FRM round the corner
21 Nov 2008
Learn Finance with the pros. Better articles, resources and screencasts for easier learning.
FRM |
Monte Carlo simulation is the preferred way to estimate counterparty risk. This owes to differences between lending and counterparty risk (i.e., bilateral contract, unknown future exposure, netting arrangements) that make counterparty risk, in a nutshell, more complex. Monte Carlo is powerful but susceptible to serious model risk. Specifically, what "engine" will be used to model the future behavior of an instrument?
Canabarro (in the FRM assigned Measuring and marking counterparty risk) lists common stochastic processes for various instruments:
These are not specific, just vague categories. For equities, from Hull, we know geometric Brownian motion (GBM) is popular, which gives normal periodic returns and therefore lognormal price levels. For interest rates, there are at least three one-factor models:
The EditGrid below simulates Vasicek process: ten Monte Carlo trials (one per column across) and 120 time steps in the rows going down (10 years x 12 months = 120 discrete steps). The inputs in yellow; note the ‘equilibrium rate' and the ‘strength of mean reversion' conspire to pull the rate gravitationally toward 6%:
Editgrid:
21 Nov 2008
20 Nov 2008
20 Nov 2008
Comments
This post is helpful. Thanks for the sample.
Leave a Comment