Transform MCMC schemes for sampling intractable factor copula models

In financial risk management, modeling dependency within a random vector is crucial and a standard approach is the use of a copula model. A flexible family of copulas, known as the factor copulas, is formed by the copulas extracted from factor models. Sampling from a factor copula is equivalent to sampling from the factor model and applying the cumulative distribution function (c.d.f.) to each component of the sample. Nonetheless, in many models of interest the c.d.f.’s are not explicitly known. In this talk I’ll present theoretical and numerical properties of a transform Markov Chain Monte Carlo (MCMC) scheme developed to efficiently compute expectations conditional to rare events in which the unconditional distribution is given by an intractable factor copula.

Bio
Targino is an Assistant Professor (Lecturer) in Statistics at the School of Applied Mathematics (EMAp), at the Getulio Vargas Foundation (FGV). Targino is currently on a sabbatical from FGV EMAp, working as a Visiting Associate Professor at the Department of Statistics and Applied Probability (PSTAT) at the University of California, Santa Barbara (UCSB). Since 2021, he is an Associate Editor of the Brazilian Review of Finance (RBFin). From 2012 to 2016, he was a PhD candidate at the Department of Statistical Science, at the University College London (UCL). After a completing a BSc in Applied Mathematics and a MSc in Statistics (both from Federal University of Rio de Janeiro, Brazil) Targino spent 2.5 years working at the financial industry in Brazil, first as a Credit Risk Modeling Analyst at Itaú-Unibanco Bank and then as a Market Risk Analyst at Credit-Suisse Hedging-Griffo. During the Masters he also collaborated on a IMPA / Petrobras research project, mainly focused on Real Options problems. 

personal website: https://rtargino.netlify.app