House under construction

Housing Perspectives

Research, trends, and perspective from the Harvard Joint Center for Housing Studies

The New Proposed Capital Rule for Freddie Mac & Fannie Mae: Ten Quick Reactions

On Wednesday, May 20 the Federal Housing Finance Agency (FHFA), the regulator of Freddie Mac and Fannie Mae, the two government-sponsored enterprises (GSEs), issued a proposal for a new rulemaking to establish modernized minimum regulatory capital requirements for the two companies. This supersedes an earlier proposal from 2018, when the FHFA was under different leadership.

Since the GSEs own approximately 45 percent of all the outstanding single-family mortgages in America, the capital rule has the potential to greatly impact the American housing finance system in how it will drive the companies to act in the future, both strategically and tactically, and which will likely translate directly into an increased cost for the typical mortgage. What emerges from the proposal is a vision that looks to be politically or ideologically driven in several key ways and, curiously, a reversion to an earlier era of financial system design. This, of course, fits perfectly well into how highly politicized most things about housing finance and the GSEs have long been. Surprisingly, the capital proposal works directly against the GSEs being able to raise equity to exit conservatorship, despite Director Calabria, the FHFA’s head, having indicated that such an exit is his highest priority.

Director Calabria’s one-year tenure is well-known to be marked by tension between his initial ideologically-driven position on various issues versus the realities and data which have occasionally led to revisions in his stances; he has admitted this is his personal modus operandi. We shall see if the many comments on the proposal that will undoubtedly be received will similarly lead to revisions, which I regard as absolutely warranted.

Against this backdrop, here are ten quick and high-level reactions to the proposal to strategically orient anyone interested in understanding it.

1. The proposal is thorough, thoughtful, and very complicated. The document is 424 pages long, and on each topic the discussion seems to be well-considered, thoughtful, and in-depth. While it builds upon the 2018 proposal, it is mostly new and changed in significant ways. It asks for feedback on 106 specific questions, plus the classic “anything else” for the 107th. It also seems to aggressively seek complexity, despite some protestations to the contrary: every major topic seems to have multiple approaches with a “higher of” requirement, adjustments and minimums to override the working of specified formulae, or complex tables. The 60-day comment period (the minimum allowed) should be extended if FHFA is truly interested in equally-thoughtful feedback.

2. A strange mix of technical specificity and judgmental arbitrariness. The proposal has very specific, complex, and lengthy calculations for a risk-based capital requirement calculation. It then applies no less than three different defined capital requirements (in addition to the statutory requirement, which is functionally obsolete but can’t be changed without legislation) to come up with a “highest of” number, and using 9/30/19 as a baseline, the highest figure for the two GSEs together is $135 billion (the lowest is $76 billion, the average $105 billion). It then adds buffers to that number which total another $99 billion, for an overall total of $234 billion. Yes, almost half of the total (42 percent) is rough judgment numbers that seem to overwhelm the massively complex formulae.

There is also an accounting-driven leverage requirement. It is currently higher than the risk-based calculation, and thus would be “binding,” which presents some problems (as discussed below); it puts the total required capital up to $243 billion. Another judgmental buffer ($91 billion) is 37 percent of that total.

Make no mistake, despite all the complex formulae and intricacies throughout the document that give it an image of scientific precision, it is these judgmental buffers that drive the grand totals to the levels that will be required. This puts an aura of arbitrariness into it, which is not a good thing.

3. The proposed capital rule and the Federal Reserve stress tests seem incompatible. On 8/15/19, FHFA published the latest results of the annual Dodd-Frank Act Stress Tests (DFAST), which are designed to show the loss assuming a “severely adverse scenario” set by the Federal Reserve, which uses the same scenario in setting capital requirements for large banks. The results actually showed two losses for the combined GSEs: $18 billion and $43 billion, depending upon a key issue related to taxes. That means the $243 billion proposed capital requirement is 13.5 times greater than the calculated loss in the severely adverse scenario using the $18 billion loss; it is also 5.6 times greater when using the $43 billion. And while a large financial institution is indeed supposed to have more capital than simply what is necessary to clear a severely adverse scenario loss (the extra amount is generally referred to as a “going-concern buffer”) 13.5 times and 5.6 times do seem rather high.

