We will develop an R package for two families of skew-t distributions that have different tail behavior for left and right tails, namely the family of the asymmetric t-distributions (AST) distributions introduced by Zhu and Galbraith (2010), and the family of generalized asymmetric t-distributions (GAT) introduced by Baker (2018). The importance of these two families is that they go beyond the symmetric tail behaviors of the skew-t distributions, as described in Azzalini and Capitanio (2014), and hence can provide better fits for certain data arising in applications, especially for asset returns. The resulting skew-t package st, will compute not only the skew-t MLEs but also basic computations of the skew-t probability density, cumulative distribution functions. Furthermore, the package will compute confidence intervals for the parameter estimates, and hypothesis tests concerning the parameters. Due to the complexity of the AST and GAT distributions, and the desire to use them in empirical asset pricing studies with large cross-sections (e.g., 1000 to 5000 stocks), we plan to use the Rcpp package to integrate C++ code to obtain high-performance MLE computations.

Student

Daniel Xia

Mentors

  • Richard Martin
  • Aleksandr Aravkin
  • Alexios Galanos
close

2019