A PARSIMONIOUS MULTI-ASSET HESTON MODEL: CALIBRATION AND DERIVATIVE PRICING

Abstract

We propose a parsimonious multi-asset Heston model and provide an easy-to-implement calibration algorithm. The model is customized to pricing multi-asset options in markets with liquidly traded single-asset options but no liquidly traded cross-asset options. In this situation, single-asset model parameters can be calibrated from option price data, however, cross-asset parameters cannot. We formulate a parsimonious model specification such that all single-asset models are Heston models, which are affine allowing for efficient calibration of the respective parameters. The single-asset models are correlated using cross-asset correlations only. Cross-asset correlations are observable, in contrast to correlations of latent variables such as volatilities, and serve as basis for calibration. A hybrid calibration approach for identifying the model parameters consistent with option price data and asset price data is outlined and illustrated by a case study. In banking practice the approach is referred to as correlation adjustment.