By the Numbers
Comparing inverse IO to interest rate floors
Chris Helwig | October 25, 2024
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors. This material does not constitute research.
Investors looking to add key rate duration on the front end of the yield curve can do that in a number of different ways, including through exposure to inverse IO in MBS or interest rate floors in the options markets. Similarities between these two approaches invite a comparison to assess which one is more efficient either for taking a view on lower rates or hedging floating-rate exposures.
Constructing an inverse IO
CMO structuring creates an inverse interest-only bond, or inverse IO (IIO), by splitting a fixed-rate pass-through or CMO into two components:
- A floating-rate principal and interest bond, and
- An IIO that gets any interest leftover after paying the floating-rate coupon.
The floater typically gets all the par principal and gets interest up to a cap, which ensures enough cash flow from the fixed-rate collateral to pay the floating-rate coupon. Without a cap, the floating coupon could rise to a level where it exceeded the interest available from the fixed-rate bond. As a result, the IIO is equivalent to a position in the fixed-rate collateral with maximum structural leverage or financing through the sale of the floater. The IIO coupon consequently is floored at zero, when the floater is getting all the interest, and reaches its maximum when the floater is getting its lowest coupon. That maximum equals the floating-rate cap less the stated margin on the floater, commonly referred to as the ‘strike.’
Comparing inverse IO to interest rate floors
As an inverse IO is simply a stream of interest payments calculated using a notional balance and a strike it is somewhat akin to an interest rate floor. An interest rate floor is a derivative contract that pays the buyer of that contract a stream of interest payments calculated as the difference between an agreed upon strike and a spot benchmark index rate based on a fixed notional balance to a specified maturity date. The floor buyer will pay an up-front premium for the contract, which may or may not be entitled to current cash flow depending on whether the strike is in or out-of-the-money versus the spot benchmark rate. The floor is identical to the inverse IO in that its coupon will reach a maximum terminal value if the benchmark rate is zero and will receive no coupon if the spot benchmark rate is above the strike. For example, both an IIO and a floor with a 5.35% strike will receive the same coupon relative to the benchmark rate (Exhibit 1).
Exhibit 1: Floor and Inverse IO coupon across benchmark rates
Comparing inverse IO to interest rate floors
The critical difference between exposures is both the notional and maturity of the floor will be well defined, while those elements for an inverse IO are subject to the level of prepayments. Given this, depending on the shape of the yield curve into a rally, the protection afforded by an inverse IO may be substantially less than that of an interest rate floor.
If long rates fall in conjunction with lower front-end floating rates, prepayment rates on the collateral underpinning the inverse IO will rise and the notional balance of the bond will drop more rapidly (Exhibit 2). Given this, while both IIO and floors with the same strike will generate the same coupon at a given SOFR rate, the inverse IO will return less absolute cash flow to the bond holder into a rally. As a result, the cost or premium on floors will be higher than prices of IIO with the same strike given the fact that the IIO holder is effectively receiving a premium for selling the prepayment option.
Exhibit 2: A comparison of IIO and floor notional balances
A framework for valuation
Comparing the two instruments involves pricing the inverse IO to multiple prepayment scenarios. In lieu of using a series of fixed prepayment assumptions, a prepayment model is dialed from 70% to 130% of the base case projection in 10% increments to derive an average life of the bond. The base case average life of the bond is 4.4 years. The bond extends to a 6.3 year at 70% of model and shortens to a 3.7 year at 130% of the model.
After deriving the various average lives of the inverse IO, a series of floors with the same strike as the IIO can be constructed. The floors’ maturities will match the average lives of the inverse IO given the different prepayment assumptions. Using bullet notionals will generate a mismatch between the floors and the inverse IO but the overstatement of the balance of the floor relative to the inverse IO on the front end of the average life should be, in large part, offset by the understatement of the balance on the back half. This, in turn, should allow for a reasonable comparison of the two instruments.
Comparing prices across prepayment and interest rate shocks
Evaluating the relationship between prices of the inverse IO to the comparable floors shows the disparity between the two is most pronounced to slower prepayment assumptions and longer average lives and converges substantially to faster ones. Convergence around shorter average lives is largely a function of two factors. First, the price of the inverse IO begins to flatten out as it approaches terminal prepay assumptions where it is not likely to be further curtailed. Additionally, the price on the floor falls substantially as the tenor of protection declines and exposure to longer tenors of volatility are no longer being priced into the cost of the option (Exhibit 3).
Exhibit 3: Comparing IIO and floor prices across average lives
While in all scenarios, inverse IO is nominally cheap to floors, there is likely a more nuanced way to value the relationship between the two instruments. Changes in prepayment assumptions drive not only changes in speeds and average lives but also the option cost of the inverse IO with the option cost increasing substantially as the prepay model is dialed faster. Arguably, this suggests that as option cost rises on the IIO but the price differential between the two instruments compresses, that investors may elect exposure to inverse IO rather than buying a floor.
For example, at 100% of the prepayment model, a 5.35% strike inverse IO backed by G2SF 6.5% collateral has an OAS of 1450 bp and an option cost of 510 bp at a price of $3.9. Increasing the prepayment model multiplier to 130% of the model and repricing using a constant OAS yields a price of $3.3 and an option cost roughly 670 bp. Given the higher option cost, the price between the two instruments compresses materially as the value of the option the investor is shorting in the inverse IO increases. Against the backdrop of elevated prepayment assumptions, investors may choose to use inverse IO as the cash flow:
- Has already been materially curtailed.
- Prices more comparably to a short, bullet floor.
- May have optionality to slower speeds, cash flow extension and price appreciation.
An alternate approach
The above analysis uses a prepay model multiplier to both curtail and extend the inverse IO cash flow and construct and price floors with maturities that match the average lives across scenarios. An alternative approach would be to observe the price relationship between the inverse IO to that solely of a 4.4-year maturity floor across a series of interest rate shocks. Under this framework, the price relationship between the two instruments begins to decouple as rates rally and decouples materially given a 200 bp rally in rates as the difference in notional balance and average life between the two becomes more substantial. Given this, investors who believe rates will rally significantly from current levels may choose to buy longer dated floors rather than inverse IO. (Exhibit 4)
Exhibit 4: Floor and inverse IO pricing across rate shocks
Breaking down partial duration and curve exposure
The presence of the short prepayment option in inverse IO manifests itself when evaluating differences in curve exposure across the two instruments as well. Looking at both overall effective and partial duration of an inverse IO and floor with the same strike and average lives shows that the floor will exhibit longer overall effective duration. This is, in part, because the duration of the inverse IO will be the net of positive duration to front end key rates and negative duration to longer key rates, akin to traditional IO which will exhibit negative duration resultant from price appreciation as interest rates rise and the cash flow extends (Exhibit 5).
Exhibit 5: Evaluating key rate durations across exposures
The amount of front-end key rate duration exhibited by the floor will, in large part, be driven by the tenor of the instrument. The example above matches the tenor of the floor to the base case average life of the inverse IO, 4.4 years and as a result maintains a substantial amount of 5-year key rate duration. Front end key rate duration in inverse IO will largely be a function of the strike. Lower strike inverse IO will generally exhibit greater amounts of front-end key rate duration as they will experience greater price appreciation as the strike becomes in-the-money than higher strike bonds whose strikes are already in-the-money.
Differences in curve exposure should drive investor decisioning when choosing between inverse IO and interest rate floor exposures. Given a set path of forward rates, coupons on both instruments will be the same given the same strike. However, investors that are of the opinion that long rates may stay relatively elevated while short end rates decline may prefer inverse IO to owning floors given additional potential for price appreciation.