So, can the proposed capital rule really be compatible with the published DFAST results? Even using the more conservative (i.e. higher) $43 billion DFAST loss, a total capital requirement of $243 billion, by simple subtraction, makes the going concern buffer equal to $200 billion. That’s over 80 percent of the total, which certainly appears excessive. This doesn’t add up somehow. The proposed rule does not address this major inconsistency with the DFAST results.

4.  A whiff of ideological reverse engineering? For several years now, as I have written in other articles, the more extreme free-market end of conservative think tanks have been suggesting that the GSEs should be subject to a simple 5 percent-of-assets capital requirement, just like banks, despite the fact that the GSEs are far less risk-intensive per dollar of asset. Using the usual $5 trillion dollar estimate of assets, that was usually rough-estimated at $250 billion. The more accurate figure for 9/30/19 (the baseline used by FHFA in the proposed capital rule) is actually $283 billion.

FHFA’s proposed capital rule, by adding in its large buffers, comes up to $243 billion. Since Director Calabria and most of his recently-hired staff have this free-market ideological background, there is an unfortunate, but real, whiff that this all began with an ideologically-driven answer—something in the $250 billion range—and worked backwards to hit that target. The incompatibility of the proposal with the DFAST results seems to add to the case that the proposal is just too high, likely by a lot.

5. Real world implications on guarantee fees (g-fees) and market share. The GSEs are monoline secondary-market mortgage companies that purchase, securitize, and then guarantee mortgage credit risk. For their single-family business, the largest, single most-watched policy variable is the average guarantee fee (“g-fee”). The obvious question to ask, then, is: what will be the impact of the capital rule on g-fees?

While this can’t be answered with any real accuracy, a general impact estimate can be made. In 2019, a very good year for the GSEs in terms of their financial performance, the two companies together produced profit of $21.8 billion. This generates a return on the $243 billion of required capital of 9 percent, which is a reasonable return to shareholders. A first-cut conclusion would therefore be that this all works out, that the companies can carry the capital requirement without having to change much of what they do.

However, the 2019 profit was not just higher than usual, but needs to be adjusted to reflect post-conservatorship standards, in particular payment of a fee to the US Treasury for its financial support. This fee has not been decided —or if it has, it’s a well-kept secret— but we need to estimate it. The most commonly-suggested figure during the many years of housing finance reform discussions would amount to about $4 billion after tax. However, when the GSEs are fully capitalized, this seems excessive, so I will use an estimate of $2 billion after tax (to also include some smaller, off-setting positive impacts). Throw in a modest reduction in earnings to reflect a typical year rather than an excellent one (I assumed a 20 percent reduction), and I roughly estimate the return on required capital is in the 6.5 percent range, clearly inadequate for investors, who will want a minimum of approximately 8 to 9 percent, even in today’s ultra-low interest rate environment.

This exercise makes many assumptions, and is indeed very rough. It would be false precision to make more assumptions to translate this back into a specific required g-fee increase. However, it is clear that g-fees will need to go up, possibly by a lot. I will leave it to others to make their own calculations, and construct a more accurate estimate as more is learned. At this point, I broadly estimate they need to go up by 20 to 40 percent (i.e. about 10 basis points to 20 basis points), which will directly translate into higher mortgage interest costs to many homeowners.

This is a large enough increase that it will also likely materially reduce the 45 percent market share held by the GSEs. In turn, this will move a significant percentage of mortgages to the Federal Housing Administration (FHA) and bank balance sheets. The Secretary of the Treasury has indicated publicly that he does not like the idea of the FHA having more market share as an unintended consequence of GSE reform because there is no private capital in front of the taxpayer. It is unclear whether banks would welcome this increase. Many free-market advocates believe the private label securitization market share would grow significantly as a result, but that market has such fundamental problems it is unclear how much it would benefit, if at all.

