I\u2019m sure we\u2019ve all been there: We\u2019re in hot competition with another bank over a deal. As the deal evolves, our trading team starts getting pushed around the market, and it dawns on us that the other bank\u2019s pricing is better than ours, at least for this class of deals. We could fix this problem by inventing a universal method for achieving the best possible prices for all deal types. That topic will be covered in a future column, next to the column on Elvis sightings. Here we focus on a sin- gle class of deals, the constant maturity swaps, caps, and floors. We develop a framework that leads to the standard methodology for pric- ing these deals, and then use this framework to systematically improve the pricing.

We distinguish between zero coupon bonds and discount factors to remind ourselves that discount factors are not random, we can always obtain the current discount factorsD( T)by stripping the yield curve, while zero coupon bondsZ( t, T) remain random until the present catches up to datet. We also use

val, not the start date, and the fixing date\u03c4j forRj is spot lag business days before the intervale n d s attj . As before,\u03b4j is the coverage for thejth interval using the day count basisdcbpay specified in the contract. Stan- dard practice is to use the 30360 basis for USD CMS legs.

The value of the CMS swap, cap, or floor is just the sum of the values of each payment. Any margin paymentsmcan also be valued easily. So all we need do is value a single payment of the three types,

Here the reference rateRsis the par rate for a standard swap that starts at dates0, and ends N years later atsn. To express this rate mathe- matically, lets1, s2, . . . , sn be the swap\u2019s (fixed leg) pay dates. Then a swap with rateRfix has the fixed leg payments

is the coverage (fraction of a year) for each periodj, anddcbsw is the stan- dard swap basis. In return for making these payments, the payer receives the floating leg payments. Neglecting any basis spread, the floating leg is worth1paid at the start dates0 , minus 1paid at the end datesn . At any datet, then, the value of the swap to the payer is

Crudely speaking, the levelL( t) represents the value at timet of receiving $1 per year (paid annually or semiannually, according to the swap\u2019s fre- quency) forN years. With this definition, the value of the swap is

According to the theory of arbitrage free pricing, we can choose any freely tradeable instrument as ourn u m e ra i re. Examining 1.8 shows that the levelL( t) is just the value of a collection zero coupon bonds, since the coverages\u03b1j are just fixed numbers. These are clearly freely tradeable instruments, so we can choose the levelL( t) as our numeraire.1 The usual theorems then guarantee that there exists a probability measure such that the valueV( t) of any freely tradeable deal divided by the numeraire is a Martingale. So

It is helpful to examine the valuation of a plain vanilla swaption. Con- sider a standard European option on the reference swap. The exercise date of such an option is the swap\u2019s fixing date\u03c4 , which is spot-lag busi- ness days before the start dates0. At this exercise date, the payoff is the value of the swap, provided this value is positive, so

Moreover, 1.9b shows that the par swap rateRs( t) is the value of a freely tradable instrument (two zero coupon bonds) divided by our numeraire. So the swap rate must also a Martingale, and

To complete the pricing, one now has to invoke a mathematical model (Black\u2019s model, Heston\u2019s model, the SABR model,. . .) for howRs (\u03c4 ) is distributed around its mean valueR0s . In Black\u2019s model, for example, the swap rate is distributed according to

On the swap\u2019s fixing date\u03c4, the par swap rateRs is set and the payoff is known to be[Rs(\u03c4 )\u2212 K ]+Z(\u03c4; tp), since the payment is made ontp. Evalu- ating 2.1 atT=\u03c4 yields

There are two steps in evaluating the convexity correction. The first step is tomodel the yield curve movements in a way that allows us to re- write the level L(\u03c4 )and the zero coupon bond Z(\u03c4; tp)in terms of the swap rateRs. (One obvious model is to allow only parallel shifts of the yield curve.) Then we can write

In the appendix we start with the street-standard model for express- ingL(\u03c4 ) andZ(\u03c4; tp) in terms of the swap rateRs. This model uses bond math to obtain

is the fraction of a period between the swap\u2019s start dates0and the pay datetp. For deals \u201cset-in-arrears\u201d\ue000=0. For deals \u201cset-in-advance,\u201d if the CMS leg datest0, t1, . . . are quarterly, thentp is 3 months after the start dates0 , so\ue000=12 if the swap is semiannual and\ue000=14 if it is annual.

In the apprendix we also consider increasingly sophisticated models for expressingL(\u03c4 ) andZ(\u03c4; tp) in terms of the swap rateRs, and obtain increasingly sophisticated functionsG( Rs).

We can carry out the second step byreplicating the payoff in 2.12 in terms of payer swaptions. For any smooth functionf(Rs) withf( K)=0, we can write