Yesterday I submitted my comment on FHFA’s September 16 proposed amendments to its “Enterprise Regulatory Capital Framework” (ERCF) rule. That comment is reproduced below.
On September 16, 2021, the Federal Housing Finance Agency (FHFA) requested comment on a notice of proposed rulemaking “that would amend the Enterprise Regulatory Capital Framework (ERCF) by refining the prescribed leverage buffer amount (PLBA) and credit risk transfer (CRT) securitization framework for [Fannie Mae and Freddie Mac]…and also make technical corrections to various provisions of the ERCF that was published on December 17, 2020.”
The proposed amendments not only ignore but would build on, and enshrine, the glaring inconsistences between the hugely excessive amount of capital required of Fannie and Freddie by the ERCF, the actual risks of the companies’ business as reflected in the results of FHFA’s Dodd-Frank stress tests for 2020 and 2021, and the structure and economics of their current CRT programs as discussed in FHFA’s May 17, 2021 report, “Performance of Fannie Mae’s and Freddie Mac’s Credit Risk Transfer.” If adopted, these amendments would actually reduce the companies’ ability to withstand future credit stresses. FHFA therefore must withdraw them, and instead devote its efforts to bringing Fannie and Freddie’s risk, capital requirements, and credit risk transfer programs into proper economic alignment.
Two publications by FHFA this year—its May CRT performance report and the August 13 release of its 2020 and 2021 Dodd-Frank stress tests on Fannie and Freddie—should have set off alarm bells at the agency that the ERCF’s capital requirements were unreasonably and unjustifiably high, and that former Director Mark Calabria had allowed his ideology to override economics when he replaced FHFA’s June 2018 capital standard with the ERCF.
The June 2018 capital proposal had two main flaws: its risk-based component was tied to current value loan-to-value ratios, which made it procyclical (with capital requirements falling in strong housing markets and rising in weak ones), and it unreasonably assumed that Fannie and Freddie’s guaranty fee income would not offset any credit losses during a period of stress—that is, all stress-period losses had to be covered by initial capital. (It may not have been a coincidence that this latter assumption boosted the companies’ required capital to 3.24 percent of total assets and off-balance sheet guarantees as of September 30, 2017, virtually identical to the 3.25 percent capital percentage proposed by the firm Moelis & Company in its 2016 “Blueprint for Restoring Safety and Soundness to the GSEs,” which was being widely discussed on a bipartisan basis at the time.) Many commenters noted, and criticized, the procyclicality feature and the exclusion of guaranty fees in calculating required stress capital, and urged correction of these flaws after a new Director of FHFA was appointed by President Trump.
That new director was Mark Calabria. When he took office in April of 2019, his views on Fannie and Freddie were well known. In an essay titled “Coming Full Circle on Mortgage Finance,” done for the Urban Institute’s 2016 “Housing Finance Reform Incubator” project (for which I also submitted an essay, “Fixing What Works”), Calabria wrote, “Securitization is a false god that failed us,” conflating the private-label securitization process in which no participant bears any risk of loss—and which was the cause of the 2008 mortgage crisis—with entity-based securitization as done by Fannie and Freddie, who do take risk. Calabria’s prescription for mortgage reform was that, “A more stable and affordable housing market would be best served by returning to an originate-and-hold model of mortgage finance,” and consistent with that objective said, “To retain whatever value there is [in Fannie and Freddie], the current GSE charters should be converted to national bank charters and the GSEs reorganized as bank holding companies (BHCs).” Then, shortly after joining FHFA he said in an interview with Fox Business News, “I think our objective over time is that you have capital levels at Fannie and Freddie that are comparable to other large financial institutions,” adding that 4.5 percent capital was “kind of in the neighborhood of where we’re looking at.”
By the time Calabria put out his initial capital re-proposal for Fannie and Freddie in June of 2020, the actual amount of credit risk at both companies had fallen significantly from where it had been when FHFA’s June 2018 standard was promulgated. One measure of this was the annual Dodd-Frank stress tests run on the companies each year, that replicate the impact of a severe credit shock comparable to the Great Financial Crisis, including an approximate 25 percent decline in home prices. To pass the 2017 stress test, run on year-end 2016 data, Fannie and Freddie had needed capital of 66 basis points of their combined total assets. To pass the 2019 Dodd-Frank stress test run on year-end 2018 data, however, they needed only half that amount of capital—33 basis points of total assets.
