This paper examines a new model of credit risk measurement, the Variance Gamma- Merton one, which seems to be adequate for describing single default occurrence and default correlation in turbulent times. It is based on the notion of business time. Business time runs faster than calendar time when the market is very active and a lot of information arrives; it runs at a slower pace than calendar time when few information arrives. We report a calibration to USA spread data, which shows the accurateness of the model at the single default level; we also compare the performance with a traditional structural model at the joint default level