Capital Fact and Fiction

Late last year, the Federal Housing Finance Agency (FHFA) formally adopted a regulation proposed by former Director Mark Calabria that raised Fannie Mae and Freddie Mac’s required capital by more than 80 percent compared with what would have been in effect under the FHFA standard proposed in June of 2018. Soon afterwards, this new capital rule was incorporated into a January 14, 2021 letter agreement between Calabria and former Treasury Secretary Steven Mnuchin, stipulating that before the companies could be eligible for release from conservatorship under a consent decree they would need to hold core (or tier 1) capital of at least 3 percent of their adjusted total assets—at June 30, 2021, about $230 billion—for two or more consecutive quarters. And then this June, the controversial ruling by the Supreme Court that the net worth sweep was a legal action by FHFA allowed Treasury’s $195 billion in senior preferred stock in the companies to remain outstanding. Because Treasury’s senior preferred is not considered core capital under the Housing and Economic Recovery Act (HERA), this put Fannie and Freddie’s combined core capital at June 30, 2021 at a negative $134 billion—more than $360 billion below the threshold for their conditional release from conservatorship under the Calabria-Mnuchin agreement.

Together, these developments left Fannie and Freddie in an impossible position. Their capital problem is obvious. With the net worth sweep blocking their access to the equity markets, it would take them nearly two decades to retain the earnings they need to fill a $360 billion capital gap, and longer if their business grows during that time. Less obvious, but no less serious, is their business problem. For the very long time they would remain in conservatorship, they would have no alternative but to price their credit guarantees to capital requirements that are grossly disproportionate to their credit risk, greatly raising the cost of and restricting the access to homeownership for the low- moderate- and middle-income homebuyers they are chartered to serve.

Fannie and Freddie’s situation is the result of a lengthy sequence of events that began with the decision in the last year of the Bush administration to force them into conservatorship, against their will and while each exceeded its statutory capital requirement, for policy reasons but under the pretense of a rescue. Throughout the Obama administration and for much of the Trump administration, FHFA and Treasury managed Fannie and Freddie in conservatorship in a manner intended to encourage Congress to replace them legislatively with a “free market” alternative—crippling the companies financially by forcing them to take $187 billion in non-repayable Treasury senior preferred stock at a 10 percent after-tax annual dividend that they did not request and did not need, falsely blaming them for the financial crisis, and burdening them with “bank-like” capital requirements unrelated to the risks of the one business they are permitted to engage in. There now is a consensus that replacing Fannie and Freddie with a free-market alternative that exists in theory but not in practice is neither achievable nor desirable. Yet the companies remain entangled in the consequences of past actions and fictions aimed at accomplishing that objective.  

It has fallen to the senior economic officials of the Biden administration to sort this out. Fannie and Freddie are the cornerstones of the $13.6 trillion U.S. residential mortgage market. They have been successfully “conserved,” but because of decisions made in prior administrations—whose objectives the current administration does not share—they are locked indefinitely in conservatorship, and unable to help the president meet his housing policy objectives. There is, however, a simple way for the Biden administration to end the companies’ state of limbo: acknowledge the facts about Fannie and Freddie, admit that their de facto nationalization in 2008 was unjustified and a mistake, and then undo it.  

The “undoing” can be achieved through two complementary actions. First, declare that Fannie and Freddie have paid back all of the $187 billion they were forced to draw during the crisis, including 10 percent interest (which they have done), and deem Treasury’s senior preferred stock to have been repaid and cancel it, along with Treasury’s liquidation preference. This would change Fannie and Freddie’s core capital at June 30, 2021 from a negative $134 billion to a positive $60 billion. Then, appoint a new Director of FHFA, and task him or her with replacing the Calabria capital rule with one based on the companies’ actual business and credit risks. As depicted and discussed in the sections below—which draw on publicly available past and current data from Fannie—there is a huge difference between the capital required by a risk-based standard for Fannie and Freddie based on fact, and one based on fictions invented by those who oppose the companies on ideological or competitive grounds. And, as we also will see, developing and implementing a fact-based capital rule for the companies is astonishingly easy. 

