Tucked away in Fannie Mae’s third quarter 2023 10Q were three factoids about what has been happening with its credit-risk transfer (CRT) programs since short- and long-term interest rates began increasing in the spring of 2022. The first was Fannie noting, “For our credit risk transfer transactions executed during the first nine months of 2023, a weaker credit profile of the reference pools, the current higher interest rate environment, and investor expectations regarding a slowdown in home price appreciation have generally resulted in either increased premium costs or the retention by Fannie Mae of a higher first loss position compared with similar transactions in the first nine months of 2022.” (It had earlier said, in its 2022 10K, that “Taking into account the increasing cost of these [CRT] transactions, the resulting capital relief, and the credit risk transfer market capacity, we have chosen to increase the first loss position retained by Fannie Mae compared with prior transactions.”) Second, Fannie revealed that through September 30, 2023, the annual cost of its CRT programs had reached an all-time high of nearly $1.5 billion, while the lifetime expected value of its “freestanding credit enhancement receivables” had fallen to just $265 million, compared with $565 million at the end of 2022. And finally, despite a $4.7 billion increase in net worth in the third quarter, Fannie only was able to trim its capital deficit by $1.8 billion, because even with its total assets rising just 16 basis points its risk-weighted assets rose by 2.3 percent, likely caused by diminished capital credit given to its CRTs, due to the company’s taking more risk on recent issuances.

I have been a long-time critic of Fannie and Freddie’s CRT programs, which began with a directive from the Federal Housing Finance Agency (FHFA) in its 2012 Strategic Plan for Enterprise Conservatorships to “[shift] mortgage credit risk from the Enterprises (and, thereby, taxpayers) to private investors,” and since that time has led FHFA to give annual targets or directives to both Fannie and Freddie for credit-risk transfer securities (CAS and STACRs, respectively) issuance, and insurance/reinsurance (CIRT and ACIS) transactions. The essence of my criticisms of these programs—particularly CAS and STACRs—is that the companies pay far too much to insure high-quality books of business against losses that have only a remote chance of occurring, and that investors will not buy new CRT securities during periods of financial stress, when they are most likely to be of value to the issuer.   

I thus was very pleased, and surprised, when in May of 2021 FHFA itself released a report titled Performance of Fannie Mae’s and Freddie Mac’s Credit Risk Transfer that not only agreed with my observations and conclusions but also offered data in support of them. The report first gave the economics of the companies’ CRT programs to date: “As of February 2021, the Enterprises had paid approximately $15.0 billion in interest and premiums to CRT investors and counterparties and the Enterprises had received approximately $0.05 billion via investor write-downs and counterparty reimbursements.” (That’s $1 of benefits for every $300 in premiums.) But it was the simulations of future CRT performance done by FHFA—using the residential mortgage model of a consulting firm, Milliman—that were the real eye-openers. FHFA ran both a “Baseline” and a “2007 Replay” scenario. In the Baseline scenario, Fannie and Freddie’s lifetime CRT costs were $33.60 billion and their “ultimate benefits” were $1.06 billion ($1 of benefits for every $32 in premiums), while in the 2007 Replay the lifetime CRT costs were $30.72 billion and the ultimate benefits were $10.10 billion ($1 in benefits for every $3 in premiums).

Those results weren’t just non-economic, as I had been saying, they were staggeringly non-economic. And I kept waiting to see what FHFA would say about them. But all it said was, “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.” That made no sense. What could be left to study? In a blog post about this (FHFA’s CRT Report) I wrote, “This encapsulates what’s so wrong with the Calabria-led Federal Housing Finance Agency: they tout themselves as a ‘world-class regulator,’ yet they seem to be operating in a world divorced from reality. Calabria and others at FHFA refer to CRTs as a triumph of ‘bringing private capital into the mortgage market,’ but their staff has put out a report showing the truth to be just the opposite: by requiring huge amounts of annual interest payments while providing minimal absorption of credit losses in return, CRTs siphon tremendous amounts of capital out of the mortgage market, and weaken the companies FHFA regulates.”

