On May 20, the Federal Housing Financing Agency (FHFA) released its re-proposed capital rule for Fannie Mae and Freddie Mac. The new standard was bank-like both qualitatively—expressing required capital amounts not as percentages but as “Basel risk weights”—and quantitatively, requiring the same 4.0 percent minimum capital for the companies’ credit guaranty business as banks must hold for the mortgages they retain in portfolio (and on which they take liquidity and interest rate risk as well as credit risk). In addition, FHFA added a number of new cushions and elements of conservatism to the June 2018 risk-based standard to bring its required capital up to a level approximating the proposed new minimum. Under the May 20 proposal, total combined required capital for Fannie and Freddie as of September 31, 2019 was 71 percent above the June 2018 requirement.
FHFA requested comments on the May 20 proposal, due on August 31. The majority of the 80 detailed comments posted to the FHFA website were critical, and most critics made the same four points: that (1) Fannie and Freddie were not banks and did not have the risks that banks do, so applying Basel bank capital requirements to them was not appropriate; (2) a 4.0 percent minimum capital requirement that generally would be binding on the companies gave them a perverse incentive to take risk, and discouraged the use of credit risk transfers (CRTs); (3) the large amount of add-ons and conservatism in the risk-based standard made it not risk-based in practice, and (4) FHFA had not given sufficient capital credit for the use of CRTs.
Tellingly, even the American Bankers Association and the Mortgage Bankers Association were critical. The former said, “While we believe the re-proposal addresses some concerns raised by ABA and others with regard to the previous proposal, the re-proposal raises concerns of its own, particularly with regard to the implications for the primary market and our members’ continued ability to sell loans to the GSEs in the revised GSE marketplace implied by the re-proposal.” And the MBA stated flatly, “The level of required capital implied by the framework is too high and may be determined too frequently by a leverage ratio rather than risk-based standards.”
I had speculated in an earlier post (Now We Know) that Director Calabria might not be receptive to suggestions that he change either the structure of the May 20 capital rule or its required capital levels. Two recent developments indicate that this indeed will be the case. First, in testimony before the House Financial Services Committee on September 16 he defended the 4.0 percent minimum capital number, made no mention of any criticisms of it or the risk-based standard, and gratuitously said about the companies’ managements, with no context or elaboration, “Fannie and Freddie have what I would consider some of the worst corporate cultures I’ve ever seen in corporate America.” Then, shortly after this hearing we learned that Calabria had sought, and received, an endorsement of his capital rule from the Financial Stability Oversight Council (FSOC), a group of financial regulators chaired by the Secretary of the Treasury and including three more bank regulators—the Federal Reserve, the Comptroller of the Currency and the FDIC—as well as FHFA, four other regulatory bodies (the SEC, the CFPB, the CFTC and the NCUA) and “an independent member with insurance expertise.”
In a four-page statement issued on September 25, the FSOC said, “The proposed [FHFA capital] rule would require aggregate credit risk capital on mortgage exposures that, as of September 2019, would lead to a substantially lower risk-based capital requirement than the bank capital framework…which would create an advantage that could maintain significant concentration of risk with the Enterprises.” And Calabria himself summarized the FSOC findings (in remarks he gave at the FSOC meeting) by saying, “As the Council found, 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. In other findings and recommendations related to the capital rule, the Council confirms the importance of ensuring that each Enterprise is capitalized to remain a viable going concern both during and after a severe economic downturn. A ‘claims paying capacity’ or similar standard is not appropriate for financial institutions of this size and importance. The Council also affirms the necessity of a dedicated capital buffer that is tailored to mitigate the potential stability risk posed by an Enterprise.”
There is little question that the Director intends the FSOC review to serve as a rebuttal to the major substantive criticisms made of the May 20 standard, and thus a rationale for not changing it (with the exception of credit for CRTs, which he may make more generous). The FSOC endorsement, however, has two disqualifying weaknesses. The first is its source. It is hardly news that a group dominated by bank regulators would prescribe bank-type and bank-level capital for Fannie and Freddie. They have done so for at least three decades, and I strongly suspect that there is little institutional recognition at any of the FSOC-member institutions that the one (weak) rationale that used to exist for applying bank-like capital requirements to the companies—that they, like banks, held large amounts of mortgages in portfolio, funded by debt—no longer is true. And that leads to the second disqualifier of the FSOC statement: it contains almost no documented facts, and literally no risk-related data; its conclusions and recommendations all stem from unsupported assertions and generalities.