Capital fact

Thirteen years after the Great Financial Crisis began, we know how much initial capital Fannie would have required to cover the credit losses of the $2.531 trillion book of single-family business at December 31, 2007 that went through it. There are two ways this “stress capital” amount can be expressed: for the liquidating book, and on a going concern basis. The former tracks credit losses on the 2007 book over its entire life, and uses as offsets to credit losses only guaranty fee income, net of administrative expenses, from loans in this book that do not prepay. In contrast, the going-concern method incorporates income (and losses) from new business, and runs only until annual total net guaranty fees, from both existing and new business, exceed annual total credit losses, and profitability is restored.

Fannie’s published data make it possible to track the losses on its December 2007 book of single-family business with precision through 2016, and to make very close estimates in the years beyond that. Through December 31, 2012—just before Fannie’s total annual guaranty fees exceeded its annual credit losses as a going concern in 2013—the losses on its 2007 single-family book were $70.0 billion, or 2.77 percent of the December 31, 2007 book. Those losses rose to $91.2 billion at December 31, 2020, for an ever-to-date credit loss rate of 3.60 percent. (Losses from the 2007 book since then have been de minimis.)

We also can use published data to make a conservative approximation of the amount of these losses that were absorbed by guaranty fees. The average guaranty fee rate on Fannie’s December 2007 single-family book was 24.2 basis points, which after 8.0 basis points in administrative expenses left a net guaranty fee of 16.2 basis points. The 2007 book prepaid at a 19.9 percent annual rate through the end of 2012, and at an overall annual rate of 20.9 percent through the end of 2020. From past experience, we know that lower-risk, lower-fee business prepays faster than higher-risk, higher-fee business, so assuming that the average guaranty fee on Fannie’s 2007 book stayed level over time produces a conservative (that is, low) estimate of cumulative guaranty fee income of $14.7 billion through 2012, and $19.3 billion through 2020. For Fannie’s liquidating book, $19.3 billion in net guaranty fees would have absorbed 76 basis points of its 360 basis points of ever-to-to date credit losses, leaving 284 basis points to be covered by initial capital.

On a going-concern basis, the initial capital requirement for Fannie’s 2007 book is almost 100 basis points lower, because it only has to cover stress credit losses until 2013, when its net guaranty fees of $7.2 billion exceed its total single-family credit losses of $4.5 billion (and then continue to), and because some losses can be absorbed by revenues from new business. As a going concern, the relevant numbers for Fannie’s 2007 book of single-family business during the stress years of 2008-2012 are: credit losses from the original 2007 book of $70.0 billion; credit losses from business done after 2007 (principally from the 2008 book) of $5.6 billion; net guaranty fees from the 2007 book of $14.7 billion, and net guaranty fees from post-2007 new business of $13.0 billion. Together, these result in total credit losses of $75.6 billion, total net guaranty fees of $27.7 billion, and losses that need to be covered by initial capital of $47.9 billion—equal to 189 basis points of the $2.531 trillion in single-family mortgages Fannie owned or guaranteed at December 31, 2007.  

But the story does not end there. Since 2007, the credit quality of the company’s single-family business has improved markedly, and its net guaranty fee has more than doubled. The credit quality increase came about because of the sweeping changes in residential mortgage underwriting standards since the crisis, many mandated by the 2010 Dodd-Frank legislation. Data put out by Fannie indicate that between 50 and 55 percent of the credit losses on its 2007 book were from Alt-A or interest-only loans, now prohibited by Dodd-Frank. Taking just these two loan types (about $425 billion) and their associated losses (about $48 billion) out of Fannie’s 2007 single-family book lowers its lifetime credit loss rate from 360 to 205 basis points. And incorporating all underwriting improvements since 2007 lowers the loss rate still further. In its 2018 capital proposal, FHFA stated that “using current acquisition criteria” the credit loss rate on Fannie’s 2007 book of business through September 2017 would have been just 1.5 percent. Meanwhile, on the revenue side, the 16.2 basis-point average net single-family guaranty fee on Fannie’s December 2007 book has risen by 21 basis points, to 37.2 basis points at June 30, 2021.