We now know from former director Calabria’s book, “Shelter from the Storm,” that he did understand that Fannie and Freddie’s CRTs were huge wastes of money; but instead of eliminating the FHFA directives and capital incentives to the companies to issue them, he blamed Wall Street. He said this in his book:

“Once I had established an independent research and evaluation function within the FHFA, I asked that team to evaluate CRT performance…. Unfortunately, that analysis was not completed until after the capital rule was finalized. I was stunned by the results. As the housing market had been mostly strong during the use of CRTs, I certainly expected to see short-term costs exceed short-term benefits. But I was not expecting the efforts to have a net cost of around $15 billion. That is real money. The analysis included stress test results as well. Most shocking was that if we again experienced a 2008 scenario, the net costs would go up, not down. [Note: that’s actually not true.]

The clearest way to think about CRT is as insurance. You pay premiums, and if something bad happens, the insurance covers the cost. Yet the way that CRT worked was that you paid massive insurance premiums and then when something bad happened, your coverage would immediately expire [note: that’s also not true] and you would get almost nothing back….It slowly dawned on me that the CRTs were structured to almost never pay.

In the late spring of 2021, after it became apparent that the companies had essentially been looted, we started to investigate the details of how CRT transactions were underwritten. Our first shock was learning that the Wall Street underwriters earned fees that were multiples of the net values of CRTs to the companies. These Wall Street underwriters—you can guess the firms—were supposed to protect Fannie and Freddie, but instead they joined in robbing their clients. Of course, the CRT investors, primarily hedge funds and real estate investment trusts, were also clients of these Wall Street firms.

We had just begun putting the pieces together when the Collins decision…was made and my tenure came to an end. Given the connections between the Biden administration and the Wall Street players profiting from CRT, I suspect that any investigation has come to an end. In fact, one probable reason the Biden administration wanted me out so quickly was to end it.”

Calabria’s self-depiction as the thwarted potential savior of Fannie and Freddie from the evil CRTs devised by Wall Street does not withstand even the most superficial scrutiny. (A more likely explanation is that, as a longstanding ideological opponent of the companies, he was perfectly content to leave the money-losing CRT programs in place, while blaming Wall Street and a Democratic administration for not ending them.) And in any event, the initial visible responses of FHFA to its May 2021 CRT report were to (a) increase (not reduce or eliminate) the capital credit given to CRTs in the revisions to the Enterprise Regulatory Capital Framework proposed that September and made final later, and (b) change the CRT language in the companies’ 2020 and 2021 Scorecards saying “Continue to transfer a significant amount of credit risk to private markets in a commercially reasonable and safe and sound manner” to “Transfer a significant amount of risk to private investors, reducing risk to taxpayers” in their 2022 Scorecard. (The deleted clause “in a commercially reasonable and safe and sound manner” was restored in the 2023 Scorecard, perhaps because no one had noticed its absence in the Scorecard for the previous year.)

Short- and long-term U.S. interest rates began to rise sharply in the spring of 2022, about a year after the work on FHFA’s CRT report—which covered the period between Freddie’s and Fannie’s first CRT issues (in July 2012 and October 2013, respectively) and February 2021—had been completed. And since that time, the issuance environment for CRTs has deteriorated dramatically. So as bad as the economics of the companies’ CRTs issued from 2012 (or 2013) through 2020 were, the economics of the CRTs issued from 2022 to now, and for the foreseeable future, almost certainly will be worse. FHFA has been silent on this, and Fannie only hints at it, but the reasons are straightforward, and are manifested in the changes to the structures and costs of the CAS issued by Fannie over the past two years.