Actual mortgage market and credit risk data paint a much different picture of Fannie and Freddie’s potential risks to the financial system than Calabria and the FSOC have put forth. To begin with, they reveal how low single-family residential mortgage credit losses are in a normal environment. Between the time Fannie was spun out of the government in 1968 through the year before the financial crisis, 2007—a near four-decade period that includes six recessions—the highest the company’s single-family credit loss rate (net credit losses as a percentage of total mortgages owned or guaranteed) ever got was 11 basis points, in 1988. And during the fifteen years I was Fannie’s CFO, from 1990 through 2004, its average annual credit loss rate was only 2.5 basis points per year. But then the bottom fell out, for all mortgage lenders, in 2008. What happened?
There have been two periods in the past 100 years when a sharp and protracted decline in U.S. home prices has occurred nationwide. The first was during the Great Depression, when home prices are believed to have fallen by about one-third (reliable data for this period do not exist), and the second was between mid-2006 and mid-2011, when home prices fell by about 25 percent. Each episode was the consequence of unique and identifiable circumstances.
The home price decline during the Depression was triggered by the collapse of the stock market, a national unemployment rate approaching 25 percent, and the failure of many regional banks. This last was problematic because the predominant mortgage at the time was a balloon loan which had to be repaid or refinanced within 3 to 5 years; with so many banks failing there were few lenders able or willing to roll over the maturing loans, leading to widespread defaults. The government responded by creating the FHA, the 30-year fixed-rate fully amortizing mortgage, and Fannie Mae. These innovations kept the mortgage market stable and home prices rising, with no annual declines, for almost 70 years.
Advocates of bank-like capital for Fannie and Freddie’s credit guaranty business pretend not to know what triggered the collapse in home prices in the mid-2000s, but the record is clear. Disastrous decisions by Treasury and the Federal Reserve in the early 2000s first to not regulate subprime lending practices and then to promote the development of a private-label securities (PLS) financing mechanism that put few if any restrictions on the risks of the loans it accepted led to a collapse in underwriting standards and near-unlimited access to mortgages for unqualified borrowers, which in turn fueled an unsustainable boom in home sales, construction and prices. When the PLS market finally collapsed in the fall of 2007—and banks effectively ceased mortgage lending because of soaring delinquencies—housing sales and starts plummeted, and home prices fell by 25 percent peak-to-trough before they could stabilize.
As after the Depression, policymakers responded to the causes of the meltdown. The Fed and Treasury acknowledged that their deregulatory posture was an error. And Congress in 2010 passed the Dodd-Frank Act, requiring lenders to apply an “ability to repay” standard to mortgage borrowers, and through its qualified mortgage standard effectively prohibited the riskiest mortgage products and loan features that proliferated during the PLS bubble. Reflecting these reforms, the credit loss rate on loans Fannie has purchased or guaranteed since 2009, which now make up 95 percent of its current book of business, has averaged 2.7 basis points over the last five years—virtually the same as during the 15 years before PLS became the predominant source of mortgage financing a decade and a half ago.
This quick review of history highlights three important points: first, the fundamental credit quality of the single-family mortgage is very high; second, the stress test used to determine Fannie and Freddie’s new capital requirement is extremely severe, and absent a repeat of the pre-crisis policy mistakes highly unlikely to recur; and third, because of the first two points there is little justification for the amount and type of capital buffers, add-ons and conservatism in the capital rule proposed by FHFA and endorsed by the FSOC, particularly when one considers the loss data and the realities of the companies’ business, as Calabria and the FSOC conspicuously do not.
FHFA’s May 20 capital proposal gives two dollar amounts for Fannie and Freddie’s “net credit losses”—that is, stress test losses after private mortgage insurance, but before any credits from securitized risk transfers—on their September 30, 2019 books of business: $109.1 billion and $134.9 billion. The latter figure, however, incorporates FHFA’s proposal to set a floor of 1.2 percent on credit losses for all loans, and thus is an inflated number. Assuming that $109 billion (or 1.80 percent of adjusted assets) is an accurate estimate of the lifetime credit losses Fannie and Freddie would incur today in response to a 25 percent decline in home prices (and it may still be high), FHFA adds another $124.8 billion through various cushions and buffers to get to the risk-based capital requirement of $233.9 billion (3.85 percent of adjusted assets), then a further $9.0 billion through the 4.0 percent minimum capital requirement, which was binding on September 30, 2019, to reach a total capital requirement of $242.9 billion for the companies.
As numerous commenters on the capital rule pointed out, the $133.8 billion in combined cushions and add-ons to the risk-based requirement in the May 20 rule exceeds by a large margin the $109.1 billion in worst-case credit losses being cushioned or added on to. FHFA and the FSOC claim this is necessary to cover risks other than credit—market, operations, and model or measurement risk—and to enable the companies to survive the stress period as going concerns. But here the disconnect with market reality becomes untenable.
