|
The yield on the 10-year pushed higher over the course of December, and is currently at its highs of early August in the low 3.80%s. The move pushed the yield curve (measured as the 2-10 year spread) to its all-time recorded steepness of +287 basis points, while the 5-year butterfly (2-5-10s) also moved out, although it remains well inside its highest level seen last spring. (See Chart 1 below.)
The sharpness of the move in the long end pushed the 10-year yield well outside of its upper 2 standard deviation bands, at least temporarily. Chart 2 shows the daily close of the 10-year versus the bands, using 40- and 60-day lookback periods. What’s also interesting is that the bands themselves have widened out. As illustrated in chart 3, the width of the bands (i.e., the difference between the upper and lower bands, in basis points) increased sharply last month, from their tightest levels since the spring of 2007. If one thinks of these bands as defining a trading range, it suggests that the 10-year yield has a lot more room to adjust before triggering the types of trades done in a breakout (i.e., duration adjustments).
While the focus has been in the long end of the curve, the short end has had an interesting move. As Chart 4 indicates, the front end of the Eurodollar curve (i.e., 3-month forward LIBOR) is lower through September of 2010 than it was in early December, before the Treasury market sold off. The curves only cross over after March 2011. One interpretation of this is that the selloff in the long end was more driven by supply and/or inflation fears than by expectations of tightening by the Fed. 
Mortgages have had a mixed run during the selloff. Chart 5 below shows daily duration-neutral performance (versus the 10-year Treasury, using YieldBook effective durations) since the beginning of last December, adjusted for dollar rolls. The chart has two notable aspects: MBS performance was spotty during the selloff, and the daily relative performance was fairly volatile in both directions. Put differently, MBS did not track Treasuries particularly well during the move. It’s also interesting that the cumulative performance of these coupons was either neutral (for Fannie 4.5s) or negative. (Over the five-week period, Fannie 4.0s lagged the 10-year by roughly a quarter-point, while 5.5s lagged by 3/8 of a point.) 
MBS’ surprisingly poor performance was also highlighted by their empirical durations. During the selloff, the empirical duration of MBS have extended substantially; as shown in Chart 6, the 20-day empirical durations (versus the 5-year) of the entire Fannie coupon stack have extended, with Fannie 4.0s extending by almost 2/3rds. 
Finally, the Fed’s purchases have begun to trail off, with $12 billion in net purchases for the week ending January 6th. By my count, the Fed has purchased a net total of $1.127 trillion in agency MBS since the beginning of the program, leaving a little over $123 billion left to be purchased (assuming that the program is not extended). With 11 weeks left in the quarter, this suggests average net weekly purchases of $11.2 billion, down from the 15.8 billion/week taken down from mid-November until Christmas. While it’s a very limited example, it may be that the reduced purchases from the Fed may have had a roll in the languid performance of MBS over the past few weeks.
 |
Subscribe to this comment's feed