Before getting into the details of some of these securities, it’s helpful to lay out a framework for the analysis. Fannie’s CAS and Freddie’s STACRs are structured as senior-subordinated securities. After (usually but not always) an initial amount of credit losses absorbed by the issuer, at some level of credit loss rate, say 50 basis points, the owner of the most junior CAS or STACR tranche begins to absorb the issuer’s credit losses, up to a “release point,” say 140 basis points. Then, the owner of the next most junior tranche begins to absorb losses, up to its release point, say 250 basis points. A CAS or STACR structure typically will have one or two additional loss-absorbing tranches, covering losses up to anywhere between 350 and 550 basis points, at which point any remaining losses fall upon the issuer again.

A second important feature of CAS and STACRs is that each tranche is priced at a spread over 1-month SOFR (the Secured Overnight Financing Rate, formerly LIBOR, the London Interbank Offered Rate). SOFR is a floating rate that moves closely with the federal funds rate, and this always has been a weakness of the CAS and STACR structures. Just as funding a portfolio of 30-year fixed-rate mortgages with short-term debt will lead to narrowing margins or losses when short rates rise, buying credit insurance on pools of mortgages that have fixed guaranty fees using floating-rate CRTs will make the cost of that insurance go up when short rates do, with the CRT issuer having no way to offset that extra cost with higher revenues. Up until the spring of 2022 this unfortunate feature was muted by the fact that 1-month LIBOR or SOFR averaged less than 100 basis points during the 2012-2021 period; it was under 20 basis points during five of those years, and only exceeded 200 basis points in one year, 2019 (topping out at just over 250 basis points). Recently, though, the story has been different. One-month SOFR crossed the 300 basis-point threshold in October of 2022, rose above 400 basis points that December, and now stands at 5.35 percent. The forward yield curve for 1-month SOFR has it staying above 3.00 percent for the next several years.

A much higher SOFR is only one of the challenges facing CRT issuers. Higher mortgage rates also have reduced activity in the housing market, caused slower home price growth, and led CRT investors to require higher spreads over SOFR on the CRTs they buy. With higher SOFR base rates and increased spreads over SOFR required by CRT investors, the only way Fannie (or Freddie) has to keep any control over the cost of its CRTs is to insure a lesser percentage of its credit risk, and that’s what it has been doing.

There is no such thing as a standard CRT structure, and the interactions of their component parts are very complex. But there is a fairly reliable metric that gauges the efficiency of Fannie’s different CAS issuances, and a combination of three good metrics that track their costs. The efficiency metric is the percentage of the first 250 basis points of credit loss transferred to CRT buyers. (The reason for the cut-off at 250 basis points is that nearly all CAS issues transfer at least that percentage of credit losses, and losses in excess of it are increasingly unlikely.) The three cost metrics are: (a) 1-month SOFR at the time of pricing, (b) the weighted average spread over SOFR of all CAS tranches (the amount of each sold tranche times its spread to SOFR, divided by the total dollar amount of the sold tranches) at the time of pricing, and (c) the initial cost of the CAS in basis points (SOFR plus the weighted average spread times the dollar amount of the sold tranches, divided by the size of the mortgage pool these CAS issues are insuring). None of these measurements are perfect. The efficiency metric ignores where within the 250 basis-point range the losses are transferred, while all three cost metrics can, and do, change over time (including the weighted average spread over SOFR, which rises as the more senior, lower-spread tranches pay off). But the metrics do offer insight into what has been happening with Fannie’s recent CAS issues.

