On June 12 the Federal Housing Finance Agency (FHFA) issued a notice of proposed rulemaking on “Enterprise [Fannie Mae and Freddie Mac] Capital Requirements.” The proposal is lengthy—263 pages, plus another 105 pages of proposed rule text. Because comments on it are not due until 60 days after it is published in the Federal Register (which as of yet it has not been), I did not attempt to work my way though it before leaving on vacation last month. Now that I’ve been able to finish reading and analyzing it, however, I feel compelled to make some initial comments about it, in advance of the formal comment I will send to FHFA later on in the summer.
It perhaps should not have surprised me, but I found FHFA’s risk-based capital proposal for single-family mortgages to have a pronounced, troubling and entirely avoidable pro-Financial Establishment and anti-homebuyer bias. As I discuss below, the biases fall into three categories: inappropriately basing the risk-based standard on banks’ Basel standard; excessive and counterproductive conservatism in the risk-based standard, and flaws in the 2.5 percent minimum capital alternative and the treatment of credit risk transfers. Taken together, these make it difficult to avoid the conclusion that FHFA backed into its capital numbers, and engineered its proposal to produce what Fannie and Freddie competitors and critics (including careerists at Treasury) have long been asking for: requiring the companies to adhere to capital standards dictated by and advantaging large banks and Wall Street firms, and disadvantaging homebuyers.
Inappropriately basing the risk-based standard on banks’ Basel standard
Despite the fact that banks’ Basel capital standards are designed to cover a very wide range of asset categories that Fannie and Freddie are not permitted to deal in, and that have much higher and less predictable loss rates than home mortgages, “FHFA’s rule is based on a capital framework that is generally consistent with the regulatory capital framework for large banks.” FHFA claims that its proposed capital rule for Fannie and Freddie “appropriately differentiates from other capital requirements based on the actual risks associated with the Enterprises’ businesses,” but even a cursory analysis reveals it comes nowhere close to doing that.
The one place in the proposal where FHFA attempts to numerically “recognize the lower risk” of Fannie and Freddie’s business relative to banks is in its determination of a minimum capital ratio for the companies. And there, instead of using actual loss data, FHFA states bewilderingly, “Risk is defined using Basel risk weights.” In Basel, mortgages get a 50 percent risk weight, while FHFA calculates that the average risk weight of the top 34 bank holding companies (who are subject to enhanced capital standards) is 72 percent. Using these figures, FHFA asserts, “This suggests that the risk-weighted asset density for [Fannie and Freddie’s] assets is about 69 percent (calculated as 50 percent divided by 72 percent) of the risk-weighted asset density for the largest bank holding companies.” Put differently, FHFA is claiming, based on the Basel risk weights, that banks’ assets are only one and a half times as risky as the mortgages owned or guaranteed by Fannie and Freddie.
That is preposterous. There is no justification for FHFA to use Basel risk weights to assess the relative riskiness of Fannie and Freddie’s assets when historical loss data are readily available for the same purpose. And the differences using real data are staggering. In the 20 years before the 2008 mortgage crisis, credit losses at FDIC-insured banks averaged 82 basis points of total assets per year, twenty times the 4 basis point average credit loss rate at Fannie and Freddie. The 2008 mortgage crisis was a true “stress event” that affected losses at Fannie and Freddie (whose assets are limited to mortgages) disproportionately to banks, but even including post-2008 credit losses in the comparison Fannie and Freddie’s 1988-2017 average credit loss rate of 14 basis points is just one-sixth the 84 basis point loss rate of the banks over the same period. And if you adjust Fannie and Freddie’s 2008-2017 losses to exclude the loan products and risk features they no longer are allowed to finance (as even FHFA does in its analyses), the companies’ 1988-2017 average loss rate falls to 8 basis points—one-tenth the comparable average loss rate of commercial banks.
As of the end of 2017, the banking industry as a whole, with $16.2 trillion in assets, had a total equity-to-assets ratio of 11.2 percent. (The four largest banks, with over half of banking industry assets at $8.2 trillion, had average equity capital of 10.2 percent.) The banking industry’s average capital percentage is about 13 times its average annual credit loss rate. It is axiomatic in finance that capital must be related to risk. If FHFA were to use the same 13:1 ratio of capital to annual credit losses for Fannie and Freddie as exists for the banks—which would overstate Fannie and Freddie’s required capital because these two companies take nowhere near as much interest rate risk as banks do—a true “bank-like” capital ratio for them would be between 100 and 180 basis points, with a figure at the lower end of this range being more consistent with the residential mortgage types Dodd-Frank permits the companies to finance today.
