By the Numbers

Loan assumptions likely to remain sparse in GNPLs

Mary Beth Fisher, PhD | September 8, 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.

Agency single- and multifamily residential mortgages have always been assumable by new borrowers. Those qualified and willing to endure the application and vetting process can take over the existing loan. Because loan assumptions do not trigger prepayments, they can extend the maturity of the mortgage-backed securities and lower prepayment speeds. The extension risk is of particular concern for investors in low-coupon Ginnie Mae project loans, whose final maturity is typically 40 years. Luckily, assumptions of project loans have been relatively infrequent—and not entirely driven by interest rates—even in periods when assumption activity has increased. The benefit of assuming a very low coupon mortgage is tempered by other economic and tax factors. This will likely keep assumption activity low in the coming years, even as multifamily sales spring back to life.

Assumptions in Ginnie Mae project loans have declined to near zero

The Federal Housing Agency (FHA) has been insuring multifamily mortgages since the 1930s, but the Ginnie Mae project loan (GNPL) securitization program didn’t get going until decades later (Exhibit 1). Monthly issuance averaged $250 million in project loan pools through the 1990s, then began to ramp up in the early 2000s. Issuance peaked at over $5 billion in March of 2021, and has since retreated to about $1.5 billion per month through the first half of 2023. The increase in modified loan pool issuance (LM, green) is primarily a result of the interest rate reduction program, which began in 2013. This is a streamlined refinance option for GNPL borrowers.

Exhibit 1: Ginnie Mae project loan issuance

Note: The primary pool types are PN, which is a standard, non-level payment project loan; CL which is a construction loan; and LM which is a modified loan. PLs are a level payment loan, with the most recent PL pool issued in February 2008. Data is monthly, through 8/2023.
Source: Ginnie Mae, Intex, Santander US Capital Markets

Loan assumptions occur when there is a change in property ownership and a new borrower takes over a loan. This is called a Transfer of Physical Assets (TPA) and, for FHA-insured loans, requires approval by the US Department of Housing and Urban Development (HUD). Loan assumptions and prepayments both result in terminations of the original FHA insurance policy. Even though loan assumptions do not result in a prepayment, the new borrower must be approved for an insurance policy and the previous policy is extinguished when the loan is assumed.

Based on strictly financial considerations, loan assumptions, which result in the assignment of a loan to a new borrower, should be more attractive when the loan has a lower coupon than the current mortgage rate. When interest rates rise and loans are deep out-of-the-money to refinance, assignments should increase. Prepayments and loan modifications due to interest rate reductions should to the reverse: increase when interest rates fall and loans are in-the-money to refinance, and decline in periods of rising rates. This has historically roughly been the case (Exhibit 2).

Exhibit 2: Prepayments, modifications and assignments in GNPL

Note: A modification of a GNPL pool results in a prepayment. The modified loan is then reissued as an LM pool and can be put into a new deal. Data is monthly, through 12/2022.
Source: FHA, Ginnie Mae, Intex, Santander US Capital Markets

Since the late 1990s, the number of loan assignments has been relatively low compared to prepayments and modifications. GNPL assumptions peaked at 75 loans during one month in 1991 and have mostly declined ever since despite an enormous surge in issuance. For the past decade assumptions have been rare, averaging one or two per month while the program has grown to 44,000 loans outstanding. That’s perhaps not surprising given that long-term interest rates had been in a 40-year bull market until recently (Exhibit 3).

However, there are several periods where the number of loan assignments in FHA-insured multifamily mortgages increased, and were at or above those of prepayments:

• In the early 1970s, when mortgage interest rates rose rapidly
• For a few years in the 1989-1992 timeframe, and
• From 2000-2002, though assignments never exceeded prepayments.

These merit further examination, to determine if it’s likely that assignments could suddenly become more popular.

Loan assumptions eclipsed prepayments in the 1970s

A longer-term look at FHA-insured multifamily mortgage interest rates shows the 350 bp rise from 1970 to 1972 (Exhibit 3). Any loan issued at the average rate of 4.75% in the 1960s was deeply out-of-the-money to refinance for the next, well, five decades until 2010. Assignments did begin to rise and exceeded prepayments in 1974-1976, but prepayments didn’t vanish. Throughout the late 1970s through the early 1980s, assignments fell below prepayments despite mortgage rates continuing to rise.

Exhibit 3: FHA multifamily loan rates

Note: Data is the monthly average loan rate of FHA-insured mortgages that have been terminated. Data is through January 2020, since most pandemic-era loans are still outstanding.
Source: FHA, Santander US Capital Markets

Assigned and prepaid loans throughout the 1970s and until rates peaked in the mid-1980s tended to have deeply negative refinance incentives (Exhibit 4). Interestingly, through most of that period, the average refi incentive of the assigned loans was consistently less negative than that of the loans that were prepaying. This is relatively unusual, since after 1980 the refi incentives of both groups tend to be fairly consistent.

