Valuation Commentary - June 2006

The MBS Basis Risk
by Alex Levin

MBS basis risk is the risk of a widening or a tightening of an MBS spread to the swap (agency, Treasury, etc) benchmark curve. The most common measure of the basis is either the nominal spread or the OAS reported for a current-coupon MBS. In my opinion, the latter is the preferred measure as it removes an indirect volatility effect.

It is widely perceived that MBS investors are willing to bear mortgage-specific risk without quantifying it or hedging. This practice is now becoming less and less feasible. Vanilla products don't yield much over non-MBS notes, whereas mortgage derivatives are loaded with prepayment risk. Therefore, quantifying the risk and considering hedging is necessary to identify truly cheap instruments and "lock into OAS." Let us consider how the basis risk comes into the hedging game and how fundamentally the concept of prOAS (AKA prepay risk-neutrality) alters the traditional view.

Traditional View of the Basis Risk

According to the traditional view, hedging against the basis risk simply requires taking a delta-neutral position to the MBS spread to the benchmark curve. Hence, an IO or MSR investor is advised to add a long TBA (or similar) hedging position to the portfolio. If the MBS market rate tightens to swaps (with all else unchanged), the IO would lose due to prepay acceleration, but the gain of the TBA will offset the loss.

It does not matter what causes the basis tightening. Aside from the apparent prepayment effect, in some analytical systems, basis widening or tightening can have a direct additive or subtractive affect on the OAS of other MBS. The rationale behind this view is that the "entire mortgage market" widens or tightens. For example, agency OAS widening for the current-coupon TBA can follow accounting scandals of other credit events and, as such, should be propagated to other TBAs and agency MBS. In essence, the OAS level for the current-coupon MBS is employed as the starting point for the OAS of other MBS.

Events of 2003

In May of 2003 the rates fell to their 40-yr record; refinancing panic dominated the MBS market. In the AD&Co. model, interpolated CC LOAS widened to 27 bps, but premium MBS were at a progressively wider spread (http://www.ad-co.com/secure/market_analysis/2003/qa-5-30-2003.html). On the same day, Trust IOs got dramatically reduced to single-point values; OAS levels exploded to 1000 - 1500 bps. Consider MSR managers and hedge fund IO investors. They used the TBAs to hedge their assets and ended up getting hammered in both places. Sure, they hedged the basis risk out and may even have had the false sense of being in calm waters, except their basis risk model missed a major point: prepay model risk.

The prOAS Angle View of the Basis Risk

According to our view, a material portion of the basis widening and tightening can be attributed to the market's prepayment revisions and changes in the price of prepay model risk. These events are most likely the result of a sudden large (long unseen) change in interest rates, but may also be associated with other economic indicators, such as new home sales, economic trends and home prices. It is the relative position of the current MBS rate and the outstanding MBS stack that drives the risk.

If rates drop, the outstanding MBS volume will be at a premium, clearly triggering refinancing concerns. In such an event, the current-coupon MBS will widen in recognition of potentially more aggressive (and efficient) exercising of the prepay option. However, this widening of the MBS basis may affect other instruments completely differently. Premium MBS should widen to a much greater extent, IOs may be hammered whereas discount MBS won't be affected. This is what happened in 2003.

If rates rise quickly, the outstanding mortgage volume will find itself at discounts leading to concerns about the speed of housing turnover. Here, many may be surprised to learn that the current-coupon MBS will widen again - this time in recognition of potential losses caused by a slow turnover if the sellout continues. Of course, this widening will be leveraged in discount MBS and POs, but not in the premium MBS or IOs. In fact, IOs should tighten because they can be viewed as hedges against slowing turnover. In exhibit 1 we show a recent, but rather typical, prOAS calibration diagram with two major principal components of prepayment risk (red line, pink line).

Exhibit 1

As seen, the current-coupon MBS (around 6.5 as of lately) carries positive risk compensation for both refinancing and turnover risk factors. Should one of these risk factors go up or down, the current-coupon OAS ("MBS basis") will also fluctuate, causing non-trivial effects in the MBS markets described above.

If the cause of widening (tightening) is unrelated to the prepayment views, such as change of credit or liquidity, only then should the traditional view of the basis risk be used.

Hedging the Basis Risk

The anatomy of the MBS basis risk is multi-dimensional, and a good hedging strategy should include static or dynamic replication of the prepay model risk factors. Exhibit 2 taken from my 2006 AD&Co Conference presentation is an example of how to protect an IO.

Exhibit 2. Hedging an IO

Fannie Mae 350 IO strip is replicated synthetically by taking a long position in TBA5.0 and a long position in TBA6.0, i.e. effectively selling an IO. Both prepay model risk factors got hedged to near perfection. As for the remaining interest rate risk, it is to be hedged using usual rate derivatives. Assuming prOAS of 300 bps for the IO, the hedged position will generate $0.68 of annual access return per $3.36 of investment (i.e. about 20%).

How would this hedging strategy work in the 2003 events? Refinancing panic would reduce the value of the IO, but the price difference between TBA6.0 and TBA5.0 would shrink, too. Hence, losses of the hedged portfolio would be limited. Of course, this strategy remains exposed to some residual risk factors such as TBA-IO dislocation.

Conclusions

A large portion of the MBS basis is related to the prepayment expectation and the price of prepay model risk. MBS basis widening does not necessarily benefit IOs and MSR; the traditional hedging method - going long TBA - may lead to a disastrous result. Instead, a rigorous investor should realize that the MBS basis responds to prepayment views and considers hedging against prepay model risk factors, turnover and refinancing.