In the first year of the COVID-19 pandemic, I headed the Federal Housing Finance Agency (FHFA), the federal agency that monitors the giant housing enterprises like Fannie Mae and Freddie Mac that coordinate most private financing of mortgages. As such, I was responsible for designing the emergency housing payment relief program that many mortgage holders utilized in the pandemic’s early months. I was determined that this relief would avoid the serious problems that had plagued housing finance assistance programs in the previous national economic emergency, the Great Recession.

COVID not only posed challenges to U.S. public health policy, but also to economic and broader domestic policy. Lockdowns, fear of exposure, and uncertainty surrounding both the economy and the government’s evolving response resulted in the sharpest labor market decline in American history. Total nonfarm employment dropped by 22 million jobs from February to April 2020. In comparison, just under 9 million jobs were lost in the Great Recession, and that occurred over two years, from 2008 to 2010.

With job loss often comes housing distress. The usual mechanism for helping workers weather a period of joblessness is the unemployment insurance (UI) system, a partnership between states and the federal government. But UI requires ongoing demonstration that applicants qualify for benefits, and that takes time to compile and process before benefits are received. Whether the job market would recover in three months or three years, policymakers during COVID needed to create a financial bridge to stabilize families’ financial health, at least until the unemployment benefits kicked in.

The UI system typically covers about half of lost earnings and about half of all workers. There are a variety of reasons why some workers are not covered, but the most common is that they are exempt because they are self-employed or part-time workers. The duration of coverage is within the discretion of each state government but is generally less than six months. During severe economic downturns, however, Congress will typically increase both the percentage of lost earnings that are covered and the duration of coverage. Congress followed this pattern during COVID and extended coverage to some workers normally outside the UI system.

For typical renters and homeowners, UI is generally sufficient to cover monthly housing expenses. But because UI benefits are capped, recipients with particularly high housing costs may have trouble fully covering those costs. Moreover, there is no requirement that those receiving UI spend it on housing, although most appear to do so. Still, without UI, it is likely that foreclosures and eviction rates would be two to three times higher during an economic downturn.

Because UI benefits are often slow to reach the unemployed, that can leave those with little savings struggling to cover their housing costs, at least temporarily. For policymakers, the solution to this problem is to provide borrowers with loan forbearance, not forgiveness. Forbearance gives borrowers a “time-out” on their monthly payment, allowing them to add missed payments back into their loan balance so the money will eventually be repaid. The hope is that the borrowers will be back on their feet within a few months or else they will have started to receive UI benefits, so they can resume their mortgage payments. That is what we aspired to do during the COVID emergency.

Learning from the Mortgage Crisis

During the 2008 financial crisis, I was a senior adviser to the U.S. Senate Committee on Banking, Housing, and Urban Affairs. From that perch, I observed up-close federal policymakers’ efforts to provide relief to mortgage payers. I became convinced that much of the federal response was poorly structured, especially the mortgage assistance programs, the Home Affordable Modification Program (HAMP) and the Home Affordable Refinance Program (HARP). When we began work on COVID mortgage relief, I was determined to not repeat the design flaws of HAMP and HARP. We would do it right, or at least better, this time.

I also wished to avoid structuring assistance in a way that would discourage work. The Great Recession witnessed the weakest job growth of any post–World War II recession. This was puzzling because of the seeming disconnect between overall consumer spending and the job market. The economy was weak in the immediate aftermath of the 2008 financial crisis, but consumer spending recovered relatively quickly. Normally, one would expect such an increase in spending to translate into a similar-size recovery in employment. But that didn’t happen.

Another puzzle was the breakdown in the usual relationship between unemployment and job vacancies. Job postings steadily increased once the economy hit bottom in the summer of 2009, but with little effect on the unemployment rate. Many in the economics profession, at least the professional forecasters, saw the weak job market as a function of weak demand in the overall economy. I was among the minority of economists around 2009 and 2010 who believed that we were instead facing structural changes in the labor market. During a Senate hearing at the time, I suggested that mortgage assistance programs were locking workers in place, perhaps discouraging moves from weak job markets to stronger job markets. The Great Recession, for instance, was one of the few recessions in which mobility decreased. Recessions rarely affect all parts of the country equally. In 2010, for instance, Nevada had an unemployment rate of almost 15 percent, but North Dakota’s rate was just 3.9 percent. There were likely unemployed construction workers in Nevada who stayed there because of the structure of mortgage assistance programs, rather than taking work in the energy boom occurring in North Dakota.