Fannie and Freddie’s capital required by FHFA’s June 2018 standard declined significantly over this period as well. When FHFA made its June 2020 capital re-proposal, it revealed that the capital required of the companies by the June 2018 standard of 324 basis points of total assets and off-balance sheet guarantees at September 30, 2017 had fallen to only 225 basis points of “adjusted total assets” (a somewhat larger denominator) at September 30, 2019. This nearly 100 basis-point capital reduction was driven by the same improvements in credit quality as the Dodd-Frank stress tests were reflecting, as well as the procyclical effect of a reduction in the current loan-to-value ratios of the companies’ guaranteed loans during a period of strong home price appreciation.
Calabria, however, wanted Fannie and Freddie’s required capital to be higher, not lower, irrespective of risk. To this end, he added a “prescribed leverage buffer amount” (PLBA) of 1.5 percent to the 2.5 percent minimum capital requirement of “Alternative 1” in the 2018 standard, bringing Fannie and Freddie’s total minimum capital requirement up to the Basel 4.0 percent bank leverage standard (as he had indicated he would). And for the risk-based standard, he made only a technical adjustment to the procyclicality of the 2018 rule, still did not count any guaranty fees as offsets to credit losses, then added enough other buffers, capital minimums and non-risk-based capital charges to raise required capital for his risk-based standard up to 3.85 percent of adjusted total assets (or 4.20 percent of actual total assets). When many commenters said that having minimum capital higher than risk-based capital would encourage excessive risk-taking, Calabria responded not by lowering the minimum percentage but by adding still more conservatism to the risk-based standard, to raise it in the final capital rule, the ERCF, to 4.27 percent of adjusted total assets (and 4.65 percent of actual total assets) as of June 30, 2020.
From 2016 through 2019, FHFA had released the results of its Dodd-Frank stress tests for Fannie and Freddie in August. The results of the 2020 stress test (based on year-end 2019 data) were expected to be released that August as well, during the comment period for the June 2020 capital rule. Calabria did not release them then, or at any other time last year. Instead, FHFA put out a statement saying, “achievement of the purposes of the Safety and Soundness Act will be adversely affected if each Enterprise’s publication of the summary of its Dodd-Frank Act stress test results is not delayed so that each Enterprise may include the alternative [Covid-19] scenarios considered by the Board.” Commenters on the capital rule made their comments without the benefit of the latest Dodd-Frank stress test results.
When FHFA finally did release the 2020 stress test results on August 13, 2021—the same day as the results of the 2021 test (run on the year-end 2020 books) were put out—there was no Covid-related loss scenario, and in the 2020 “severely adverse scenario,” with a 28 percent home price decline, Fannie was able to survive with no initial capital, while Freddie needed just 32 basis points of total assets as capital (combined, they needed 12 basis points of capital). The results of the 2021 stress test were even better: neither company needed any initial capital to survive the 23.5 percent home price decline in this year’s “severely adverse scenario,” and during the stress period they were able together to accumulate and retain earnings equal to 16 basis points of their combined total assets.
No one paying attention, including at FHFA, should have missed the fact that while FHFA’s Dodd-Frank stress tests based on a repeat of the Great Financial Crisis were showing that Fannie and Freddie had gone from needing 66 basis points of capital to survive their stress test to generating 16 basis points of retained earnings as it unfolded, Director Calabria had been using a host of cushions, buffers and add-ons to set a “risk-based” capital requirement for the companies of more than 460 basis points—double the capital required by the 2018 FHFA rule—to survive essentially the same scenario.
