Valuation Commentary

2004 Valuation Round-up
by Alex Levin

Last December, the year-closing article was titled "The Lessons of an Eventful Year". This year did not warrant such an adjective - it was quiet and stable. However, stability has implication in financial markets - it reduces the value of risk and makes risky instruments (read: MBS) expensive. Hence, the MBS market did generally well, but left fewer investment options to new buyers and changed the flavor of the market.

The curve is flatter and the rates are less volatile

With the overall rates staying roughly where they were (Exhibit 1), the swap curve flattened considerably. The 10-2 swap spread started the year at 255 bps and ended it at 115 bps. This change alone has made mortgage market effectively shorter by 0.1-0.2 years and caused value impairment to IOs and MSRs. The absolute volatility index that we introduced in June of 2003 (http://www.ad-co.com/newsletter/Jun03/Value.htm) started the year at 120 bps and ended below 100 bps.

Fixed-rate MBS: freezing temperatures

The main trend in the TBA market could be characterized as "tight, tighter, and extremely tight" (Exhibit 1).

Exhibit 1

By the end of the year, LOAS for current-coupon agency MBS was down to zero, which is not a historical exception. In the absence of disturbing events, the MBS market often dives into zero or even sub-zero temperatures (see Levin [2001]). What was more striking is that high-premiums (7.5s and 8s) and low-discounts (4.5s) became traded virtually flat to swaps as well. In our view, this flat LOAS profile witnesses a complete ignorance of both refinancing risk and turnover risk. To illustrate this point, let us take a look at Exhibit 2 where we depict the history "risk-neutral" tunings of the AD&Co agency model (since March 19, 2004).

Exhibit 2

Those who read our articles and follow our research systematically got acquainted with this jargon that has become the main theme of the year (A. Levin [2003-04]). I realize that all others need introduction. To be very brief, risk-neutral tunings show how to alter our empirical prepayment model to eliminate an OAS difference between MBS and agency debentures. Risk-neutral tunings account for the model risk that gives rise to OAS. Pronounced over-1 refinancing tuning indicates considerable risk of understated refinancing priced in by the MBS market. Likewise, under-1 turnover tuning reflects a concern that the modeled turnover can be too fast. Close-to-1 tunings suggest that the market does not appreciate and reward bearing a risk. OAS computed to using a risk-neutral prepay model is called prepay-risk-and-option-adjusted spread (prOAS).

As seen from Exhibit 2, the year started with concerns on both ends and ended with neither. We agree with Gabaix at el [2004] and expressed similar views ourselves that the difference between the current coupon and the weighted average rate of outstanding volume influences the risk. Hence, every sharp change in rates immediately violates the balance and inflates the risk. To see that, please spot the changes occurred on May 7, July 30, or August 6, 2004. However, a slow market drift does not induce concerns. The uneventful fall of 2004 resulted in both tunings stabilized close to their risk-less level, 1.

Somewhat independent evidence of our theory can be found in the prepay derivatives market brokered by ICAP. According to ICAP, prepay derivatives have cheapened lately with 1-month implied volatility having fallen from 4 CPR to 3 CPR. Several consecutive auctions predicted the actual speed well. Although ICAP's derivatives are mostly for prepay surprises, they do carry long-term valuation implications. MBS prepayment predictability reduces the price of prepayment risk, which is tautology to MBS richness.

Strong evidence that the tight year-end OAS numbers are caused by curtailed risk aversion is found in the Trust IO pricing. Please read below.

Trust IOs: elusive and risky; caution and sophistication are required

Traditionally wide OAS on IOs reflect prepay risk leverage. On the other hand, one can gauge this risk using the TBA market as shown in Exhibit 2. What if we apply these risk-neutral tunings to value Trust IOs? The results can be counter-intuitive and not always inline with brokers' assessments and solicitations. To illustrate these findings, let us compare the markets of April 2 and December 17, 2004; the MBS rates were virtually identical for these two market days. The curve got flatter and less volatile, but even more importantly, the perception of prepay risk changed. It is the latter that most likely would elude investors.

Exhibit 3

As seen from the table above, OAS for the Trust IO market has tightened by 313 basis points, on average. At the same time, the risk-neutral tunings changed drastically. Thus, the refinancing tuning dropped from 1.511 (serious concern) to 1.068 (risk ignorance). As a result, the risk-adjusted tightening has been only 190 basis points. On the same basis, 8% IOs have even widened.

In general, IOs are rich today by historical standards, even on a prepay-risk-adjusted basis. However, they saw cheaper days as we just illustrated.

Other themes

Agency hybrids have been traded wider than fixed rates. A relatively new product, GNMA hybrids, are traded 1-2 price points cheaper than comparable conventional instruments, but have tighter margin and reset caps. According to lehmanlive.com, they are wider by a good 20 bps OAS. Valuation at 90% of the conventional hybrid model's speed confirms this cheapness.

Pay-up has been another popular theme. It is impossible to trace feature-specific pay-up dynamics (such as low-balance, high-FICO, NY concentration, etc.) without being on the market actively, but it is possible to develop tools that track the trends. Read the previous-month article and see examples. Ask us about case studies using our OAS system; it handily cracks massive broker inventories and sorts it by "practical arbitrage". In each such listing, we have seen pools beating the market pay-ups - even with just over a 6-month holding period. Given limited investment attractiveness of TBAs, using specified pools as a long position against TBAs could have been a good idea.

References

A. Levin, Mortgage Spread Dynamics, in F.Fabozzi (ed.) Professional Perspectives on Fixed Income Portfolio Management, vol. 2, FJF Associates, 2001.

X. Gabaix, A. Krisknamurthy, and O. Vigneron, Limit of Arbitrage: Theory and Evidence from Mortgage-Backed Securities Market (working paper), 2004.

A. Levin, Divide and Conquer: Exploring New OAS Horizons, in 3 parts, Quantitative Perspectives, AD&Co, 2003-2004.

---