It was not until I had started reading the work of University of Chicago economist Casey Mulligan on the Great Recession that I fully appreciated what was happening. Mulligan’s thesis is a simple one, even if he uses a lot of complex math to demonstrate it. The idea is that the massive expansion of means-tested assistance programs, starting in 2008 but ramping up in 2009 and 2010, resulted in very large increases in work disincentives in terms of government benefits lost if the recipients experienced income increases. At the center of this expansion were the mortgage assistance programs established in the Great Recession.

The admirable objective of these assistance programs was to keep families in their homes by making their mortgage payments affordable. Policymakers decided to cap program participants’ monthly mortgage payments at 31 percent of income, a figure that was a somewhat arbitrary choice. That created a nasty disincentive: It added a tax of 31 percent on each additional dollar one earned, because one’s monthly mortgage payment would increase by 31 cents for every new dollar. Worse, above a certain income threshold, borrowers lost program eligibility, so in some circumstances people could see an implied tax rate of 100 percent or more. Mulligan estimates that in some unusual instances, borrowers could face an implied tax rate of almost 400 percent.

Combined with other expanded means-tested assistance programs, such as unemployment assistance, the total loss of government benefits for each additional dollar earned could be staggering. Mulligan’s calculations indicated that these program expansions and their very strong work disincentives were the primary reasons for the weak job recovery of the Great Recession. With that research in mind, I was determined that COVID housing finance relief would assist families but not discourage work.

The FHFA Plan

Accepting that any program is going to have costs and benefits, the immediate objective was to buy time for mortgage holders to get back on track, or in the parlance of the early months of the pandemic, to “flatten the curve” of mortgage defaults. Because of that objective, we decided to base mortgage forbearance not on current income but on time. While a loss of income was the immediate qualification, beneficiaries would not lose forbearance if they went back to work or earned more income.

The length of assistance would be limited to our best guess as to how long the pandemic would adversely affect the job market. In March 2020, our best estimate was a three- to six-month downturn in the job market. Accordingly, we initially offered forbearance with the requirement that every three months the lender would check in on the financial status of the borrower. When Congress later codified our plan into statute as part of the Coronavirus Aid, Relief, and Economic Security (CARES) Act, the maximum assistance period was fixed at 12 months.

Another problem of the 2008 mortgage crisis and subsequent Great Recession was the paper chase associated with the various mortgage assistance programs. Generally, for borrowers to be eligible for forbearance or other mortgage assistance, a substantial amount of paperwork was required. There are countless stories of those materials being lost or wrongly rejected. In some instances, forms were submitted with false information by either the lender or the borrower. As law professors Kathleen Engel and Patricia McCoy described in their 2016 book The Subprime Virus: Reckless Credit, Regulatory Failure, and Next Steps: “Servicers were inundated with paperwork. Loan files got lost or were incomplete. Borrowers would repeatedly submit the needed information only to have servicers repeatedly lose the documents in the frenzy.”

Such paperwork requirements were not without reason. If anyone could apply for mortgage forgiveness or forbearance, a lot of borrowers who did not need help would take it, potentially overwhelming the system and adding billions of dollars to costs that might ultimately be borne by the taxpayer. In addition to the direct monetary costs, this would have pulled human resources away from those who needed them. But in a financial crisis, you must triage. So yes, the relief programs were imperfectly executed, but there was no getting around the tradeoff between timeliness and targeting.

Another reason for the stringent application process during the Great Recession was to determine how much assistance to grant. Those programs were generally means-tested, with benefits being reduced as one’s income grew. That said, there was a big issue to address: any income information would be stale. That issue reappeared in the COVID emergency: the rapid rate of job loss meant that using last year’s tax return information would likely not offer an accurate picture of a borrower’s current need. So, we greatly minimized the qualification process, which resulted in assistance that was delivered with the speed we believed was necessary. As researchers in the Federal Reserve System later recognized, “Forbearance was especially effective due to its timeliness and the ease with which borrowers were able to take advantage of it.” Those, of course, were two of our main objectives in designing the programs.

