LIBOR Market Model and Forward-Rate Simulation
The Brace–Gątarek–Musiela model describes forward rates for specified accrual periods and links them through a correlation structure. Historical cap and swaption implementations can be compared by testing discretisation error and whether simulated forward rates have the correct drift under the chosen measure. An overnight rate compounded in arrears has a different fixing convention, for which the forward market model provides a related framework.
How it works
The historical BGM model specifies rates for fixed accrual periods. Each forward is lognormal under its own forward measure in the basic version, which supports Black caplet pricing. Simulation under one common measure requires a drift adjustment and a structure for correlation between the forward rates.
Worked example
With a half-year accrual, a forward rate of 3% and strike of 2%, a caplet pays 0.5 × (0.03 − 0.02) = 0.005 per unit of notional if the modelled rate fixes there. Under the forward measure, the simulation and Black formula should agree on the expectation of that payoff across many draws.
Validation and limitations
Check each rate’s drift under the chosen simulation measure and reprice initial quotes. Watch discretisation error at long horizons and correlation sensitivity for swap options. The historical forward setting differs from current compounded in-arrears benchmarks. See the forward market model.
Further reading
- The Market Model of Interest Rate Dynamics
Alan Brace, Dariusz Gątarek and Marek Musiela · Journal record
- Looking Forward to Backward-Looking Rates
Andrei Lyashenko and Fabio Mercurio · Working-paper PDF
- The End of LIBOR (official benchmark history)
Financial Conduct Authority