CMStatistics 2017: Start Registration
View Submission - CFE
Title: Asset pricing in the quantile-frequency domain Authors:  Matej Nevrla - Czech Academy of Sciences (Czech Republic) [presenting]
Jozef Barunik - UTIA AV CR vvi (Czech Republic)
Abstract: Despite many attempts to understand cross-section of asset returns, there is no consensus on the functional form of the pricing kernel, cornerstone of the asset pricing theory. Many simplifying assumptions, such as quadratic preferences in CAPM model, led to results not supported by the real-world data. In the recent literature, two fruitful approaches emerged trying to explain risk premium of the assets. One with emphasis on asymmetric features of the asset returns based on the notion that agents put more weight on downside risk then upside potential when pricing an asset. The other approach stresses that risk aggregated over all investment horizons is not an adequate measure and emphasises the importance of frequency-specific risk and its implications for agents caring differently over various economic cycles. We aim to unify these two approaches and show that downside risk possesses complex structure and differs over various cycles. We do this by defining quantile cross-spectral betas that capture exposure to the downside risk over different investment horizons. Downside risk in our approach is characterized by the quantile dependency, which is further decomposed into frequency domain. We apply it to the wide range of asset classes and asses how the model jointly prices them.