That left us with a tough question that could not be avoided: What would be required of borrowers? At the time, the FHFA was an independent regulator, so the decision about what would be asked of millions of borrowers seeking mortgage relief rested on my shoulders. The White House had not decided what they were going to do about the government-backed mortgage programs, such as those managed by the Department of Housing and Urban Development, the Department of Agriculture, and the Department of Veterans Affairs. Having previously worked in the White House, I understood that the deliberative process within any administration is not designed to move quickly. Most of the time, that is for the best so that various stakeholders can be heard and decisions made purposefully. Facing a pandemic, though, I decided to move quickly and alone, if necessary.

The decision was to take borrowers at their word. Mortgage servicers were directed to only orally inquire whether a borrower requesting assistance had suffered a financial hardship as a direct result of COVID, such as job loss. This was a calculated risk. Borrowers who were not truly in need of assistance could have easily taken advantage of the program, which had the potential to quickly overwhelm the mortgage market.

In my media appearances, I made sure not only to inform homeowners that assistance was available and how to get it, but also to discourage those who did not need assistance from applying—perhaps a first among policymakers. We had limited resources. For starters, the FHFA, Fannie, Freddie, and the mortgage servicing industry had only so many employees. Attention paid to borrowers who did not need help would pull attention away from those who did.

Establishing assistance programs was just the beginning. Perhaps to be expected, there was some confusion about these programs on the parts of borrowers, renters, lenders, and the general public. With a handful of commentators calling for mortgage and rent forgiveness rather than forbearance, some families approached their lenders with the hope that any paused mortgage payments would be forgiven.

Whether it would have been better to just forgive rent and mortgage payments was a question far outside our scope at the FHFA. Most importantly, we did not have the legal authority to do so. It would have to be the decision of Congress, not us. Furthermore, we simply did not have the resources. Even a targeted, modest forgiveness program would have bankrupted Fannie and Freddie. They were already massively leveraged and on the verge of failure without adding tens of billions of dollars of additional losses. What’s more, the FHFA’s first responsibility was to oversee the safety and soundness of Fannie, Freddie, and the Federal Home Loan Banks. Forcing any of the entities into massive losses would have been the exact opposite of our legal responsibilities. The FHFA is not the Department of Housing and Urban Development or the Federal Emergency Management Agency. The FHFA is a safety and soundness regulator, not a grantmaking or household assistance agency.

Perhaps as importantly, broad mortgage forgiveness was also not needed. Job losses were concentrated predominantly among renters. Fannie and Freddie do not have relationships with renters. Their exposure was directly to homeowners and only indirectly to renters via the landlords. As for the homeowners, our internal data clearly showed that the overwhelming majority of Fannie and Freddie borrowers had significant equity in their homes. Less than 1 percent of Fannie and Freddie forbearance borrowers had loan-to-value ratios over 97 percent. The typical mortgage holders in forbearance had 20 to 30 percent equity in their homes. That is, they had the ability to pay back any paused mortgage payments. They also had extremely strong incentives to pay them back. For borrowers without that ability, we had other options.

Requiring borrowers to pay back any forbearance also reduced the incentive of anyone who did not need the assistance to take advantage of it. While politicians might prefer giveaways to all, we did not have that option (not that we would have pursued it if we had). Also, Fannie and Freddie are private companies, even if chartered by Congress and in conservatorship. Appropriately, there was no broader public expectation that private companies should voluntarily suffer losses or give away their products for free because of COVID. Fannie and Freddie operated under the same set of rules.

Payment deferral / One issue that we wanted to avoid was any payment shock when a borrower exited forbearance. One theory behind the 2008 crisis rests upon borrowers being pushed into default when the interest rates on their mortgages were reset. Various mortgage products—such as so-called “5/​1s” and “2/28s”—would function by having low fixed rates, such as for the first five years in the case of a 5/1, and then adjust to a market-driven rate, often some spread over the London Interbank Offered Rate (LIBOR). If the market rate at the time of reset was above the introductory rate, the borrower’s monthly payment would increase. Depending on the rates, this rise could be quite dramatic and in theory could push a borrower into default.