It is understandable that competitors and critics of Fannie and Freddie would want to ignore actual comparable loss experience and insist that the companies’ capital be based on the Basel bank standards, but it is not acceptable for Fannie and Freddie’s safety and soundness regulator to do the same thing. And grossly overstating the risk of the companies’ business leads FHFA to make its next major mistake—adding excessive, redundant and unjustified conservatism to its risk-based standard in order to produce a capital number that approaches the simplistic 4.0 percent risk-weighted leverage requirement applied by Basel to bank mortgage holdings.
Excessive and counterproductive conservatism in the risk-based standard
For Fannie and Freddie’s single-family credit guaranty business (the main focus of this comment), there is a simple, effective, and nondistortive way to implement the risk-based capital directive in the Housing and Economic Recovery Act (HERA). FHFA would take the companies’ single-family books at the end of each quarter, and apply the stress loss rates it developed for the various types of loans and risks (in the proposal these are done to a high degree of granularity) over a defined and realistic period of time, and incorporate projected income and expenses to determine the amount of initial capital the companies would need to survive that stress. The minimum capital percentage, and not the risk-based standard, would incorporate any capital cushions deemed appropriate, taking into account the fact that HERA gives the regulator authority to take prompt corrective action when capital falls below either the minimum or the risk-based amounts.
FHFA does determine stress loss rates by loan and risk category, but it then adds numerous and redundant elements of conservatism, both when applying the stress test and in addition to it. These include: (a) declining to count guaranty fee income as an offset to loan losses, (b) not allowing for the tax deductibility of those losses, (c) adding a fixed “going concern” buffer to all assets, irrespective of their risk, and (d) adding a reserve for deferred tax assets, or DTAs, despite the fact that the going concern buffer is designed to ensure the companies’ continued solvency, so that their DTAs always would have value. (FHFA also adds a “market risk” capital charge, but this applies only to the portfolio business.) In a fact sheet FHFA produced that shows the “impact of the proposed capital rule,” the refusal to count income as an offset to credit losses adds an estimated 113 basis points to the risk-based standard for single-family credit guarantees, while the going concern buffer adds another 75 basis points to it.
FHFA’s rationales for not counting revenues in its stress test are exceedingly weak. Its first is that the Basel capital standards don’t count them. That’s true, because Basel uses simple ratios, not a stress test. The Dodd-Frank stress test for banks does count revenues, as any legitimate stress test does and should. A second reason FHFA gives is that, “Inclusion of revenues could result in a very low or zero-risk based capital requirements for specific portfolio segments.” That’s also true, if the loans aren’t very risky. But that’s precisely why FHFA includes a minimum capital level, which kicks in if the aggregate risk-based capital numbers drop below the minimum.
The impact of not counting revenues is huge. Without revenues, FHFA calculates that stress credit losses for Fannie and Freddie’s September 31, 2017 book would be $112.0 billion, or 201 basis points of their combined assets and off-balance sheet guarantees. But with the companies’ current average net guaranty fee of 30 basis points (deducting both administrative expenses and the payroll tax fee), and the stress prepayment rates from the 2008-2012 period, guaranty fee income from that same book would be an estimated $63 billion, leaving credit losses net of revenues of $49 billion, or 88 basis points. That’s the accurate risk-based capital percentage.
There clearly is merit to the concept of a going concern buffer, but FHFA errs badly in adding it to the risk-based standard; it must instead be a component of the minimum, due to the interaction of a risk-based standard that can change over time and the provision in HERA for prompt corrective action by the regulator if Fannie and Freddie do not meet both their risk-based and minimum capital standards.
Fannie and Freddie’s minimum capital standard will be a fixed percentage (FHFA’s two proposed alternatives are discussed briefly below), while the risk-based number will change counter-cyclically—requiring less capital when home prices are rising and more when they are falling. Fannie and Freddie executives know that when home prices are falling it is difficult, expensive and sometimes not possible to raise capital. For that reason they almost certainly will want to hold significantly more capital during good times than is required by their risk-based standard, because when that risk-based requirement rises during periods of stress (and it can rise quickly and considerably) they will want to continue to be in compliance with it to avoid adverse regulatory action, including restrictions on their business.