6. The leverage requirement being higher than the risk-based one is highly problematic. The proposed rule shows the minimum leverage requirement of $243 billion as being higher than the risk-based one at $234 billion. The explanation from FHFA is that, to be credible, a leverage requirement must be binding (i.e. the higher of the two) from time to time, and that at the current time it makes sense.

I could not disagree more. Having a leverage ratio be binding means that the capital system is incenting an abandonment of economically-driven risk-taking that is the hallmark of a well-run financial institution. Instead, risk-taking would be based upon accounting, where all assets are incorrectly treated as having the same risk, with all the distortions this caused historically (I examined this last fall in What the FHFA Needs to Get Right in its Capital Rule). This will include, of particular relevance to the GSEs, the push towards higher-risk lending and the rendering of credit risk transfer as falsely uneconomic and thus not worth doing.

Do we really have to repeat the sins of the past like this?

In fact, the only valid purpose of a leverage ratio in a capital framework in this day and age, when measures of risk are far more precise than the 1980s or previously, is to be a back-up that only becomes binding if the risk-based system goes awry and produces very low numbers. Starting with it binding, and expecting it to be so cyclically, is simply not correct financial regulation as I understand it.

7. A risk-transfer rose by any other name. The proposal addresses the two primary vehicles by which the GSEs have single-family credit risk transferred away from them: mortgage insurance (MI) and credit risk transfer (CRT). Counterparty risk is a core to this, but other considerations are also important.

Unfortunately, for some unfathomable reason, the approaches adopted are incompatible even though they represent substantively the same economic transaction. For CRT, the proposal calls for reduction after reduction in the capital relief given due to CRT, with the reasons seeming fairly arbitrary. It looks a lot like piling on. It shrinks by almost 50 percent the capital benefit of CRT transactions versus the 2018 proposal, probably rendering any such transaction uneconomic. Its approach to MI is not just incompatible, it’s incomparable. However, MI’s capital benefit is only reduced by about 5 percent, not 50 percent. The proposal does note that its MI approach is more favorable—far more, in fact—than that used by bank regulators.

This comes across as terribly biased (CRT being treated punitively, MI laxly) and unfortunately statements by Director Calabria for some time have indicated that this was his pre-existing viewpoint. The reduction in the concentration of mortgage credit risk via CRT, which is so important to reducing the systemic risk of the country’s financial system, has become recognized by many (including at Treasury) as a signal success of the conservatorship. This would completely undo that progress. (MI does not reduce this concentration risk, as the risk is still held by mortgage monolines, rather than broadly diversified throughout the financial system.) There will likely be many comments on this aspect of the proposal, and hopefully FHFA will listen and constructively revise it. In particular, having incompatible and non-comparable approaches for the same economic risk just seems below the professional standards of a proper regulatory capital framework.

8. The single security at risk? One of the major improvements in housing finance during conservatorship has been the implementation of the single security, i.e. that Freddie Mac and Fannie Mae single-family mortgage-backed securities (MBS) trade together. This improves liquidity, which translates into lower mortgage rates and increased competition between the two companies.

However, the mechanism by which the single security keeps the two MBS trading together does result in each company guaranteeing not only their own loans but large amounts of the other’s MBS. So, for example, Fannie Mae has to guarantee MBS issued by Freddie Mac, relying upon Freddie Mac’s guarantee of the underlying loans to support Fannie Mae’s guarantee, and of course, vice versa. This has been considered a nil risk exposure, however, as the MBS of both companies are supported by strong agreements (but not a full guarantee) from Treasury. In the 2018 capital proposal, this cross-exposure was considered to require no capital to support, as would be true for direct obligations of Treasury. In the current proposal, there is a requirement for capital support, in line with how bank regulation treats such near-but-not-quite Treasury risks. This creates a significant friction when friction-free seems to be required for the single security to work as intended.

