A sharp drop in the quality of TBA
admin | February 1, 2019
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
MBS investors know well that the note rate paid by a borrower is the most important driver of prepayments speeds. Typically there is a relatively stable spread between the weighted average coupon or gross WAC of a pool and the pool’s coupon, but lately that has changed. Fannie Mae and Freddie Mac give originators significant leeway to deliver loans into different coupons, allowing the spread to widen. And widen, it has. The rising gross WAC on new conventional pools has started to erode the quality of the TBA deliverable, and some of the market’s most widely used relative value models have yet to catch up.
The bid for excess servicing is high
Discussions with originators have indicated that the bid for excess servicing is extremely high. If servicing valuations exceed the amount received by selling into a higher coupon then the best execution is delivery into the lower coupon. Originators can deliver loans with a note rate up to 250 bp above the pool’s coupon. Exhibit 1 (below) shows the spread between the gross WAC of the loans and the net coupon of the pool for generic (non-specified) 30-year deliverable pools.
Exhibit 1: Gross/net coupon spreads have widened since April
This spread has widened significantly since April 2018. By July spreads had returned to levels (roughly 75 bp) seen in 2016, and held in that range through the end of October. Starting in November, however, gross/net coupon spreads widened further to roughly 85 bp on average. The most egregious example was deliveries into the 3.5% December Fannie Major pool, which has a spread of 110 bp. These pools will have worse convexity than a typical pool, raising the possibility that even a small amount of production could become TBA worst-to-deliver and lower TBA prices.
While the bid for excess servicing has been very strong, coupon swaps have sold off since the start of November. This is to be expected since interest rates moved significantly lower, but they are priced well below the last time rates were at these levels. For example, the mortgage current coupon averaged 3.52% in January 2019, and the 4.0%/3.5% swap was 64/32s and the 4.5%/4.0% swap was 52/32s. Rates were very similar in February through April 2018, yet the coupon swaps averaged 82/32s and 72/32s, respectively. This has likely contributed to lower coupons becoming best execution for loans.
The GSEs are participating, too
Exhibit 1 (above) shows the gross/net coupon spread for three different pool types:
- Single issuer pools
- Multi-issuer “major” pools (borrowing Fannie Mae’s name)
- Multi-issuer cash window pools
The cash window pools are especially interesting, since Fannie Mae and Freddie Mac make the pooling decisions. In a cash window sale the originator can retain at most 50 bp of servicing, so the fact that spreads on cash window pools have tracked wider alongside the major and single issuer pools—in these cases, the originator picks the coupon and retains the excess—shows that the GSEs have chosen to profit from the strong bid for excess servicing, possibly to the detriment of the overall MBS market. Although it is worth mentioning that neither GSE sold a 3.5% cash window pool in December.
The problem is this hurts the TBA deliverable
The TBA price has to reflect the valuation of the worst-to-deliver pool, so the creation of these high gross WAC pools should push down TBA prices. YieldBook has not yet updated its TBA assumptions, and its gross/net coupon spreads are roughly 60 bp to 65 bp. That is roughly 20 bp lower than current levels. Exhibit 2 (below) shows that adjusting gross WACs 20 bp higher should lower the TBA price by 5.6/32s for 3.5%s, 7.9/32s for 4.0%s, and 8.6/32s for the 4.5%s.
Exhibit 2: Higher gross WAC should hurt TBA prices (as of 1/30/2019)
Some of this should manifest itself in higher pay-ups for a variety of more seasoned pools that have lower gross/net coupon spreads. However, many investors look at pay-ups as a percent of theoretical value, so the actual pay-up increase will be less than the theoretical value. For example, take a pool with a 20/32 theoretical pay-up and 8/32 actual pay-up. If the TBA drops by 6/32s then the new theoretical pay-up would be 26/32s, but the new actual pay-up would be , a 2.4/32 increase. Therefore the entire market suffers if the TBA deliverable worsens.
Ginnie Mae MBS do not suffer this problem
Ginnie Mae does not offer the same leeway to servicers that the GSEs provide. Within the Ginnie Mae II program the gross/net coupon spread must be between 25 bp and 75 bp (the older Ginnie Mae I program requires the spread to be 50 bp exactly). This means that originators have no option to deliver a loan into a different coupon. Therefore, any spread variation is small from month-to-month and depends only on the mix of loans being originated in a given month.