The Big Idea
The new, new era in MBS
Steven Abrahams | January 28, 2023
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.
When the Fed started letting MBS roll off its balance sheet last year and banks, too, backed away, it tipped the market into a new era for MBS. For the first time since the start of the Global Financial Crisis, the market found itself without a big marginal buyer drawn to MBS for policy value rather than investment returns. The new, new era continues. And without a policy bid, MBS option-adjusted spreads seem headed back toward a level around 30 bp wider than where they stand today.
Market structure in the BGFC and AGFC
Before last year’s downturn in Fed and bank demand, the structure of the MBS market split neatly around the Global Financial Crisis, or GFC. Before the GFC, private capital almost completely set the marginal price of MBS. But after the GFC, public capital flooded the market and systemically tightened MBS spreads. The market in the next few years looks likely to give back some of that richness.
The split in market structure shows up nicely in the history of key investors’ share of outstanding agency obligations—MBS and debt (Exhibit 1). This share is readily available from the Fed’s quarterly Financial Accounts of the United States. Share of MBS alone is harder to come by. But the patterns in either series should be broadly similar.
Exhibit 1: Shifting shares of agency obligations mark different eras
From 1990 through 2008, banks and the Fannie Mae and Freddie Mac portfolios consistently held between 30% and 40% of agency obligations and invested for private returns. Some specifics:
- The Fannie Mae and Freddie Mac portfolios before the GFC invested for private shareholders, with common stock that traded on the NYSE. Each agency held 2.5 cents in capital for every $1 invested in MBS and funded at times as much as 20 bp below LIBOR across the curve. Except for a small liquidity portfolio, the agencies invested almost exclusively in MBS, and mainly in agency MBS. By the late 1990s, the agencies would commonly report pre-tax ROEs in the high teens or better based largely on portfolio returns. In 2003, Fannie Mae ran into some hedging problems that raised concern about systemic risk, and then both Fannie Mae and Freddie Mac ran into accounting problems that raised governance concerns. Aggressive portfolio growth stopped. The Fannie Mae and Freddie Mac share of agency obligations rose from nearly 0% in 1990 to nearly 20% in the early 2000s before dropping back toward 10% just before the GFC.
- Bank investment portfolios before the GFC invested largely for interest income and liquidity, with some portfolios that used the TBA dollar roll for funding and some that occasionally traded MBS for total return. Bank share of MBS tended to rise during and after recession—such as the 1991 and 2001 recessions—as loan demand cooled, deposit balances rose and banks needed to add an earning asset. Nevertheless, bank share of agency obligations fell almost in mirror image of the Fannie Mae and Freddie Mac rise, starting in 1990 at 33% and dropping to a low in the early 2000s near 17%, standing nearly 15% just before the GFC.
From 2008 through mid-2022, the Fed through Quantitative Easing and banks with regulatory needs for high quality liquid assets have dominated MBS, with their joint share rising start to finish from 15% to 57%. Some specifics:
- The Fed has used QE to add cash to the financial system after policy rates approached and stayed near zero. The Fed has allowed MBS to roll off its portfolio after QE1 and after QE3, but bank share of agency obligations continued to rise during both episodes. Fed share from start to finish has run from 0% to 25%.
- Post-GFC regulations have required banks, among other things, to add liquidity to the balance sheet. Banks subject to standard liquidity cover ratio or LCR tests lifted holdings of high quality liquid assets—largely US Treasury debt and agency MBS—from less than 5% of total assets before the GFC to nearly 20% before formal LCR implementation on January 1, 2016. Banks subject to modified LCR tests raised high quality liquid assets to a peak around 12% of total assets. Standard and modified LCR banks hold the vast majority of bank assets. Banks continued to use agency MBS for interest income, too, but use of the dollar roll and trading for total return fell, with substantial parts of bank MBS portfolios ending up passively in held-to-maturity portfolios. Bank share rose start to finish from 15% to 32%
It is especially noteworthy that the Fed and banks combined bought amounts of MBS equivalent to all of the net new MBS issued across Fannie Mae, Freddie Mac and Ginnie Mae and bought nearly $1 trillion in MBS beyond net new production. That additional $1 trillion came through MBS sales from nearly every other type of MBS portfolio.