Cutting the stress credit loss rate by over half and more than doubling the net guaranty fee have a dramatic effect on the amount of capital Fannie’s single-family book would require to survive a repeat of the Great Financial Crisis home price collapse. FHFA’s loss rate of 150 basis points through September 2017 for Fannie’s 2007 liquidating book “using current underwriting criteria” projects to 160 basis points of losses through December 2020. A net guaranty fee rate of 37.2 basis points, with prepayment rates equal to those of the 2007 book, would have produced 165 basis points of guaranty fee income during that time—5 basis points more than the credit losses. And the more conservative estimate of 205 basis points of losses from the current book’s underwriting (adjusting only for its lack of Alt-A and interest-only mortgages) still would have required Fannie’s 2007 liquidating book to have had only 40 basis points of initial capital to survive the Great Financial Crisis.        

On a going-concern basis, today’s underwriting and net guaranty fee produce equally striking stress results, even using the 205 basis-point loss estimate for the current book, and assuming that both the size of the book and average guaranty fee remain level, rather than increase by 12 and 17 percent, respectively, as they did between 2007 and 2012. A net guaranty fee of 37.2 basis points on a level $2.531 trillion book of business produces $9.42 billion of revenue per year. Using the same pattern of credit losses as Fannie actually experienced between 2007 and 2020, a 205 basis-point cumulative loss rate through 2020 results in total single-family credit losses (from old and new business) that start at $3.69 billion in 2008, rise to $13.66 billion in 2010, and fall to $8.73 billion in 2012. At a constant $9.42 billion per year, cumulative guaranty fee revenues exceed cumulative credit losses by $6.69 billion during 2008 and 2009. Revenues fall short of credit losses by $6.33 billion in 2010 and 2011, but then are greater than credit losses again in 2012 and beyond. At no time during the stress period do cumulative losses ever exceed cumulative revenues.  

The reality, therefore, is that based on Fannie’s historical experience, and incorporating its current single-family net guaranty fee and a conservative estimate of the credit quality of its current single-family book of business, the company as a going concern could survive a repeat of the 25 percent drop in nationwide home prices that characterized the Great Financial Crisis—on which the HERA risk-based capital standard is supposed to be based—without the need for any initial capital. Further, FHFA effectively confirmed this last month, when, with virtually no publicity, it released the results of the 2020 and 2021 Dodd-Frank “severely adverse” stress tests, run on Fannie’s business for the end of 2019 and 2020. The company needed no initial capital to survive those either.

Capital fiction

How, then, did former FHFA director Calabria arrive at the requirement that in order to meet his standard of safety and soundness at June 30, 2020, Fannie must have capital equal to 4.55 percent of its $3.761 trillion in total assets, or $171 billion? The short answer is that he concocted it. He ignored the directive in HERA to base Fannie and Freddie’s capital on risk, and instead began with a pre-determination to require them to hold at least the 4.0 percent minimum capital as banks hold for their on-balance sheet residential mortgages, on which they take interest rate as well as credit risk. He then used four contrivances—not counting guaranty fees in the risk-based capital stress test; subjecting low-risk loans to a minimum risk weight; misapplying the going-concern buffer, and adding a capital penalty for doing more than a small amount of credit guaranty business—to artificially engineer a result for the required amount of Fannie’s “risk-based” capital that was greater than his arbitrary minimum of 4.0 percent. Reviewing each of these briefly:

Guaranty fee income in stress tests. Fannie and Freddie are not banks; their one permitted line of business is guaranteeing the credit of residential mortgages. For a credit guarantor, the role of capital is to absorb losses that exceed the dollar amount of revenues from loans that continue to pay throughout a period of stress. This capital amount is determined by running a stress test—in Fannie’s case, one based on a 25 percent nationwide decline in home prices. Not counting any revenues during that stress test produces a meaningless answer. The Federal Reserve counts income in the Dodd-Frank stress tests it runs for large systemically important banks, and FHFA counts income in the Dodd-Frank stress tests it runs for Fannie and Freddie. FHFA’s 2018 capital rule, issued under Director Watt, did not count revenues either, and almost all commenters on that rule said that it should. FHFA calls not counting guaranty fee income in the stress test “conservative,” but it is far more than that. Conservatism in running a stress test would be increasing the prepayment rate on surviving mortgages from the 20.9 percent annual rate experienced by Fannie’s 2007 book of business to some higher annual percentage; setting it at 100 percent (which is what not counting guaranty fees at all does) is indefensible, and invalidates the result.

