Bermudan Swaptions

Interest Rate Modeling




Tutorial Contents

  1. Fixed Income Notation
  2. Fixed Income Probability Measures
  3. Multi-Currency Markets
  4. The HJM Framework
  5. Basic Fixed Income Instruments
  6. Caps, Floors and Swaptions
  7. CMS Swaps, Caps and Floors
  8. Bermudan Swaptions
  9. Structured Notes and Exotic Swaps
  10. Callable Libor Exotics and TARNs
  11. Volatility Derivatives
  12. Yield Curves and Curve Construction
  13. Managing Yield Curve Risk
  14. Vanilla Models With Local Volatility
  15. Vanilla Models With Stochastic Volatility
  16. Introduction to Term Structure Models
  17. One-Factor Short Rate Models I
  18. One-Factor Short Rate Models II
  19. Multi-Factor Short Rate Models
  20. The Libor Market Model I
  21. The Libor Market Model II
  22. Risk Management and Sensitivity Computations
  23. P&L Attribution
  24. Payoff Smoothing
  25. Vegas in the Libor Market Model


Bermudan swaptions are a common form of option on swaps. They are similar to standard vanilla swaptions, except the holder has the right to enter into a fixed-floating swap on any of a list of specified fixing dates up to expiry. For a given tenor structure

0T0<T1<<TN,τn=Tn+1Tn, 0 \le T_0 < T_1 < \cdots < T_N, \quad \tau_n = T_{n+1} - T_n,

the holder of a Bermudan swaption has the right to exercise on any of the dates {Tn}n=0N1\{T_n\}_{n=0}^{N-1}. Once exercised on date TnT_n, the option terminates, and the holder enters the swap with the first fixing date TnT_n and the final payment date TNT_N. The period leading up to T0T_0 is known as the lockout or no-call period. In common jargon, a Bermudan swaption on a 10 year swap with a 2 year lockout period is known as a 10 no-call 2 or 10nc2 Bermudan swaption.

Demand for Bermudan swaptions is driven by various segments of the fixed income market. Mortgage companies use them to hedge their mortgage pools - the flexibility that the opportunity to exercise at various times gives them allows them to match the uncertain timings of mortgage prepayments. In contrast, investors seeking higher upfront income sell Bermudan swaptions - the additional optionality acts in favour of the holders of Bermudan-style options, leading to higher premiums versus an otherwise equivalent European swaption. Bermudan swaptions are also used in order to hedge callable Libor coupon bonds.

Pricing Bermudan Swaptions

Formally, at time TnT_n, the value of a payer Bermudan swaption, if exercised, is the swap value

Un(Tn)=β(Tn)i=nN1τiETnQ[β(Ti+1)1(Li(Ti)K)]=i=nN1τiP(Tn,Ti+1)(Li(Tn)K)=An,Nn(Tn)(Sn,Nn(Tn)K). \begin{aligned} U_n(T_n) & = \beta(T_n) \sum_{i=n}^{N-1} \tau_i \mathbb{E}_{T_n}^{\mathbb{Q}}\left[ \beta(T_{i+1})^{-1} (L_i(T_i) - K) \right] \\ & = \sum_{i=n}^{N-1}\tau_i P(T_n, T_{i+1})(L_i(T_n) - K) \\ & = A_{n, N-n}(T_n)(S_{n,N-n}(T_n) - K). \end{aligned}

Loosely speaking, a Bermudan swaption is an option to choose between Un(Tn)U_n(T_n) for different n=0,,N1n=0,\ldots,N-1.

By no-arbitrage, the value V(Tn)V(T_n) of the Bermudan swaption at time TnT_n is then given by the maximum of Un(Tn)U_n(T_n) and the hold value Hn(Tn)H_n(T_n), the value of the Bermudan swaption with only the exercise dates {Ti}i=n+1N1\{T_i\}_{i=n+1}^{N-1} remaining. It is assumed that a rational investor at time TnT_n will observe the exercise value Un(Tn)U_n(T_n) along with the hold value Hn(Tn)H_n(T_n) and decide to exercise only if they stand to gain at that instant by doing so. Hence, we find that

V(TN)=UN(TN)=0,V(Tn)=max(Un(Tn),Hn(Tn)),n=0,,N1, \begin{aligned} V(T_N) & = U_N(T_N) = 0, \\ V(T_n) & = \text{max}(U_n(T_n), H_n(T_n)), \quad n=0,\ldots,N-1, \\ \end{aligned}

Meanwhile, the hold value Hn(Tn)H_n(T_n) simply represents the value V(Tn+)V(T_{n}^+) of the option at the instant after the choice not to exercise was made. For t(Tn,Tn+1)t \in (T_n, T_{n+1}), the investor has no ability to exercise, so we find that Hn(t)=V(t)H_n(t) = V(t) and hence, we have

Hn(Tn)=β(Tn)ETnQ[β(TTn+1)1V(Tn+1)] H_n(T_n) = \beta(T_n) \mathbb{E}_{T_n}^{\mathbb{Q}} \left[\beta(T_{T_{n+1}})^{-1} V(T_{n+1}) \right]

Combining the above yields

Hn(Tn)=β(Tn)ETnQ[β(Tn+1)1max(Un+1(Tn+1),Hn+1(Tn+1))],n=0,,N1,V(0)=β(0)EQ[β(T0)1V(T0)]=β(0)EQ[β(T0)1max(U0(T0),H0(T0))]. \begin{aligned} H_n(T_n) & = \beta(T_n) \mathbb{E}_{T_n}^{\mathbb{Q}} \left[\beta(T_{n+1})^{-1} \text{max}(U_{n+1}(T_{n+1}), H_{n+1}(T_{n+1})) \right], \quad n=0,\ldots, N-1, \\ V(0) & = \beta(0) \mathbb{E}^{\mathbb{Q}} \left[\beta(T_{0})^{-1} V(T_0) \right] \\ & = \beta(0) \mathbb{E}^{\mathbb{Q}} \left[\beta(T_{0})^{-1} \text{max}(U_0(T_0), H_0(T_0)) \right]. \end{aligned}

This defines a useful iteration backwards in time from the terminal time TNT_N, to the initial time 00. The idea behind this iterative scheme is known as dynamic programming. Loosely speaking, we work backwards to price Bermudan options. Hence, pricing Bermudan swaptions is a problem naturally suited to numerical schemes that proceed backwards through time, in particular, finite difference methods and tree-based methods fit this description well.

On the flipside of the above point, methods such as Monte Carlo pricing, which are inherently forward moving in time, are poorly suited to pricing Bermudan options, and more sophisticated numerical techniques must be considered to produce accurate pricing.

© 2022 Will Douglas, All Rights Reserved.