variation across researchers adds uncertainty: non-standard errors. To study them, we let 164 teams test six hypotheses on the same sample. We find that non-standard errors are sizeable, on par with standard errors. Their size (i) co-varies only weakly with team merits, reproducibility, or peer rating, (ii) declines significantly after peer-feedback, and (iii) is underestimated by participants.
Standard applications of the consumption-based asset pricing model assume that goods and services within the nondurable consumption bundle are substitutes. We estimate substitution elasticities between different consumption bundles and show that households cannot substitute energy consumption by consumption of other nondurables. As a consequence, energy consumption affects the pricing function as a separate factor. Variation in energy consumption betas explains a large part of the premia related to value, investment, and operating profitability. For example, value stocks are typically more energy-intensive than growth stocks and thus riskier, since they suffer more from the oil supply shocks that also affect households.
When estimating misspecified linear factor models for the cross-section of expected returns using GMM, the explanatory power of these models can be spuriously high when the estimated factor means are allowed to deviate substantially from the sample averages. In fact, by shifting the weights on the moment conditions, any level of cross-sectional fit can be attained. The mathematically correct global minimum of the GMM objective function can be obtained at a parameter vector that is far from the true parameters of the data-generating process. This property is not restricted to small samples, but rather holds in population. It is a feature of the GMM estimation design and applies to both strong and weak factors, as well as to all types of test assets.