It was common during the early days of the 2008 crisis to hear rate resets cast as the chief cause of the predicament. I believe that once mortgage rates dramatically declined and defaults continued to mount, it became clear that resets were at worst a modest contributor to the crisis. Nonetheless, we wished to avoid a situation in which borrowers exiting forbearance would see large rises in their monthly payments. We hoped to keep any increase modest and manageable for the borrower.

At the same time, we were trying to conserve capacity on the balance sheets of Fannie and Freddie for future workout loans. It is standard practice that if a mortgage becomes delinquent and must be modified in some manner to achieve sustainability—usually in the form of a rate reduction or extension of the term—then that loan is removed from the pool of loans that serve as collateral for agency-issued mortgage-backed securities. When that happens, the loan is generally moved onto the balance sheet.

An additional concern was to minimize any disruption to the mortgage-backed securities market, which had already experienced a rough ride by March 2020. If we modified loans on a massive scale, we would have to remove them from the mortgage pools and replace them mostly with lower-rate loans. The result could be a sudden and significant decline in the value of mortgage-backed securities. Some of this declining value was likely unavoidable given the rate environment, but we could at least slow that decline, bringing a little more price stability to that market, which ultimately determines the price that households pay for their mortgages.

The solution we settled on was to add any missed payments to the end of the loan, what we called “payment deferral.” Those payments would essentially be a “balloon.” Perhaps they would be paid back once the home was sold, coming out of the proceeds like any normal mortgage. A more likely outcome, because of the low-rate environment, was that the missed payments would be rolled into the loan amount when the loan was next refinanced. In such cases, most borrowers would still see a substantial reduction in their monthly payments, even with the higher loan balance.

Results / Ultimately, about 8 million borrowers—roughly 1 in 10 homeowners—entered mortgage forbearance during COVID. By 2022, over 90 percent of them would exit forbearance, getting back on their feet, at least in relation to their mortgage. In fact, more than half would be in forbearance for three months or less. And 15 percent would take forbearance for only a single month.

About a fifth of those borrowers continued to make their monthly payments, despite having entered forbearance. Another fifth repaid their missed payments in one lump sum. Over a fourth paid off their existing mortgage, usually by taking advantage of the record low rates to refinance.

Still, more than two years later, just under half a million borrowers who entered COVID forbearance remain behind on their mortgages. Over 70,000 of those have entered foreclosure. It should be noted that the vast majority of those were delinquent before the pandemic, but others have been unable to recover lost jobs or income.

Forbearance rates for Fannie and Freddie loans peaked around 6 percent in May 2020. In contrast, the experiences of government-backed mortgage programs were much worse. The FHA saw forbearance rates spike to almost 15 percent and remain in double digits well into 2021. Loans guaranteed by the Department of Agriculture’s Rural Housing Service did only slightly better.

Conclusion

Mortgage forbearance offered during COVID was a success, especially compared with similar efforts during the Great Recession. But the parameters of the mortgage response, however appropriate for a pandemic, might not be the best response for the next housing finance emergency. And there will be another emergency of some sort. Most likely, it will match the historical trend of a monetary-induced increase in mortgage rates leading to a decline in housing demand, followed by weakness in housing prices, which when mixed with a spike in unemployment leads to mortgage distress.

The precedents set during COVID that are likely to be useful next time include the following:

  • basing duration of forbearance on a fixed time frame, not on changes in income or employment;
  • allowing borrowers to attest to hardship first, with servicers verifying eligibility later; and
  • requiring that missed payments be paid back, but not all at once.

In short, be quick and clean and directly address the issue at hand.

