Integrating multi-market risk models

Peter G. Shepard

Abstract

ABSTRACT

We present a method of consistent integration of factor models across multiple markets. By aggregating successively broader models, it allows the inclusion of many more risk factors than would normally be possible with a limited time series of data. The method achieves local market consistency by exploiting natural corrections to factor exposures, providing advantages over a previous approach to integrated models. The framework is also extended to allow specific cross-market correlations among individual market factors, taking into account the relationships not captured by the coarse-grained global model.