Social Sharing
Extended Viewer
On the White Board - May 2008
May 15, 2008
A few months back, we discussed the integration of market and credit risk (Research Monthly). One of our conclusions was that the challenge was not so much connecting two separately modeled sources of risk, but rather providing a model for total risk over a relatively long (one-year, for example) investment horizon. At this horizon, models for jointly describing market and credit risk exist; what does not exist, however, is a standard for how to model holdings over a long horizon, where questions of aging, reinvestment, cashflows and path dependence cannot be ignored. There are lots of right answers here, and lots of wrong ones as well, depending in particular on how we define risk or capital or the investment process. One effort of ours currently is to build up somewhat of a taxonomy of modeling assumptions, in order to provide appropriate alternatives and stimulate more discussion of the topic. In the process of thinking about this taxonomy, we have also realized that clean assumptions about how an instrument moves forward in time can be just as important for horizons as short as one day. In particular, we have seen that the commonly applied assumption of "instantaneous changes" is ill defined for path dependent products. Consequently, it is better to think in terms of "steady state" portfolios, a concept that is equivalent to instantaneous changes for simple products, but which generalizes nicely to the path dependent products.