12:00
Ignorance-Fuelled Securitization Booms
Pascal Golec (University of Amsterdam)

******

13:00
Comparing Density Forecasts in a Risk Management Context
Hao Fang (University of Amsterdam)

This paper develops a testing framework for comparing the accuracy of competing densities of aggregated marginal variables in the downside part of the support. Three proper scoring rules including conditional likelihood, censored likelihood and penalized weighted likelihood are used for assessing the predictive ability of (out-of-sample) densities, closely related to the Kullback-Leibler information criterion (KLIC). We consider distributions in the framework of skew-elliptical family which is analytically traceable under affine transformations. The common practice of forecast comparison in high-dimensional space is problematic because that a better forecast in multivariate evaluation does not necessarily correspond to a better portfolio return forecast; as illustrated by examples in the paper. An application to the daily returns of three US stocks suggests that the Student-t outperforms the Normal, Skew t and Skew Normal assumptions in the left tail of the portfolio return. The visualized dynamics of our test statistic provides a side proof for regime change over last thirty years.