The Big Idea

Lessons learned from agency MBS 2023

Brian Landy, CFA | December 15, 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. This material does not constitute research.

The agency MBS market started the year priced near its lowest level ever and weakened throughout much of the year. I expected relatively heavy supply and light demand to keep spreads wide. Supply didn’t reach my expectations, though it still topped pre-pandemic norms, and spreads generally remained wide throughout the year. Without a stabilizing buyer like the Fed and banks, spreads showed more sensitivity to market events—widening at the onset the regional bank crisis in March and in response to REIT selling in September and October; tightening as the banking crisis was resolved and as the Fed raised expectations for future rate cuts in December. Like supply, housing turnover didn’t meet my expectations but also avoided extreme low scenarios many market participants feared. Home prices rebounded after a short period of falling prices, illustrating the housing market is resilient in the face of higher mortgage rates.

Spreads stayed wide and volatile.

Mortgage spreads remained wide throughout the year, due to a combination of modestly heavier net supply from new production and effective supply as the Federal Reserve chose not to reinvest MBS paydowns and banks generally allowed their MBS portfolios to shrink. Although net supply plummeted compared to the heavy levels of 2020 and 2021, during those times the banks and Fed bought all the net supply and roughly an extra $1 trillion from other market participants.

Since 2022, mortgage demand has been led by money managers and spreads have approached the widest levels since the 2008 financial crisis. The varying value of mortgages relative to other asset classes as well as inflows and outflows from funds dictates mortgage spreads, increasing volatility. For example, from 2012 through 2019 the standard deviation of daily changes of the Bloomberg MBS index OAS was 1.7 bp. That volatility jumped to 3.3 bp in 2022 and 2023. Volatility was typically only higher during times of great stress—the financial crisis in 2008 and the few years following, and during the onset of the pandemic.  This should continue throughout 2024 and many investors may prefer MBS with lower spread duration to minimize the exposure to spread moves.

Exhibit 1. Volatility of the Bloomberg MBS Index OAS.

Option-adjusted spread vs UST. Volatility is calculated over a 6-month rolling window, annualized.
Source: Bloomberg, Santander US Capital Markets

But sometimes supply brings demand

Although typically increasing supply leads to asset underperformance, occasionally supply can elicit demand. This happened when the FDIC started to sell MBS pools acquired after the failures of Silicon Valley Bank and Signature Bank. Most of the pools were concentrated in lower coupons, and many people expected lower coupons to underperform higher coupons in order to find buyers. However, the lists traded very well. For example, the FNCL 2.0% excess returns in the Bloomberg MBS index were 0.73 in May compared to 0.16% for FNCL 5.5%s. Lower coupons ultimately underperformed later in the year as the selling came to an end. Much of the collateral was in specified pools, not generic pools, and that probably helped pique interest. Selling generic pools, like in the improbable event the Fed were to sell MBS pools, might not have found the same interest from the market.

Supply and turnover come in below expectations but avoid a total collapse

It is likely that the mortgage market has not been so far out-of-the-money in at least the last 30 years, and there is little-to-no data on prepayment speeds prior to then. The MBS market has faced refinance wave after refinance wave, separated by short periods with the market trading at a modest discount to par. Speeds for 30-year MBS averaged 5.4 CPR over the last 12 months—about 5.0 CPR in conventional and 6.3 CPR in Ginnie Mae. That is certainly slow, and below typical estimates that baseline discount speeds should be around 6 CPR to 7 CPR. On the other hand, it avoided some of the worst-case possibilities investors feared.

A couple factors helped to prevent speeds from crashing to zero despite historic levels of unaffordability. The first is home price appreciation. This is a double-edged sword—it makes homes less affordable, but also bolsters the buying power of existing homeowners. A homeowner with a mortgage earns a leveraged return, which helps when trading into a larger home. Some moves cannot be avoided despite high mortgage rates. Furthermore, some borrowers are still doing cash-out refinances. Roughly 8% of conventional loans and 15% of Ginnie Mae loans originated in the last year were cash-out refinances, which helped support speeds. The cash-out refinance might make sense for borrowers that need to consolidate debt but cannot qualify for a second lien.

High levels of home equity and a strong economy led this space to expect speeds would settle around historical levels, not below. More weight should have been given to the increase in rates, reduction in affordability, and the pace of turnover during the pandemic that pulled some moves forward. Slower than expected speeds also pulled net supply lower than expected. But turnover did not crash to zero and speeds for most of the year speeds moved in conjunction with seasonal factors and not mortgage rates. So, it is plausible that speeds will edge higher this year and return to historical norms in a couple of years, even if rates were to stay high. Hope springs eternal!

Home prices keep growing.

Home prices surprised many in the market by increasing for much of the year. Prices fell for a short time before rebounding and reached new highs this year. This team was an outlier in expecting prices to, at a minimum, not fall—the 30-year fixed rate mortgage protects existing borrowers from rate increases and there didn’t appear to be another catalyst, like a recession and growing unemployment, on the horizon to force sales. Homeowners can choose not to sell a home, which tends to prevent price declines. Investors should expect prices to keep trending higher unless the economy takes a severe turn for the worse.

Persistently fast servicers

Servicers can wield a lot of influence over the chances a borrower will refinance. Many people don’t watch mortgage rates too closely and may not realize when their loan is in-the-money to refinance. A servicer that is proactive about alerting borrowers to refinance opportunities can create prepayments and lift speeds. But it was unclear whether a servicer could influence prepayment speeds in a turnover market—how could a servicer encourage borrowers to move?  But there have been some servicers whose loans have consistently prepaid faster or slower over the last year. Quicken’s conventional loans, for example, prepaid 25% faster than its peers over the last 12 months. That’s a significant difference for pools trading below par. Small servicers, on the other hand, tended to prepay a little slower. One explanation is that servicers with larger cashout refinance businesses were able to maintain faster than average speeds.

Policymakers adjust loan pricing

Conventional and FHA loans both faced price adjustments this year. At the end of 2022 it seemed apparent that the FHA was preparing to lower annual insurance premiums, and that came to pass in February. However, the depth of the cut was surprising, taking premiums back to the levels before the 2008 financial crisis. The FHA ultimately needed a bailout from the Treasury, so a more modest cut had seemed prudent. The GSEs also seemed likely to tweak pricing to favor borrowers with lower credit scores and lower down payments, and this also happened. The timing was opportune for investors since most pools were trading at deep discounts to par and there was no reason to fear a pickup in refinancing. In fact, anything that might reduce lock-in and lift speeds would increase the value of MBS. Looking ahead it seems like there is less opportunity to make broad changes to loan pricing and the focus will center on targeted lending programs, so maybe there is less to worry about for investors. But it is never easy to predict policy changes, and the elections in November could set the stage for big changes in 2025.