The Treasury market has sold off over the last few sessions with a
violence that brought the debacle of 1994 to mind for industry veterans. The yield on the 10-year has backed up by about
24 basis points since Monday’s close.
What makes it feel even more wrenching was the fact that the market had
been trading in an increasingly tight range since October. The chart below shows the daily close of the
10-year (the black boxed line) versus bands that show 2 standard deviations of
price changes. Unlike last October’s
selloff (which pushed the 10-year to 2.40%), this move pushed the 10-year yield
decisively beyond the upper 2SD band; this is a good graphic representation of
why this selloff feels so
cataclysmic.
(Here are some additional charts of yesterday's pain from MND)
Mortgages had a tough
time toward the end of trading on Wednesday.
As rates rose and prices dropped over the last few days, the
option-adjusted durations (OADs) for 30- and 15-year passthroughs extended
dramatically. Using YieldBook’s
calculations, the OADs for FN 3.5s and 4s both extended by 0.6-0.7; the OADs
for Ginnies extended slightly less. To
put this in context, the price of Fannie 3.5s would now be expected to change around
3/32s more for a 10-basis point change in yields than on Tuesday morning (holding
everything else constant, of course).
It’s useful to consider some of the technical factors
impacting both the Treasury and MBS markets in this type of environment. The MBS duration extension discussed above triggers
the type of convexity selling I noted a few weeks ago, in which investors
lighten their positions in order to shed duration. Unfortunately, this type of selling tends to
feed on itself; investors sell duration as their portfolios extend, which
pushes prices lower, which further extends durations, etc…In this way, the MBS
market’s negative convexity impacts the overall market for fixed income by
exaggerating both selloffs and rallies.
Mortgage bankers also need to be aware of the impact of
the selloff on execution. A rule of
thumb is that production will tend to move to a higher coupon (i.e.,
originators will “pool up”) when the price of the current coupon moves below
102-16 for two months forward settlement.
(Right now, that would be FN 3.5s for May settlement.) We saw a classic illustration of this on
Wednesday. Early in the day, almost all originator
selling in Class A was seen in 3.5 coupons; and although 3.5’s vastly
outnumbered 4.0’s in terms of the daily total, by the end of trading, 4.0’s were a noticeable
contributor.
This will impact the market in a variety of ways. The price performance of TBA 4s (both
conventionals and Ginnies) will probably underperform their hedge ratios, as
selling picks up and weighs on prices. Dollar
rolls are also likely to richen, since selling will tend to be in the back
months. Among other things, wider rolls
make it less attractive for lenders to hold their inventory of loans and
deliver them into a later-month TBA trade.
For secondary market managers, the key questions in this
type of market are related to hedging decisions. Assuming rates settle into the current area
(around 2.25% on the 10-year Treasury), I’d be inclined to hedge my loans that
will be allocated into 4s using relatively long hedge ratios, given the
likelihood that originator selling pressures push their prices lower. Hedging loans that are pooled into 3.5s will
require a different strategy. I don’t
recommend trying to buy convexity in the options markets, unless 1) you are an
experienced options trader, 2) you have extremely good analytical software, and
3) you’re related to your trade counterparties.
A better solution is to very actively trade the hedge positions based on
market moves. While you don’t want to
be hyperactively buying and selling, you can let your position get long as the
market trades higher (by not immediately hedging all new locks); in selloffs, however,
immediately hedge new locks using progressively longer hedge ratios.