MBS performed well last week, with the Fannie 30-year current coupon rate tightening by four basis points on the week to interpolated Treasuries and 2 basis points to swaps.  Duration-neutral price performance over the period was also fairly good; most liquid Fannie coupons (up to 6s) outperformed Treasuries by 8-12 ticks on the week.

There are a few interesting things to note with respect to the MBS market, however.  The chart below indicates that the 30-year current coupon is approaching its tightest levels over the last six months or so; however, spreads have not pushed through the tights first seen in mid-November, and again seen in early January.  Moreover, spreads versus swaps remain at the midpoint of a range first established last September.  Depending on whether you think that swaps remain a relevant benchmark, MBS may not be as richly valued as it appears at first blush.  Moreover, if the sector is rich, I’d argue that it has a great deal to do with the effect of the Fed’s purchase program on volatility.

A brief look at the recent path of the 10-year Treasury yield indicates that it is in the middle of a steadily contracting range.  Chart 2 below shows the daily close of the 10-year yield along with 2 standard deviation upper and lower bands.  This gives a good indication of how 10-year yield volatility has shrunk since the beginning of the year. 

This point is also made by Charts 3 and 4 below.  Chart 3 shows the bands (using two different lookback periods) have contracted over time, and are currently approaching lows last seen in the spring of 2007.  A similar picture is shown in Chart 4, which shows the MOVE index, (“a yield-curve weighted index of normalized implied volatility on 1-month Treasury options,” according to Bloomberg) along with the 30-day realized vol of the 10-year Treasury.  (Note that while the 10-year realized vol tracks the index well, realized vol for the 2-year has a much lower correlation.)

These charts indicate that intermediate and long-maturity Treasury volatility continues to trend lower.  Moreover, I’d argue that the Fed purchases have had a significant impact on realized volatility of Treasuries.  The Fed has pulled the equivalent of about $650 billion of 10-year Treasuries out of the market.  By pulling this massive amount of mortgage duration (with its associated negative convexity) out of the market, the investment community collectively needs to buy and sell far less duration to stay flat when the market moves; in turn, this reduces the intensity of changes in Treasury yields.

Therefore, the Fed’s purchases have put downward pressure on mortgage spreads for two reasons—the obvious supply considerations, and the reduced level of volatility in the bond market.  This also suggests that mortgage spread may not widen dramatically once the Fed stops buying MBS later this month, as the reduced volatility should stay with the market for a while--at current production levels, it would take about two years to add an equivalent amount of mortgage duration to the market.  (I made a similar argument in my February column in Asset Securitization Report, available here.)

The February prepayment reports showed that Freddie Mac speeds reflected the delinquent loan buyouts, with some premium cohorts virtually paying off.  (2007 Gold 7s, for example, prepaid at 98% CPR.)  As noted in some articles (see here), Fannie’s buyouts have been much less organized and efficient, with a series of press releases outlining their procedures.  The result has been significant disarray in the market for premium Fannie dollar rolls.

It is interesting to use values in the roll markets attempt to deduce short-term prepayment speeds are being implied.  Exhibits 5 through 7 below show the implied one-month CPRs for the premium Fannie rolls for March/April and April/May settlement, along with break-even speeds for changes in the levels of the rolls.  (All three charts assume a 0% Cost of Funds).   Exhibit 5 shows that Fannie 6.5s are rolling at negative drops for both months, which translates into speeds in the area of 75-80 CPR.  (The Fannie 6.5 roll was trading early Monday at -10+ and -8 ticks for the Mar/Apr and Apr/May rolls, respectively.)  This reflects the likelihood that Fannie will buy loans out higher coupon pools first, as described in their release last week available here.

Dollar rolls levels for lower coupons are reflecting the likelihood that prepayment speeds in March for TBAs will not peak until April.  Exhibit 6 indicates that the current level of the Fannie 6 roll for March/April of 3 3/8s ticks, which equates to a March speed in the area of 57 CPR.  For the April/May roll, however, the roll is currently trading to a drop of -7+ ticks, which represents a speed in April of around 78 CPR.

In addition to uncertainty over the timing of buyouts, values for the Fannie 5.5 roll are apparently being skewed by tight technical condition for the coupon.  As shown in Exhibit 7, the break-even speeds for Fannie 5.5s is around 30 CPR for the March/April roll and 48 CPR for April/May, using current drops of 10 3/8 and 5 ¾ , respectively.   Both breakeven prepayment speeds seem slow in comparison to the higher coupon.

Comments (0)

Subscribe to this comment's feed