Beginning with the June 2018 version of the Fannie-Freddie capital rule, FHFA consistently has refused to acknowledge that the companies’ guaranty fees constitute revenues capable of absorbing credit losses. While some version of discounting the value of guaranty fees may be reasonable in a stress test conducted on a liquidating book of business (which is how the risk-based capital requirement is derived), ignoring guaranty fees in assessing a company’s ability to survive a stress period as a going concern is nonsensical. Those fees will be there, and moreover many of them (in Fannie’s case, over 40 percent) already have been received in cash, as loan-level price adjustments on higher-risk loans, and literally are present on the balance sheet.
In the second quarter of 2020, Fannie and Freddie’s combined guaranty fees, excluding the TCCA fees payable to Treasury, totaled $7.6 billion, or more than $30 billion annualized. After deducting annualized administrative expenses of $5.4 billion, Fannie and Freddie’s net guaranty fees currently are running at a rate of $25 billion a year. If the $109 billion in combined lifetime stress credit losses FHFA is projecting for Fannie and Freddie follow the same annual pattern as the losses from Fannie’s 2007 book—which is the one subjected to the financial crisis—they would look like this over the first five years: $7.4 billion, $14.5 billion, $24.5 billion (peak in year 3), $19.4 billion, and $14.4 billion. Note that each year’s loss is less than the companies’ current amount of annual net guaranty fees (although year three just barely), and that over the full five-year period the companies would have earned guaranty fees of $125 billion while suffering credit losses from their pre-stress period books of $80.2 billion, leaving their capital accounts not just intact, but increased. (Credit losses at both companies from new business put on during the first five years of the stress period would be only another $8.0 billion if they followed the post-2007 pattern.) That is their business reality.
Contrast this with the unreality of the FHFA capital scheme, endorsed by the FSOC. FHFA projects credit losses of $109 billion from a stress scenario that is highly unlikely to occur, ignores the fact that as the going concerns FHFA (properly) insists the companies be they would be able to cover these stress losses with guaranty fees, and then requires them to protect against “other risks” with a dollar amount of capital, $134 billion, greater than the projected stress credit losses themselves. Yet what, besides going concern risk, might those other unquantified risks be? Unlike commercial banks, Fannie and Freddie do not have the interest rate risk of funding 30-year mortgages with short-term debt; they have virtually no liquidity risk (the threat of deposit flight, or an inability to roll over debt); their market risk is low because their guaranty fees are locked in up front and tend to be recaptured (and can be increased) when mortgages refinance, and they now hold few securitized mortgages in portfolio; their business is not operationally complex and the operations risk they do have is not correlated with credit risk, and finally, in their one line of permitted business their credit losses occur with considerable advance warning, and even in a worst-case scenario are spread out over many years.
When I worked with former Fed Chairman Paul Volcker on the risk-based standard that became the basis for Fannie and Freddie’s 1992 capital legislation, he often expressed his strong view that the capital standard for any regulated financial institution must not go so far in pursuit of a subjective safety and soundness goal that it impeded the ability of a regulated entity to conduct its business on an economic basis. Director Calabria, to my knowledge, has never addressed this balance issue in any form or forum. Instead, for whatever reason he seems determined to subject Fannie and Freddie to a bank-level capital requirement that does not remotely align with the risks of the mortgages they guarantee, despite what he now knows the consequences of this will be: distorted credit pricing, greatly hindered competitiveness, a significant reduction in the volume and breadth of the companies’ business, and quite possibly discouraging investors from supplying the equity required for them to become fully free of the regulatory restrictions imposed upon them more than a dozen years ago.
The fact that Director Calabria has been able to obtain the support of the Financial Stability Oversight Council for his May 20 capital proposal may give him temporary regulatory cover for making only modest adjustments before the rule becomes final, but it does not change the reality that the standard is seriously misguided, and inconsistent with readily available historical and market data. And that inevitably will limit its lifespan. At some point Fannie and Freddie will have a regulator who does not insist that they be arbitrarily and punitively overcapitalized, and understands that hamstringing the operations of two companies at the center of a $10 trillion market critical to the health and growth of the U.S. economy, for no demonstrable reason, is disastrous public policy. For this reason, even if the May 20 FHFA capital standard goes into effect as proposed, it is highly unlikely that Fannie and Freddie will have to recapitalize solely through retained earnings, over a period as long as ten years. They will be given a sensible and workable capital requirement long before then.