Prior to the pandemic, Fannie consistently had been able to cover 76 to 86 percent of its first 250 basis points of credit losses on CAS-insured pools by limiting its first-loss position to between 25 and 50 basis points, and retaining only 5 percent of the tranches above that level, on which it paid an average tranche spread to SOFR that rarely exceeded 250 basis points. When the company resumed CAS issuance in the fall of 2021, it stayed with this low first-loss retention (at that time 25 basis points), and with SOFR at just 5 basis points and the weighted average CAS tranche spread staying under 300 basis points, Fannie’s initial pool cost on its first four post-pandemic CAS issues was less than 8 basis points. Short- and long-term interest rates began rising in March of 2022. Fannie stayed with the same basic structure for the CAS it issued on April 5, 2022, but paid an average spread of 413 basis points over SOFR (at 29 basis points), for an average pool insurance cost of 14 basis points. Then, as interest rates continued to rise, the CAS Fannie issued on June 28 with only 30 basis points of initial loss retention required a weighted average tranche spread of 502 basis points, and with SOFR at 1.52 percent the company had to retain half of its two most junior CAS tranches (on which it paid 1200 and 680 basis points over SOFR, respectively) to hold its initial pool insurance cost at 28 basis points.

With an average net guaranty fee (total guaranty fee less 10 basis points paid to Treasury and 8 basis points of administrative expense) of 44 basis points on new business, Fannie could not continue to pay 28 basis points for CRTs that, per FHFA’s own admission, provide very little credit loss protection during periods of stress. Something had to give, and what gave was the amount of first-loss exposure Fannie kept. In the CAS it issued on September 21, 2022, Fannie created two tranches it retained that took the first 125 basis points of credit losses, and it also retained 75 percent of the next most junior tranche, which took the losses between 125 and 190 basis points. Keeping this much of the credit risk allowed Fannie to lower the initial pool cost of that CAS to 12 basis points, but at the expense of only being able to able to transfer 29 percent of its credit losses up to 250 basis points—with none being transferred until Fannie had absorbed the first 125 basis points of those losses. 

SOFR continued to rise through the summer of 2023, and as it did Fannie’s CAS issuance became an exercise in tailoring the amount of the first 250 basis points of loss exposure it retained in an effort to keep the initial cost of each issue around or below 30 basis points of the total pool size. In its final two CAS issues of last year, Fannie retained 150 or 155 basis points of first-loss exposure, then an average of half the exposure in the next tranche, which covered losses to over 250 basis points. With those issues, the CAS program became the numerical opposite of what it had been before the pandemic (and FHFA’s 2021 CRT report): instead of transferring 76 to 86 percent of its first 250 basis points of credit losses to CAS investors, Fannie was retaining 76 to 86 percent of those losses, while paying an average of 21 basis points for the remaining (miniscule) loss protection CAS investors were willing to provide at that price.

Fannie’s CAS program has never been economic, but it’s now a charade. The company seems to have set an average initial pool cost it will accept (although the cost of individual CAS issues can vary considerably), and it keeps as much of a mortgage pool’s credit risk as it has to in order to sell CAS at that price, with the percentage of transferred credit losses being the variable. There is not even a pretense of these issues being effective or efficient.   

This, of course, is not Fannie’s fault; it is another stupefying consequence of the company’s conservatorship under FHFA. The former director of FHFA, Mark Calabria, has admitted that CAS are a “looting” of Fannie, yet knowing this, the current director, Sandra Thompson, continues to use both carrots (capital credits) and sticks (CRT issuance goals that reduce executive compensation if not met) to encourage CAS and STACR issuance, even as the costs of these deals soar and their benefits dwindle, making the looting even greater.

The only explanation for FHFA’s passivity in the face of the CAS charade is the same as for its passivity in addressing the inconsistency between the results of its recent Dodd-Frank stress tests and its regulatory requirement that Fannie and Freddie hold 3.0 percent capital before they can be released from conservatorship: the current FHFA leadership will not make any significant change to the status quo for the companies—no matter how obvious, urgent or desirable—unless or until it is directed to do so by some other institution or individual, whether Congress, Treasury, or a very senior economic or policy official in the administration. In the meantime, the losses from their CAS and STACR programs will slow the growth of Fannie and Freddie’s retained earnings, while the reduced capital credit given to their much less efficient new CRT issues will be a headwind against filling their capital gaps, by causing their risk-weighted assets to rise at a faster rate than their total assets.