Solving the Wrong Problem

Nearly all of the nineteen essays solicited by the Urban Institute in its “Housing Finance Reform Incubator” series have been submitted. We have all ten on single-family financing, all three on multifamily, and five of the six on affordability and access. This post focuses on the essays on single-family and affordability and access. (Contributors to the multifamily essays seem to agree that Fannie and Freddie’s programs work well, and that while improvements can be made, reforms in multifamily are less extensive and less urgent than on the single-family side).

On the whole, I found the single-family essays to be disappointing. I began my own essay by stressing the need for each proposal to have a stated objective against which it could be measured. My objective focuses on the system: “to create a capital markets-based secondary market mechanism capable of financing at least $1 trillion of 30-year fixed-rate mortgages annually throughout the business cycle, at the lowest cost to homebuyers consistent with an agreed-upon standard of taxpayer protection.” In every other essay, to the extent there is a stated objective it is some version of “fixing the problems with Fannie Mae and Freddie Mac.” How those “fixed” systems would work in practice—particularly for the low- and moderate-income homebuyers Fannie and Freddie originally were chartered to serve—is not addressed in sufficient detail in any essay to permit evaluation on the merits.

On the positive side, all of the essayists except Alex Pollock either explicitly or implicitly reject the notion that Fannie and Freddie were the causes of the mortgage crisis. Andrew Davidson and Jim Millstein both recognize and discuss the true problems, while the “principles for reform” Pat Mosser articulates in her essay make clear she has a sophisticated understanding of what took place to trigger the crisis. Still, they along with the majority of the other essayists suggest sweeping changes to the form or structure of Fannie and Freddie, all requiring legislation.

Why? Of the four essayists (other than me) that give detailed reform proposals— Millstein, Davidson, Mosser and Mark Zandi—all cite some form of incentive problem at Fannie and Freddie. Millstein says, “there is an inherent conflict of interest between the GSEs’ obligation to promote access and affordability and the private market’s imperative to maximize shareholder value,” while Davidson says, “The GSEs placed shareholder gain over risk management and were severely undercapitalized.” Mosser does not give a specific criticism of the companies, but notes that her proposed change to mutualized ownership is “designed to address the first principle of incentive alignment.” Finally, Zandi states, “our reliance on this duopoly [Fannie and Freddie] created perverse incentives that ultimately led to too much risk taking, forcing taxpayers to shoulder the resulting cost.”

The remedies the four essayists propose differ, although they have features in common. Two—Mosser and Davidson—would convert Fannie and Freddie to mutual ownership. Zandi would “merge Fannie and Freddie to form a single government corporation.” Millstein would have the companies form new credit guaranty subsidiaries that ultimately could be spun out as stand-alone entities, either with or without government guarantees, depending on what Congress is willing to accept at the time the spin-offs occur. Zandi and Davidson would transfer all (Zandi) or most (Davidson) of the responsibility for grading, pricing and managing mortgage credit risk from Fannie and Freddie to third parties, principally capital markets investors. And with the possible exception of Millstein, all believe the guarantor should have an explicit line of credit from the government to cover catastrophic risk, which it would pay for.

I have reservations or concerns about many of these proposed changes, but my main criticism of the essays is not that; it’s that they spend most of their time solving the wrong problem. The biggest problem Fannie and Freddie had leading up to the crisis wasn’t their incentive structure—that was good enough to enable them to produce and maintain a level of credit quality that was far higher than all other sources of mortgages at the time—it was capital. And the four authors spend much too little time analyzing and discussing the amount of capital that should be required in the new system, how that amount should be determined, where it would come from, the implications of the recommended (or assumed) capital scheme on the mechanics of delivering credit guarantees to borrowers, and the consequent availability and affordability of mortgages, particularly for low- and moderate income homebuyers.

Two of the essays in the affordable housing category—one by John Taylor and the other by Mike Calhoun and Sarah Wolff—correctly focus on this omission. Taylor notes, “It’s unclear whether guarantors could raise sufficient capital to support [proposed] reform models…not to mention whether that type of capitalization requirement and the guarantee fees it implies would simply price many low- and moderate-income borrowers out of the conventional market altogether.” Calhoun and Wolff build on this point, adding, “Estimates of how structural changes will affect…the rates consumers pay on their mortgage…nearly all provide a combined estimate or estimate costs for a “typical” borrower. What is lacking is analysis of how costs will be distributed” [emphasis in original].

To be effective, a secondary mortgage market financing system for 30-year fixed-rate mortgages needs to be able to tap large volumes of funding from international capital markets investors reliably and consistently, then channel those funds to a wide range of homebuyers on terms they can afford. The key to the first is giving mortgage-backed security investors a credit guaranty they trust; the key to the second is keeping the cost of that guaranty low enough that a credit guarantor can use cross-subsidization to offer affordable guaranty fees to higher-risk borrowers without pushing fees on lower-risk business so high that it gets financed elsewhere, increasing the risk on the overall book. The Urban Institute essayists agree that an explicit government guarantee can accomplish the former; they are largely silent on how best to achieve the latter.