One concern about forbearance design for next time is the possibility that borrowers who do not need assistance take it. In 2008, many distressed borrowers blamed their lenders for their situation. In some cases, there was good reason to do so. That said, such feelings made the problem harder to solve. Forbearance and foreclosure are expensive, even when successful. Fortunately, borrowers did not abuse the forbearance option during COVID, at least not Fannie and Freddie borrowers. Next time, however, if the emergency is accompanied by widespread borrower anger directed at lenders, we might see the return of strategic forbearance.

The success of forbearance during the pandemic might suggest that Fannie and Freddie should be kept in permanent conservatorship because it was the legal vehicle for implementing forbearance. Such a conclusion is unwarranted. The problematic mortgage mitigation programs of the Great Recession were conducted during the conservatorship, and they did not appear to improve the agencies’ performance. We have also witnessed similar forbearance options offered on mortgages not covered by the CARES Act. Even more telling is that similar levels and flexibility of forbearance were offered in the auto loan market, where there are no government-sponsored enterprises. Now that the FHFA has lost some of its independence, the agency’s ability to move quickly may well be compromised in the future. If anything, our mortgage market will be better prepared for the next crisis, with Fannie and Freddie fully capitalized and operating outside conservatorship.

More importantly, I fear we will lack the leadership to make unpopular choices. My telling borrowers not to take assistance if they did not need it was not popular. I could not imagine an elected official disseminating that message. Requiring missed payments to be repaid was not popular, especially when it appeared that trillions of dollars were being given away by Congress. And making sure it was all paid for—using a mortgage fee to recoup the cost of COVID assistance—was definitely not popular. While not as obvious at the time, paying for the assistance provided by Fannie and Freddie also meant that the aid was not adding to the inflationary pressures experienced later in the pandemic.

Assistance that helped keep just under 3 million families in their homes during a pandemic was provided by Fannie and Freddie at the direction of the FHFA. I remain extremely proud of that effort. In comparison, HAMP provided permanent assistance to 1.5 million borrowers, and about a third of those eventually defaulted. Perhaps most shocking was HAMP’s slow rollout. A year into the program, just over half a million borrowers had received permanent assistance, whereas the FHFA helped almost six times that number in the first year of COVID. If the sluggish rollout of HAMP was not bad enough, the program proved expensive, costing taxpayers well over $20 billion.

The FHFA plan created a funding mechanism that ensured that assistance would have no cost to the taxpayer. It produced no visible disruption in the mortgage market. Its recovery fee brought to government a level of transparency that is rarely seen.

All government choices have costs. Making those costs explicit is not an excuse for inaction. We acted, and we acted quickly. And beneficiaries ultimately paid for it, not the taxpayer. This should be the model for future responses, not the endless subsidies and bailouts that have all too often become the norm.

Readings

  • “Forbearance during COVID-19: How Many Borrowers Used It, and for How Long?” by Juan M. Sánchez and Olivia Wilkinson. On the Economy Blog, Federal Reserve Bank of St. Louis, May 31, 2022.
  • “Foreclosures, Enforcement, and Collections under the Federal Mortgage Modification Guidelines,” by Casey B. Mulligan. National Bureau of Economic Research working paper no. w15777, 2010.
  • “Lessons Learned from Mortgage-Borrower Policies and Outcomes,” by Kristopher Gerardi, Lauren Lambie-Hanson, and Paul Willen. In Recession Remedies, edited by Wendy Edelberg, Louise Sheiner, and David Wessel. Brookings Institution Press, 2021.
  • “Means-Tested Mortgage Modification: Homes Saved or Income Destroyed?” by Casey B. Mulligan. National Bureau of Economic Research working paper no. w15281, 2009.
  • “Reducing Strategic Forbearance under the CARES Act: An Experimental Approach Utilizing Recourse Attestation,” by Jackson T. Anderson, David M. Harrison, and Michael J. Seiler. Journal of Real Estate Finance and Economics 65(2): 230–260 (2022).
  • “The Effect of Perceived Lender Characteristics and Market Conditions on Strategic Mortgage Defaults,” by Michael J. Seiler. Journal of Real Estate Finance and Economics 48(2): 256–270 (2014).
  • The Subprime Virus: Reckless Credit, Regulatory Failure, and Next Steps, by Kathleen C. Engel and Patricia A. McCoy. Oxford University Press, 2016.