Net credit risk capital. In its June 2018 capital proposal, FHFA said that using the product and risk characteristics of Fannie’s September 30, 2017 book of business, the ever-to-date credit losses from Fannie’s 2007 book would have been 1.5 percent. This translates to a lifetime loss rate in the 1.65 percent range. The grids, risk multipliers and counterparty “haircuts” used in the 2018 FHFA capital rule pushed this net credit risk capital up to 2.1 percent of total assets. Calabria wanted an even higher number, however, so he put a 20 percent risk weight, or a 1.6 percent capital floor, on all mortgages, which made the net credit risk capital requirement on Fannie’s June 30, 2020 book—whose credit quality was at least as good as the September 30, 2017 book—289 basis points of total assets, or about 125 basis points more than the actual expected lifetime stress losses of that book.

Going concern capital. FHFA’s June 2018 capital rule included a “going concern buffer” of 75 basis points of capital, to “to ensure that each Enterprise would continue to be regarded as a viable going concern by creditors and other counterparties after a severe economic downturn.” Calabria’s 2020 capital rule renamed this the “stress capital buffer,” but kept it at 75 basis points. Both versions of the rule, however, conveniently ignore the facts that Fannie’s risk-based capital requirement is based on a liquidating book of business, and that shifting to a going-concern perspective results in the company having a large amount of loss-absorbing revenue from new business, and also a shorter stress period. During and after the Great Financial Crisis, Fannie needed almost 100 basis points less initial capital to survive as a going concern than it did to cover the losses on its liquidating 2007 book. Put differently, because Fannie’s risk-based capital requirement is (very conservatively) based on a liquidating book of business, it already has a going-concern buffer built into it.

Stability buffer. What Calabria calls the “stability buffer” of 110 basis points of total assets added to Fannie’s June 30, 2020 risk-based capital requirement is unrelated to risk, totally arbitrary, and based on a mischaracterization of Fannie and Freddie’s market share. They are the only two companies whose sole business is guaranteeing the credit of conventional (non-government-guaranteed) residential mortgages. Expressing their size as a percentage of the business done by entities which can originate, service or hold mortgages in portfolio, and then requiring Fannie and Freddie to hold more capital the further the volumes of their credit guarantees rise above a tiny 5 percent share of activities they are not permitted to engage in, is preposterous, and a blatant and clumsy means of penalizing their conventional credit guaranty function in favor of portfolio investing, which demonstrably is far riskier. The stability buffer has no place in the companies’ risk-based capital standard.

*                                           *                                                  *

On the day Mark Calabria stepped down as Director of FHFA (at the request of President Biden), he said, “When the housing markets experience a significant downturn, Fannie Mae and Freddie Mac will fail at their current capital levels.” He had to know this was untrue as he was saying it; he had been sitting on the unpublished results of the 2020 stress test run on Fannie, showing that it could survive the Federal Reserve’s “severely adverse scenario” of a 4 percent drop in real GDP and a 4 percent rise in unemployment with no need to draw on its capital, now $37.3 billion, at all. Yet he had used his position to mislead the public, and also to convince the Financial Stability Oversight Council to not just endorse his wholly contrived 4.55 percent risk-based capital requirement for Fannie, but to go beyond it, by saying in a joint statement with FHFA, “Risk-based capital and leverage ratio requirements materially less than those in the proposed rule would likely not adequately mitigate the potential stability risk posed by the Enterprises. Indeed, more capital might be necessary.”  

This fictional approach to Fannie and Freddie’s capital has gone unchallenged since the beginning of the conservatorships. The Biden economic team must change that, and it can start by acknowledging and embracing the incontrovertible facts about the companies’ past and present credit risk, and instructing a new Director of FHFA to replace the Calabria capital standard with one that draws on those facts. The Calabria standard is not fixable. It was constructed to produce an artificial result, and has too many unrealistic elements that interact in ways that have unintended consequences, such as the refusal to count guaranty fees in the stress test, which pushes the large majority of the cost of that artifice onto the borrowers of affordable housing loans, who are the least able to bear it. Moreover, it is completely unnecessary to have a risk-based standard that is as mind-numbingly detailed and complex as the current one. FHFA can provide an unquestionably high level of taxpayer protection, and at the same time allow Fannie and Freddie to price their credit guarantees on an economic basis, with a far simpler and fully transparent approach to capital.

