Risk-neutral pricing in a behavioural framework

This thesis investigates three issues related to risk-neutral pricing. The first aspect investigated is the effect of discretization and truncation errors on risk-neutral moments, as defined in Bakshi, Kapadia and Madan (2003). It proposes exact solutions for the finite integrals in the volatility, cubic and quartic contracts and compares its accuracy approach with the interpolation-extrapolation approach. It yields more accurate estimates for risk-neutral skewness and kurtosis for those assets which exhibit the volatility smirk. By contrast, for those assets dominated by the forward skew, the exact approach outperforms the interpolation-extrapolation approach for skewness only. The second issue investigated is the skewness preference. It seeks to explain the positive skewness preference through heterogeneous beliefs and overconfidence. An overconfident group longs more skewness in the positively skewed portfolio, over-estimates the value of this portfolio, causes heterogeneous beliefs and yields a positive skewness preference. The final issue investigated is the relation between risk-neutral kurtosis and returns. The relation can be either positive or concave and cannot be explained bu heterogeneous beliefs and overconfidence. There are important causality effects between skewness and kurtosis and evidence is presented that the relation between kurtosis and returns may not be independent.