Market spreads in the BGFC and AGFC
Spreads also split into BGFC and AGFC. The OAS on par 30-year conventional MBS in BGFC ranged around a median of 101 bp and in AGFC ranged around a median of 28 bp—a 73 bp difference. Analysis by the research staff at the Fed suggests that raising the Fed’s share of outstanding MBS from 0% to nearly 25% would tighten MBS spreads by 57 bp, an amount not inconsistent with the tightening in OAS before and after the GFC. Buying from banks to satisfy LCR requirements presumably would tighten OAS further.
Exhibit 2: Mortgage OAS dropped sharply after the GFC
Into the new new era
With the Fed on its way out of MBS and banks likely continuing to reduce MBS exposure this year and possibly beyond, OAS should move back towards levels before GFC. Note that nominal MBS spreads could still tighten as rate volatility likely comes down this year, but OAS should reflect shifting market structure and drift wider. How much wider depends on the pace at which the Fed and banks withdraw. The Fed’s work suggests that every 1% decline in Fed share of outstanding MBS should widen spreads by roughly 2 bp. Declining bank share should add to that, although by how much is an educated guess.
After QE3, Fed share bottomed out at 15%. With Fed share of MBS alone now just below 30%, a drop back to those earlier levels would suggest a widening in OAS of 30 bp. That would push OAS on par 30-year conventional MBS from the current 36 bp to 66 bp. Add a little bit for a declining bank share, and that roughly splits the difference between the Before GFC and After GFC median spreads. That seems like a reasonable expectation as MBS begins to flow back into hands managing not for policy value but for private investment return.
* * *
The view in rates
OIS forward rates continue price for peak fed funds approaching 5% in June with nearly 50 bp of cuts by the end of year. The cuts run contrary to warnings from Fed Chair Powell and other FOMC members that the Fed will hold rates high through next year. That sets up the market for a bit of volatility next year if Powell sticks to his guns. But the broad volatility trend still should be down.
Fed RRP balances closed Friday at $2.00 trillion, now down steadily since the start of the year. With the federal government at its debt ceiling and cutting back on net new issuance, rates on short Treasury bills have become much more competitive. Banks have also started competing a little more for cash by raising deposit rates. Competitive from T-bills and bank rates may be cutting into RRP.
Settings on 3-month LIBOR have closed Friday at 482 bp, roughly unchanged over the last three weeks. Setting on 3-month term SOFR closed Friday at 468 bp, up 3 bp over the last week. The spread between 3-month SOFR and LIBOR has renormalized after spiking last fall.
Further out the curve, the 2-year note closed Friday at 4.19%, rich given the risk of fed funds climbing to 5.0% or higher and remaining there into 2024. The 10-year note closed well above fundamental fair value at 3.50%.
The Treasury yield curve has finished its most recent session with 2s10s at -70 bp, unchanged over the last three weeks. The 5s30s finished the most recent session at 1 bp. The 2s10s curve looks likely to invert by around 100 bp shortly before Fed tightening comes to an end. That should come by mid-year.
Breakeven 10-year inflation finished the week at 232 bp, u 8 bp in the last week. The 10-year real rate finished the week at 120 bp, down 4 bp in the last week.
The view in spreads
Volatility has continued to fall through January and should continue with each reading on inflation and employment and each Fed meeting. The benign reading on core PCE in the last week pulled implied volatility down noticeably. Both nominal MBS and credit spreads should tighten as volatility drops. Nominal par 30-year MBS spreads to the blend of 5- and 10-year Treasury yields closed Friday at 129 bp, down 6 bp on the week. Par 30-year MBS OAS has closed Friday at 36 bp, down 5 bp on the week.
The view in credit
Most investment grade corporate and most consumer balance sheets look relatively well protected against the likely impact of Fed tightening. Fixed-rate funding largely blunts the impact of higher rates, and healthy stocks of cash and liquid assets allow these balance sheets to absorb a moderate squeeze on income. But in leveraged loans, a higher real cost of funds has already started to eat away at highly leveraged balance sheets with weak or volatile revenues. The 25% of leveraged loans rated ‘B-/B3’ are especially vulnerable to downgrade. The leveraged loan market is the bellwether to watch for broader corporate and consumer credit, and stress in leveraged loans looks likely to spill over into CLOs.