In the real world in which the companies operate, therefore, FHFA’s proposal to put multiple layers of conservatism inside the risk-based standard is highly problematic. Even though this conservatism is intended as a cushion, in reality it won’t serve as one, because the extra capital can never be drawn down—doing so would push a company under its risk-based requirement (which has the “cushions” built into it) and trigger the prompt corrective action the companies so desperately will be trying to avoid. This means that the conservative set of elements in the FHFA proposal that produced a 3.24 percent risk-based capital requirement at September 31, 2017—a time when the housing environment was very favorable—would in fact lead Fannie and Freddie to hold at least 4 percent and more likely closer to 5 percent capital, including the excess they know they will need when the cycle turns less favorable, and their (overly conservative) risk-based requirement becomes much higher.
FHFA needs to understand that if it takes a true risk-based capital number of 88 basis points, adds unnecessary cushions, and then ignores the impact of its prompt corrective action disciplines on management’s incentive to hold excess capital, the result will be actual capital holdings that cause extreme distortions in the pricing of single-family credit guarantees. Having to price mortgages that have an 8 basis point expected loss rate to earn a market return on 5 percentage points of capital will stretch the affordability of and access to credit guarantees from Fannie and Freddie beyond the breaking point. And there is no reason for it. Fannie and Freddie are not banks, and FHFA should not be trying to smother them with Basel-like required capital, unrelated to the risks of their business. Instead it should subject them to a straightforward and realistic risk-based standard, with no fudges or cushions, then add a minimum standard calibrated to the statistical risk of their business, and that incorporates a going concern buffer sized to reflect the reality that if they fall below this minimum HERA gives the FHFA director discretion to classify them as “critically undercapitalized,” and appoint FHFA as their conservator or receiver.
Flaws in minimum capital and credit risk transfer proposals
Two other aspects of the FHFA capital proposal deserve brief mention here: its alternative minimum standards and its treatment of credit risk transfers.
FHFA suggests two potential approaches for a minimum capital (or leverage) requirement: a “2.5 percent alternative,” which is 2.5 percent of all on-balance sheet assets and off-balance sheet guarantees, and what it calls a “bifurcated alternative,” which effectively is 1.5 percent of all credit guarantees and 4.0 percent of other on-or off-balance sheet items.
Between the two, it is an easy choice. FHFA derived its 2.5 percent alternative by using Basel risk weights to translate bank risk or solvency measures into Fannie and Freddie capital, thus making the egregious mistake noted earlier of ignoring actual historical risk data for Fannie, Freddie and the banks in favor of the inapplicable, imprecise and wildly inaccurate Basel risk-weight comparison. That leaves the bifurcated alternative as the only reasonable one. FHFA does not make a convincing case for why it picked the 1.5 percent and 4.0 percent numbers for this standard, but with the important caveat that these figures be viewed as including a going concern buffer, I would view them as generally acceptable. I plan to do additional work on this topic, however, and to address it in more detail in my comment letter.
FHFA makes great efforts to favor credit risk transfers (CRTs) in its risk-based capital standard. One of these is ironic but harmless: having relied heavily on Basel in deriving credit risk capital, FHFA ignores Basel in giving capital credit to CRTs (Basel does not do so). A second act of favoritism is much more serious. Assessing the worth of CRTs as capital substitutes requires measuring their expected benefits against their costs. In the risk-based standard, FHFA gives credit for the expected benefits of CRTs but does not take their cost into account at all—because it ignores guaranty fee income and expenses, including CRT interest payments. (I’m tempted to think that FHFA’s odd decision to ignore income and expense in the risk-based standard may have had its origin in a desire to create a regulatory incentive for the CRT securities it and Treasury have been forcing Fannie and Freddie to issue). No one disputes that CRT capital is greatly inferior to equity capital, yet the FHFA rule inexplicably and dangerously gives the companies a one-sided capital incentive to substitute the former for the latter. If the companies fall prey to this lure (and they may not, if they look at the economics of the transactions) it will make them riskier, so in the interest of safety and soundness FHFA’s CRT treatment must be changed.
* * *
These are my most important observations on the FHFA capital proposal. I believe it is critical that individuals and groups who have an interest in and a commitment to a mortgage finance system that works well for homebuyers read the FHFA proposal, make sure they understand it, and then give FHFA their comments on it. There can be no doubt that the supporters of large banks and Wall Street interests will flood FHFA with comments praising the wisdom and fairness of its proposal, so it will up to advocates for the ordinary consumer—the affordable housing groups, community banks and others—to point out the flaws and failings of FHFA’s capital plan, and to offer advice as to how it can be fixed in a way that reverses its pro-bank and Wall Street bias, which if not undone will deal a severe blow to the low- moderate- and middle-income homebuyers Fannie and Freddie were chartered to serve.