Exhibit 4: Refi incentives of assignments vs prepayments

Note: Refinance incentive is estimated based on average mortgage rate on the date of termination. Data is monthly, thru 12/2022.
Source: FHA, Santander US Capital Markets

The average age of assigned loans also tended to be much shorter than those of prepayments during the 1970s – early 1980s (Exhibit 5). Assignments were almost entirely for loans under 10 years compared to prepayments of loans well over 15 years. Again, although assignments continued to modestly favor shorter loans, the average age came closer together after the mid-1980s.

Exhibit 5: Average loan age of assignments vs prepayments

Note: Average loan age on the date of termination. Data is monthly, thru 12/2022.
Source: FHA, Santander US Capital Markets

There are several reasons for the difference in average loan age during the period:

• Prepayment penalties for Ginnie Mae project loans typically expire after 10 years. Assuming a loan does not require payment of penalties or reset them
• Tapping 15 to 20 years of equity versus 5 years is valuable enough to a borrower to sell or refinance the property despite much higher interest rates
• The tax consequences of assuming a loan, versus a purchase that requires entirely new financing, were possibly different during the time period.

The most recent rate-driven rise in assumptions in the late 80s/early 90s

The last time loan assumptions exceeded prepayments was in the 1988 to 1992 period. Interest rates were a bit volatile and refinance incentives, after briefly turning positive, dropped back into negative territory. Assumptions rose and exceeded the level of prepayments during the period, but the average age of both assigned and prepaid loans was about 12 years.

The non-interest rate driven wave of the early 2000s

There was no apparent rate-based incentive driving the rise in assumptions in 2000-2002, since both the loans that prepaid and those that were assumed had a positive refinance incentive, based on our estimates. The assumed loans had an average age of 12 to 15 years versus about 17 years for the prepays – both well outside the penalty period for GNPL.

It’s outside the scope of our analysis, but it’s possible that either changes in tax laws made the transfer competitive with a refinance or the assumption process itself was much less onerous at the time. Whatever the reason, something drove it and then the window closed. Post-2002 loan assumptions have dwindled down to one or two a month.

Exhibit 6: Project loan rates have risen sharply

Note: Average rates of loans issued during the month. Data through 8/2023.
Source: Ginnie Mae, Intex, Santander US Capital Markets

Assumptions never materialized in 2014-2016

If assumptions were still a viable option post-2002, particularly for low loan age loans similar to those that were assumed in the 1970s, a tick up should have materialized after the trough in rates in 2012-2013 (Exhibit 6). Prior to the pandemic, that period notched historically low rates for multifamily loans. The bounce higher in rates put 2012 and 2013 vintage loans anywhere from 150 to 250 bp out-of-the-money to refinance from 2014 through 2019, presumably making loan assumptions more attractive.

But the number of loan assumptions continued to hover near zero, while prepayments – not including those from loan modifications – remained relatively steady at 75 per month. Something structural appears to have shifted in the market that has either raised the hurdle or reduced the incentive to an assumption.

Hurdles to loan assumption have likely increased, while refis have become easier

Prepayments and assignments are both driven by sales of properties. Prepayments and loan modifications are also driven by refinance options, whose execution has become much easier and faster over the past few decades.

By contrast, loan assumptions have possibly become more difficult. Potential new owners must navigate the re-underwriting of both the insurance policy and re-examination of the loan, particularly if they intend to lever up. A second lien, allowed up to 80% LTV, must be approved by HUD. According to originators, a loan assumption can easily take six months or more and requires reams of documentation.

The upside to an assumption is that it doesn’t extend the term, change the rate or reset the prepayment penalty protections. Loan modifications that are interest rate reductions of GNPLs don’t extend the maturity, but penalties are due because the loan prepays, and the penalty protections restart. Lowering the interest rate typically improves the credit quality of the loan – for example, by increasing the debt service coverage ratio (DSCR) – which is why performing loans have little trouble taking advantage of the streamlined interest rate reduction program.

There is no free assumption

The economic benefit of assuming a far below market rate loan isn’t lost on the seller. They will likely expect a higher price for the property as compensation for transferring the low-cost financing, and the extra six months it will take to close the transaction as the potential buyer applies for an assignment. Putting in additional equity beyond the existing debt – either outright or through a subordinate loan – will also raise the overall cost of financing. Whether that raises it to the level of market mortgage rates will depend on the particular sale terms, but it’s an onerous process to go through unless the benefit is substantial.

Assumptions of Ginnie Mae project loans could tick up a bit over the next few years; but the prospect of hundreds of them materially extending maturities and keeping prepay speeds pinned near zero, even in low coupon 2021 and 2022 vintage securities like the FDIC portfolio, seems unlikely. The benefit of an assumption is relatively easy to monetize for the seller, and the hoop-jumping required of the buyer is formidable.