We analyse TBA Deliverables and their Effect on Dollar Rolls for August Settlement.
- Three-month winning streak ends.
- Rising rates crimp all things housing.
- FDIC successfully completes MBS agency pass through sales.
- Anatomy of a dollar roll squeeze.
For more data-driven insights in your Inbox, subscribe to the Refinitiv Perspectives weekly newsletter.
A streak is a streak, until it isn’t.
Supply remained steady while demand did not as U.S. agency mortgage-backed securities (MBS) closed out the summer season on a down note. Summer bummer.
Sellers took turns at every back up on rates, more frequent and sustained, while buyers balked at higher/wider entry levels as only infrequent FTQ rates rallies inspired support. Speaking of support, the FDIC liquidation lists are 85% completed, with Agency Passthroughs 100% done. Of the original USD114 billion total, just USD23 billion is left, of which USD11 billion in Agency CMBS and USD6 billion Agency CMOs.
MBS index for August lagged -50 basis points in excess returns to the duration neutral Treasury index, while posting an even more negative Total Return of -95 bps as prices tanked into summers end. Volatility measures eased back to +111 bps on the benchmark 3m10y (normalised) tenor. 10yr yields were net 8 basis points higher to 4.09%s, as the 2s10s curve “steepened” back to -75bps.
Source: Yield book
Originator supply held steady at USD2.428 billion with pipelines flushing as soon as they filled. Current production coupons shifted a bit with rates, 42% in 6%s and 33.9% on 6.5%s with 5.5%s falling off against sell offs. The 30yr FHLMC primary rate has risen 28 basis points on the month to 7.18% and Bankrate’s daily quote poked the mid 7.50% area.
Dollar rolls told an interesting story this past month as Class A settlement (30yr FNMAs and FHLMC), featured the plight of shrinking supply for TBA. Spec pools carved out by hedge funds, left little or no “cheapest to deliver” bonds for TBA settlement and caused the 6.5% roll (USD102 handle) to spike 1/4 point above its value (“cost of carry”) and 200 basis points “special” (below funding).
When the bulk of upper coupon supply is already carved out to specified pool buyers, that usually spells trouble for delivery and has too many buyers chasing too few bonds to avoid “fail to deliver” (penalised by coupon interest daily at zero funding). This is what can drive the price of the roll, the “drop”, above break-even levels (or “carry”). Our analysis to follow will further break down the particulars.
Market Perspective – Deliverable float carve out
Each month MBS settlements are done in bulk for the four (4) major classes of agency product; Class A (30yr FNMA FHLMC), Class B (All 15yrs), Class C (GNMAs), and Class D (Arms and Ballons). This is in addition to specific and non-regular settling pool trades outside of the dominant TBA (to-be-announced) contracts which comprise 90% of the volume. With USD250-USD300 billion traded daily, one can imagine the vastness of settlements and how an organised and regularly scheduled clearing and netting procedure is paramount to market sanity.
The process of regular and scheduled settlements was set up decades ago to facilitate order and ease market access and usage. Even during turbulent times over that timespan (financial crisis, wars, housing crisis), TBA trading, and settlement, has largely proceeded without a hitch.
That basically leaves supply and demand to chart its predestined course, with bids and offers subject to both shortages and surpluses as settlement nears. This past month, after long languid periods, dollar rolls finally showed a little stress.
Case in point, the Class A 6.5% roll with August settlement representing the front month on that roll, spiking 5/32nds in a matter of days and well above its “carrying” value.
The cause-too few bonds available to satisfy sellers delivery needs to the buyer. Hence, sellers of the front month had to chase down bonds, via dollar roll markets during settlement, to avoid failing to deliver on a premium coupon.
Source: LSEG Advanced Mortgage Analytics
From the table, total issuance for the past two months in 30yr UMBS (FNMA/FHLMC) 6.5%s was USD7.383 billion in 526 pools. This includes specified pools as well as generic; “cheapest to deliver”, or CTD, most likely to find its way into TBA settlement.
From that list, filtering out more valuable specified pools, the list gets pared down to USD2.217 billion in August and July production. CTD tends to omit older paper, which contains underlying value in its seasoning alone. The net yield amounts to just USD190 million in TBA-likely August production generic pools (loan age 0.4years).
There is the essence of the Class A 6.5%s shortage for delivery in August-there weren’t enough CTD bond readily available, and sellers balked at delivering more valuable pools and instead scoured the open market to acquire TBA likely paper on settlement day thereby driving up the price of the dollar roll over and above cost (of carry)
Month to date
Spreads were softer as summertime conviction was lacking after Fed events and moves faded following the beach crowd out to the beckoning fall season. After such a busy and eventful year that’s understandable, as is the free-market backstop bid (as opposed to the Fed’s mandated bid), where buyers can sit idle and wait for “their” ideal entry point, not the markets.
The 30yr current coupon (30yr CC) rose a net 16.5 basis point, in synch to higher overall interest rate moves. OAS gapped net 5.4 bps (using the Tsy curve), ZV (Zero Volatility) net 3.5 bps wider, and benched to the 5&10yr Treasury net 6.7 bps softer-consistent across all tenors and yet back from mid-month wides.
Source: Yield book
Year to date 2023
Even though spreads have softened over the past month, the recovery from the bank failure induced wides of May 26, 2023 have been impressive. Much like the liquidation of agency MBS alluded to above, the widening was more of a knee jerk reaction and nothing systematic to the underlying collateral itself. Fixed income securities do have the danger of falling behind when rates rise, and such was the case with the lower couponed and longer duration that SVB/Signature banks held, and held, and held until it was too late.
From the chart below, we have highlighted the peak of wide spreads on the aforementioned date, then the solid tightening recovery of 30 basis points into mid-June, and a gradual tightening until the late summer swoon we just experienced.
The Current Coupon itself has put on 51 basis points to 5.851%, OAS spreads wider 27bps (43.89), ZVs +22.7 (155.8bps), and the much watched 5&10yr Treasury comparison +25bps to +168 context.
On the plus side, since the bank failure marker; 30yr CC +7bps, but OAS, ZV and 5&10yr spreads have actually firmed -15 to -21 bps. Not too shabby.
Source: Yield book