The LIBOR Market Model (LMM), also known as the Brace-Gatarek-Musiela (BGM) model, is a popular interest rate model used for pricing interest rate derivatives. It is based on the evolution of forward rates through time, represented by the LIBOR (London Interbank Offered Rate) rates, which act as key reference rates in the global capital market. The main advantage of the LMM is that it accurately captures the market’s term structure dynamics, making it well-suited for pricing complex interest rate derivatives.
In the LMM, the dynamics of the forward rates are driven by a set of stochastic processes. The model is built on a series of forward LIBOR rates L(t, Ti, Ti + 1), where Ti and Ti + 1 are the start and end dates for the i-th accrual period, and t is the current time. This makes LMM a discretely tenored model in contrast to continuous tenored short-rate models such as Hull-White or Black-Derman-Toy models.
To derive the dynamics of the L(t, Ti, Ti + 1), we first express it in terms of a forward bond price ratio:
$$L(t, T_i, T_{i+1}) = \frac{1}{\delta_i} \left( \frac{P(t, T_i)}{P(t, T_{i+1})} - 1 \right), \quad \delta_i = T_{i+1} - T_i$$
Here P(t, Ti) represents the price of a zero-coupon bond at time t maturing at Ti. Next, we assume that the forward bond prices are driven by a stochastic process under a risk-neutral measure Q. The dynamics of the forward bond price ratio under a risk-neutral measure is given by:
$$\frac{dP(t, T_i)}{P(t, T_i)} = -r(t)dt + \sum_{k=1}^{n} \sigma_{ik}(t) dW_t^k$$
Where r(t) is the short rate, σik(t) are the forward rate volatilities, and dWtk are the increments of a set of n Brownian motions under the Q-measure.
To derive the dynamics of the LIBOR rate, we use Itô’s lemma for the quotients of two stochastic processes:
$$\begin{aligned}
d \left (\frac{P(t, T_i)}{P(t, T_{i+1})}\right ) &= \frac{1}{P(t, T_{i+1})} dP(t, T_i) - \frac{P(t, T_i)}{P(t, T_{i+1})^2} dP(t, T_{i+1}) \\
&\phantom{=} + \frac{\sigma_{i+1}(t, T_{i+1})\sigma_{i}(t, T_i)}{P(t, T_{i+1})^2} dt\end{aligned}$$
Substituting the dynamics for the forward bond prices we derived above, we get the stochastic process for the forward LIBOR rate under Q-measure:
$$\begin{aligned}
dL(t, T_i, T_{i+1}) &= L(t, T_i, T_{i+1}) \times\\
&\phantom{=} \left [ \sum_{k=1}^{n}\left (\sigma_{ik}(t) - \sigma_{i+1,k}(t) \right ) dW_t^k + \frac{1}{\delta_i} \sum_{k=1}^{n} \sigma_{i+1,k}(t)\sigma_{ik}(t) dt \right ]\end{aligned}$$
Now, we have a system of stochastic differential equations for the forward LIBOR rates. The LMM assumes that the forward rate volatilities σik(t) are deterministic functions of time, which implies that the LIBOR rates in the LMM are lognormally distributed. The choice of the volatility functions depends on the specific requirements of the model and the available market data, such as swaption prices and caplet prices.
In summary, the LIBOR Market Model is a flexible and powerful framework for modeling the evolution of forward interest rates and pricing interest rate derivatives. By capturing the dynamics of the entire term structure, it allows for the accurate pricing of complex interest rate products, such as interest rate swaps, swaptions, caplets, and other exotic derivatives. The model’s main limitation lies in determining the appropriate volatility functions and calibrating the model to market data, which often requires sophisticated numerical techniques, such as Monte Carlo simulations or finite difference methods.