There is a telling table in the longer “Promising Road” paper that complements the Zandi essay, giving a build-up for the guaranty fees the authors say would result under their proposal, the current system, and three alternative financing systems. In all cases, the expected annual credit loss on the loans being guaranteed is 4 basis points, which to me seems reasonable and perhaps even high. Fannie’s average realized single-family credit loss rate during the fifteen years I was CFO was less than 4 basis points per year, and its current books of business are notably better than when I was there (more on that in a moment). The “Promising Road” authors put the average single-family guaranty fee under the current system at 70 basis points (Fannie and Freddie’s charged fee actually has been 60 basis points), then show that fee to be “only” 90 basis points under their proposal, compared with fees ranging from 106 to 136 basis points for the other three systems.

But here is where there needs to be a reality check. If you have to charge between 90 and 136 basis points basis to insure against a risk that has an expected cost of 4 basis points, you don’t have a workable business model. You will have priced your product so high that you won’t have enough customers to enable you to attract the amount of equity capital you need to get that business running in the first place. The fact that 90 basis points is lower than 136 is irrelevant; 90 itself is far too high.

Even today’s credit guaranty system—with Fannie and Freddie in conservatorship and charging an average of 60 basis points in guaranty fee (including the 10 basis point payroll tax fee imposed by Congress in 2012)—is operating at the fringe of effectiveness. As several of the authors in the Urban Institute series point out, the “credit box” today is much smaller than it was before the crisis, despite (or probably because of) the much higher average guaranty fees. This is particularly notable with credit scores. In the 2000-2002 period—before underwriting standards began to collapse—37 percent of the loans Fannie purchased or guaranteed had credit scores under 700, which are typical for affordable housing borrowers; in 2013-2015, just 17 percent of Fannie’s loans had sub-700 credit scores.

The main reason for this dramatic difference almost certainly is pricing. In 2014, FHFA published a table that broke down Fannie and Freddie’s modeled and charged guaranty fees in the first quarter of that year by credit score and loan-to-value ratio. The companies’ average “fully priced” guaranty fee for loans with credit scores of 700 or higher was 61 basis points; for loans with sub-700 credit scores that fee was more than double, at 124 basis points. The companies dealt with this disparity primarily by not charging the fully priced fees. To attract at least some sub-700 credit score business, Fannie and Freddie charged only 76 basis points for it, not 124, and to avoid losing higher-quality business they cut those fees, too, from 61 basis points to 57 basis points. As a consequence, their average charged fee was only 60 basis points, 12 basis points less than the fully priced fee of 72 basis points. And even then, the companies still served a much narrower segment of the borrowing public than they had been able to do prior to the crisis.

This situation will not get any better under any of the current popular reform proposals. A credit guarantor operating under a binding capital standard that requires a 72 basis point fee to earn its target return will be much less willing to cut guaranty fees to attract business than Fannie and Freddie have been. That guarantor will face the alternative of trying to entice at least some lower-credit score business at a 124 basis point fee, or cutting this fee and having to make up for it by charging more than 61 basis points for higher-quality business (at the risk of losing it, and driving up the risk on their total book). Whichever the guarantor chooses, it will have an even lower percentage of lower-credit score business than Fannie and Freddie have now, and lower overall business volumes as well.

Given this current reality, mortgage reformers shouldn’t be trying to “fix Fannie and Freddie’s incentive structure” in a vacuum; they should be trying to figure out a way to make the secondary mortgage market finance system work in a post-crisis world.

The key is capital. Many reformers seem to have gotten off track on this in 2013, when they endorsed the Corker-Warner legislation with its capital requirement of 10 percent. Ten percent always was an arbitrary number—and it always was unreasonable and unworkable given the risks in the credit guaranty business—but with that as the starting point, it made the capital issue look too easy. Lowering capital requirements to 4 or 5 percent seemed like a huge concession, and all that needed to be done. It wasn’t, and isn’t. More capital is not better; what’s needed is the right amount of capital: enough so there is only a miniscule chance of triggering the government’s catastrophic risk guaranty, but not so much that the resulting guaranty fees have to be so high that the guarantor cannot produce a sufficiently large and diverse volume of business to attract the capital required to back it, while meeting expectations for breadth of service to affordable housing constituents.

So far, the Urban Institute’s essayists have done part of their task—coming up with ideas for the business structure of the guarantor—but they’ve done it out of context. Next, they should come back with their ideas on capital and pricing: how much capital credit guarantors should have and why that’s the right amount; the specific mechanism they envision for transmitting guaranty fees and mortgage rates consistently, efficiently and affordably to the populations Fannie and Freddie have traditionally served; and, if there is a significant transition from the current system to the envisioned one, how that transition would work in practice. As the authors go through this exercise, the correct structure for the guarantor should become apparent to them: it will be the one that makes the overall system work best for homebuyers and the government alike.