This disconnect between the reality of Fannie and Freddie’s actual creditworthiness and their assumed, but fictious, need to cover more than 400 basis points of credit losses in a severe stress scenario was inescapable when FHFA published its May 2021 performance report evaluating the companies’ credit-risk transfer programs. In it, FHFA said it had asked a consulting firm, Milliman, to simulate the performance of the companies’ CRTs on $126 billion of risk in force as of April 30, 2021 under two sets of conditions, a “Baseline scenario” and a “2007 Replay” intended to mimic the credit stress experienced during the Great Financial Crisis (and the Dodd-Frank stress tests). Milliman found that in the baseline scenario Fannie and Freddie’s lifetime CRT costs were $33.60 billion and their “ultimate benefits,” or credit loss reimbursements, were $1.06 billion, for a net CRT cost of $32.55 billion. And in the 2007 Replay, Milliman projected lifetime CRT costs of $30.72 billion, ultimate benefits of $10.10 billion, and a net CRT cost of $20.63 billion.
FHFA gave the results of the Milliman CRT performance simulations without comment or conclusions; instead, it simply said, “FHFA continues to assess the CRT programs, including their costs and benefits as well as the benefits and risks to the safety and soundness of the Enterprises, the Enterprises’ ability to perform their statutory mission, and the liquidity, efficiency, competitiveness and resiliency of the national housing finance markets.” Yet the problem FHFA dodged in its CRT report is obvious. The reason that Fannie and Freddie will make (according to Milliman) 30 dollars in CRT interest payments for every 1 dollar of credit loss transferred in a normal environment, and pay 3 dollars in interest for every 1 dollar in credit losses transferred even in an environment of extreme credit stress, is that the companies’ CRT programs are calibrated to wildly overstated levels of potential credit loss, and have been since their inception. The large majority of the CRTs they issue are pure giveaways to the investment community.
And FHFA knows this, at least at the staff level. In its June 2018 capital proposal, FHFA said that the credit loss rate of Fannie’s 2007 book of business through September 30, 2017 “using current acquisition criteria”—that is, without the Alt A loans, interest-only ARMs and risk layering that resulted in over half of that book’s losses—would have been only 1.5 percent. Fannie and Freddie can cover a 9-year cumulative loss rate of 1.5 percent with the income from their current average annual guaranty fee (net of administrative expenses) of 36 basis points, as evidenced by the most recent results of their Dodd-Frank stress tests. And with a 9-year cumulative stress loss rate for the companies of 1.5 percent, the Milliman CRT performance results make perfect sense. Typically, Fannie and Freddie’s CRTs do not transfer any losses before they exceed 50 basis points of a covered pool of loans, and they continue to provide coverage up to 400 basis points or more. With the expected loss rates of Fannie and Freddie’s post-2007 loans in the range of 2 to 5 basis points per year, only a very small portion of covered pools in a “baseline scenario” will have credit losses in excess of 50 basis points while the CRTs issued against them remain outstanding (as they can, and do, prepay). And even in a repeat of the Great Financial Crisis, only the bottom third of the CRT coverage range of 0.5 percent to 4.0 percent (or more) of a pool balance has any risk of experiencing credit losses.
Once these actual data, from FHFA, are introduced into the analysis, it becomes obvious why the agency’s September 16 ERCF capital amendments (and “technical corrections”) are such a bad idea. They use the lure of a reduction in capital requirements—more risk-based CRT credit, and a reduction in the PLBA—from levels that are indefensibly high, and based on wholly fictitious notions of Fannie and Freddie’s credit risk, to effectively penalize the companies for not issuing CRTs that are virtually certain to lose them tremendous amounts of money under any set of circumstances, thus greatly reducing their ability to handle the credit stress they may one day face in reality. This is the opposite of FHFA’s professed goal.
Because FHFA’s September 16 amendments would weaken the companies, they must be withdrawn. But that will not be sufficient; the disconnect between the ERCF, the results of the annual Dodd-Frank stress tests run on Fannie and Freddie, and the economics of their credit risk transfer problems will persist until FHFA acts to fix it. And it is clear what needs to be done. The 1.5 percent stress loss rate for Fannie and Freddie’s 2007 book of business “using current acquisition criteria” through September 2017, the Dodd-Frank “severely adverse scenario” stress test results for 2020 and 2021, and the Milliman performance simulations of the companies’ April 2021 CRT books all are based on real data. Calabria’s ERCF is not.