A rigorous and highly effective capital regime for Fannie and Freddie can be built with just three elements: (a) a true risk-based capital requirement based on a stress test run on each company’s book of business every quarter, with no cushions or add-ons; (b) a single “all purpose” capital cushion, calculated as a percentage of this true risk-based requirement, and (c) a minimum capital percentage. Fannie and Freddie’s required capital would be the greater of the risk-based amount (plus the capital cushion) and the minimum percentage. To implement this regime, FHFA only would need to decide on two numbers: the multiple to be applied to the stress test results (which will change each quarter, based on changes in the companies’ mix of business) that determines the size of the risk-based capital cushion, and the minimum capital percentage. And for each of these, there is one compelling choice.

As basic principles, the multiple FHFA should apply to the results of Fannie and Freddie’s quarterly stress tests should be large enough to provide meaningful protection against unanticipated events or model error in the stress test, but not so large as to override the true risk-based element, and distort the companies’ credit pricing. Thirty percent—or a 1.3 multiple—seems about as large as a capital cushion could be without violating the second principle. Also, in determining the size of the cushion, it is critical to bear in mind that the amount of capital Fannie and Freddie actually will hold—and that will serve as the basis for the pricing of their credit guarantees—will be more than the greater of the minimum or the risk-based standard. To avoid regulatory sanction, company management must meet both their minimum and risk-based standards. If they draw on their capital cushion, whatever its size, they will be subject to prompt corrective action. To avoid that, they invariably will hold excess capital, typically 10 to 20 percent of their required amount. During periods of stress they will first draw down that excess capital, then, if necessary, raise new equity.

For Fannie and Freddie’s minimum capital requirement, the clear choice is 2.5 percent, for three reasons. First, a 2.5 percent capital requirement for the companies’ on-balance sheet assets already is in HERA, as is a 45 basis-point requirement for their off-balance sheet credit guarantees (on which they take only credit risk). When the FASB required Fannie and Freddie to put credit guarantees on the balance sheet in 2010, FHFA sensibly kept their minimum capital at 45 basis points by regulation, since their economic substance hadn’t changed. If FHFA simply removes that regulatory waiver, the result will be to more than quintuple the required capital on credit guarantees relative to pre-crisis times—“serious reform” by any measure. Second, in FHFA’s June 2018 capital proposal, “2.5 percent of on-balance sheet assets and off-balance sheet guarantees” was the more conservative of the two alternatives for Fannie and Freddie’s minimum capital (the other was “1.5 percent of trust assets [MBS] and 4 percent of non-trust assets”). And third, 2.5 percent also is the minimum in the 2020 final Calabria standard—only without the (excessive and unjustified) “prescribed leverage buffer amount” of 1.5 percent, which brings his total to 4.0 percent.

Were this capital regime and minimum percentage to be adopted and implemented today, the 2.5 percent minimum would be far above Fannie and Freddie’s risk-based requirement, and thus be binding. At June 30, 2021, 2.5 percent of their total assets (not “adjusted total assets,” a concept FHFA should drop) was $175 billion. With the Biden administration at the same time deeming Fannie and Freddie’s senior preferred stock repaid and canceling Treasury’s liquidation preference, the companies’ core capital would rise to a positive $60 billion, leaving them with a gap to adequate capitalization of $115 billion.

Fannie and Freddie’s after-tax earnings over the past four years have averaged more than $20 billion per year. With only a $115 billion capital gap, a capital standard that permits them to price their credit guarantees fairly, and without the net worth sweep, the resulting strong investor demand for their equity would open up many paths out of conservatorship for the companies. One would be for FHFA, with Treasury approval, to set a new threshold for their release under a consent decree—say 1.5 percent of total assets—that would take effect upon settlement of an equity issue that meets or exceeds this threshold. Then, each company would be allowed to decide the mix of retained earnings, issues of common stock, or issues of noncumulative junior preferred stock to use to eliminate its remaining capital gap, and obtain its release from conservatorship unconditionally. Other alternatives for a quick exit from conservatorship undoubtedly would be proposed as well.

During previous administrations, following the fictions about Fannie and Freddie has led to their interminable conservatorships. To get them out, the Biden administration merely has to change course and follow the facts. When it does, all of the companies’ stakeholders—and particularly low- moderate- and middle-income homebuyers—will reap the benefits.