The impact of this friction on the single security is not yet known. But it is not good and could, in fact, sink the entire project. More thought and homework are definitely needed on this. Replicating the simple bank regulation approach (a 20 percent risk-weight, as it is known), as FHFA did, may be inappropriate in this situation, and FHFA should consider something customized, as it did many times in the capital proposal with respect to other risks.

9. A backwards-looking strategic vision. During conservatorship, with the awful experience of the 2008 collapse of the financial system fresh in memories, a different vision of the GSEs emerged. Instead of being massively concentrated pools of a single risk (i.e. mortgage credit) that will always come under pressure in downturns (especially any concentrated in housing), the GSEs would become channels through which credit risk would be transferred broadly throughout the financial markets, just as interest rate and liquidity risk are distributed via pass-through MBS. This would materially improve the overall strength of the country’s financial system. This also appeared to be more capital efficient, as credit risk was laid off via CRT at comparatively low cost. This vision was held under multiple FHFA directors, representing both side of the political spectrum.

FHFA’s capital rule proposal undoes all this. It goes back to the business model that existed prior to 2008, with all its concentrated mortgage credit risk, and it deals with the resulting systemic risk impact by requiring massive amounts of capital to support it in the name of safety and soundness. And if the only concern one has is safety and soundness, perhaps it all looks good. But regulators should also be worried about financial system efficiency. The amounts of capital to be tied up by the proposal are inefficient, and to society that is a waste. The pass-through model being implemented in conservatorship (pass-through credit risk on top of the existing pass-through of interest rate and liquidity risk) is far more economically efficient, not just in terms of an individual company’s view but in terms of perhaps $100 billion of the country’s wealth sitting idle to meet unduly excessive regulatory capital requirements, when it could be employed elsewhere in the economy more productively. That’s roughly worth $10 billion of income every year to the country, or about $75 to $100 per family annually.

And since that idle capital requires a return, it means g-fees must go up too.

Add in the possible ending of the single security, as described above, and the vision, whether intended or not, of the GSE capital proposal is to return to the past, but with a lot more capital. It’s simply backwards looking. And the unintended consequences, especially more mortgage credit risk flowing into the FHA, which has no private capital in any form sitting in front of the taxpayer, could be highly regrettable.

10. This is going to hurt the equity-raising needed to exit conservatorship. Director Calabria has said many times that his top priority is to end the conservatorships, and have the two GSEs raise enough equity to be well-capitalized. This capital proposal, however, will make that harder.

First, the companies will need to raise equity in amounts never before done, by an even larger margin than expected. Second, the projected return-on-equity (ROE) of the companies, which is the single most important measure of financial success for large financial institutions, will be under-market, and thus render the shares in any IPO unattractive, until enough years have gone by to show that increased g-fees and other impacts of the capital rule proposal are able to generate an ROE of at least 8 to 9 percent after tax. Third, the increased pricing required to deliver a return on the $243 billion level of capital will shrink the GSEs’ market share, leading to declining revenues and profits, which investors will not like.

Buried in the proposal as well is the notion that the full $243 billion isn’t really officially required. It’s only required if the GSEs wish to escape restrictions on “capital distributions and discretionary bonus payments.” They can be considered adequately capitalized at a lower level of capital. That’s nice in a regulatory sense, but in the real world, especially when it comes to capital raising time, it’s a meaningless distinction. No investors will want to put money into a company that has such restrictions on it: the full $243 billion is what they will need. In fact, a large financial institution can’t run at exactly the level of capital required by its regulator but needs a working cushion. Research shows something like a 5 percent buffer is needed, raising the actual required capital for normal operations post-conservatorship to be $255 billion.

To conclude, this proposal has a lot of good detail and depth to it. However, at key junctures it is strategically backward-looking or driven by ideology more than is appropriate. I hope it receives many comments from interested parties to generate significant reworking, with the result that it develops a reputation as professional rather than political, right down the middle of the fairway. Otherwise, it risks becoming a political football, like so much about the GSEs and housing finance, likely to be significantly revised every time a new FHFA director from the other party is appointed.