In fact, since the beginning of the conservatorships, proposals for Fannie and Freddie’s capital have never been linked to their risk; they have been driven by the intent of the companies’ critics and competitors to use overcapitalization in the name of safety and soundness to push their guaranty fees to noneconomic levels, and drive business to “free market” alternatives. In 2013, for example, FHFA Acting Director Ed DeMarco required Fannie and Freddie to raise their guaranty fees by 10 basis points not because of risk but to “encourage more private sector participation” and to “reduce [their] market share.” And as recently as April of 2014, the Johnson-Crapo bill from the Senate Banking Committee would have required the credit guarantors who were to replace Fannie and Freddie to hold 10 percent capital to back their credit guarantees—with no reference at all to risk, other than to say that the 10 percent capital amount could be reduced if the guarantors transferred it. The ERCF is only the latest example of a non-risk-based approach to Fannie and Freddie’s capital, but it is the one that currently is binding on them, so it is the one that FHFA needs to repeal and redo.
The persistent and deliberate overcapitalization of Fannie and Freddie has had several negative, but predictable, consequences. Most obviously, you have two companies who today have extremely high-quality books of business and earn some $20 billion per year after-tax, but have no hope of exiting conservatorship in the foreseeable future because they have a core capital shortfall to the grossly inflated levels required by the ERCF of nearly half a trillion dollars, and no access to the capital markets because Treasury and FHFA have elected not to cancel the net worth sweep, which was imposed before the two agencies realized that the correct resolution of the companies’ indeterminate limbo was not to replace them, but to recapitalize them.
Second, Fannie and Freddie’s guaranty fees since the conservatorships have risen by over 20 basis points, and could rise dramatically further if the ERCF remains in place. In order to earn a modest after-tax return of 9.0 percent on 465 basis points of capital, the companies would need to charge an average of 65 basis points on their new credit guarantees, another 21 basis points more than their average gross fee (net of TCCA) in 2020 of 44 basis points. This is a ticking time bomb that everyone would prefer to think does not exist. And even the current level of Fannie and Freddie’s guaranty fees has had a profound effect on their ability to do affordable housing business. In 2007, 36 percent of the loans they purchased or guaranteed had credit scores less than 700; in 2020, just 12 percent of their combined business had credit scores that low.
Finally, and not surprisingly, banks’ holdings, and share, of 1-4 family first mortgages and mortgage-backed securities (MBS) have soared since the conservatorships. At December 31, 2007, banks held $2.23 trillion in 1-4 family first mortgages and MBS, for a 22.2 percent share of that $10.04 trillion market. Outstanding 1-4 family first mortgages and MBS were 15.7 percent higher at June 30, 2021, at $11.62 trillion, but banks’ holdings of them then were nearly double, at $4.42 trillion, for a 38.0 percent market share. This may have been good for the banks—and what they wanted to have happen—but shifting these volumes of mortgage holdings from capital markets investors such as pension funds and life insurance companies to leveraged commercial banks, who are funding them with consumer deposits and short-term purchased funds at a time of record low interest rates, increases systemic risk markedly.
None of these effects are ones senior economic officials in the Biden administration, when they focus on them, will support, or wish to have continue. The change of administration thus puts FHFA in an excellent position to take the lead in breaking free of the misguided, fiction-based policies of previous administrations towards Fannie and Freddie, and shifting to policies based on fact. FHFA must be bold in making this change, and not ignore the need for it or pretend it isn’t necessary, as the September 16 proposed capital amendments do.
And the required changes are straightforward. First, FHFA and Treasury must agree to declare that Fannie and Freddie have paid back all of the $187 billion they were forced to draw during the financial 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. Then, FHFA must replace Calabria’s ERCF with a rule based on the companies’ actual business and credit risks. As I discuss in “Capital Fact and Fiction” on Howard on Mortgage Finance, 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 aligned with the risk-based capital requirement.
Only when the ERCF has been replaced should FHFA turn to the task of determining how much capital credit to give to Fannie and Freddie’s (redesigned and recalibrated) credit risk transfers. Changing the CRT credit before then would be a waste of FHFA’s time, and worse, result in a great waste of the companies’ money.