Risk Transfer and Reform

On September 26 I participated in a conference call hosted by the Washington D.C. investment firm Compass Point, on the topic of “Mortgage Finance Reform and Credit Risk Transfers.” Below is the text of my prepared remarks for that call.


The topic I’d like to address this morning is credit risk transfers, and the role they might play in a reformed mortgage finance system.

Today there are two competing approaches to resolving the conservatorships of Fannie and Freddie. The first is what Treasury and those I call the “Financial Establishment”—commercial and investment banks and their supporters—have sought from the day the conservatorships began: to wind Fannie and Freddie down, and have Congress replace them with some mechanism more to the liking of large primary market lenders. Over the past few years there have been a number of proposals for doing this: Corker-Warner, Johnson-Crapo, proposals from the Urban Institute and the Milken Institute, and most recently a white paper put out in April by the Mortgage Bankers Association.

But since the election of President Trump and the appointment of Steve Mnuchin as Treasury Secretary, an alternative approach has emerged: administrative reform that would preserve and strengthen Fannie and Freddie, without legislation. This idea is backed by community lenders, affordable housing groups and plaintiffs in the lawsuits against the net worth sweep, and in June it was turned into a concrete proposal by the investment firm Moelis & Company.

The status quo of keeping Fannie and Freddie in conservatorship is very convenient for the government, but because of the net worth sweep lawsuits I don’t think it can be sustained indefinitely. And when the conservatorships do end I believe it’s more likely to be through administrative action than legislation. The deep divisions within and between the parties—as well as the House and Senate—over how an alternative to Fannie and Freddie would operate make it dauntingly difficult to come up with legislation in as complex and critical an area as our $10 trillion mortgage market. By contrast, if Treasury follows an administrative reform process it can build on two companies that have a proven record of success, use negotiation with plaintiffs to settle the lawsuits—which will be essential in any reform effort—and by keeping Fannie and Freddie at the center of the mortgage finance system extract about $100 billion in value from the warrants it holds on the companies’ common shares.

So far, though, the Mnuchin Treasury has shown no sign of moving in this direction. It hasn’t changed any of the policies put in place by the Bush and Obama Treasuries that were designed to make it easier to replace Fannie and Freddie by “winding down” their profiles in the market. One of those policies, which I’ll focus on today, is having the Federal Housing Finance Agency (FHFA) mandate the use of credit risk transfers, or CRTs, to move credit risk, and revenues, away from the companies to private mortgage insurers and investors in CRT securities.

We now know from discovery in one of the court cases that in December of 2011 Treasury proposed to “Develop a plan with FHFA to transition the GSEs from their current business model of direct guarantor to a model more aligned with our longer term vision of housing finance.” A key element of this plan was “to sell a first-loss position (or the majority of the credit risk) to the private market on all of their new guarantee book business within a five- or seven-year time period.” There was no requirement that these risk transfers make economic sense for Fannie or Freddie.

For the past four years the large majority of the credit risk transfers done under this policy have been securitized CRTs—Fannie’s Connecticut Avenue Securities (or CAS) and Freddie’s Structured Agency Credit Risk notes (or STACRs). Private mortgage insurers and reinsurers have seen only small increases in business from Fannie and Freddie, mainly, I think, because of their capital constraints.

I’ve been a vocal critic of CAS and STACRs because they’ve been so non-economic: the companies are paying huge amounts of money to insure very high-quality books of business against levels of credit loss that have extremely low probabilities of occurring. But what I hadn’t realized until I read the Treasury policy document a couple of months ago is that Treasury and FHFA both know this, and don’t care. Their goal has been, to use their word, to “de-risk” Fannie and Freddie, and in fact it’s easier to transfer business to the private market if it’s not economic for the companies—that is, if private investors get lots of revenue and not much risk of having to absorb credit losses. And that’s exactly what’s been happening.

But Treasury can’t continue to support doing CRTs this way. When we finally agree on a proposal for secondary market mortgage reform, the credit guarantors in the new system—whether Fannie and Freddie or some other entities—will be required to prudently manage the credit risk of the loans they guarantee. For the reformed guarantors, CRTs can’t be giveaways; they’ll have to be conceived of and used in a way that strengthens, not weakens, them.

Whoever the credit guarantors are, they’ll be given an updated statutory capital requirement, and either Treasury or Congress will have to determine how credit risk transfers fit into that requirement. There really are only two alternatives, which I call the CRT model and the equity model. In the CRT model, some fixed portion of a guarantor’s capital requirement, say half, must be made up of credit risk transfers, with the rest being common or preferred equity. In the equity model, all capital must be shareholders’ equity, although a guarantor at its discretion, and with the permission of its regulator, can substitute CRTs for equity when it thinks that makes economic sense.

Over the last few years a number of prominent reform proposals, including those from the Urban Institute and the Milken Institute, have incorporated the CRT model. Wall Street firms and capital markets investors—both of which make a lot of money from the CAS and STACRs being issued today—support the CRT model as well. But making credit risk transfers mandatory for our future credit guarantors would be a disastrous mistake.

The fatal flaw of the CRT model is that while Congress or a regulator can require a credit guarantor to transfer a fixed percentage of its credit risk, they can’t require an investor or an institution to take it. Institutions or investors will take the risk in good times, but when home price growth is slowing or prices are actually falling—and mortgage delinquencies and defaults are rising—they won’t.

The housing and mortgage markets are cyclical, and in downturns credit losses are concentrated in the most recently acquired loans; older business generally performs much better, because it’s had time to build up equity. What’s been called the “Great Recession” was no exception. Loans acquired before 2005 made up 46 percent of Fannie’s book of business at December 31, 2007, but they were responsible for just 13 percent of company’s credit losses between 2008 and 2016. In contrast, loans from 2006 and 2007, which were only 38 percent of Fannie’s December 2007 book, contributed 69 percent of those losses. (Loans from 2005 were 16 percent of the book, and 18 percent of the losses.)

This sort of skewed loss distribution wreaks havoc with a mandatory CRT program. A good way to see that is by assuming Fannie had issued CRT securities against all of the loans it owned or guaranteed before 2008. To simplify the analysis—without distorting it—we’ll look just at what Fannie calls its “M-2” security, which takes losses beginning at 1.0 percent of the initial principal balance of the pool it covers, and continues taking them up to 2.75 percent of that pool. We then can examine how these M-2 tranches would have performed with our three different loan groups: all loans prior to 2005, loans acquired in 2005, and loans acquired in 2006 and 2007.

For loans acquired before 2005, Fannie would have paid a total of about $15 billion in interest on its M-2 tranches, but because of their 1.0 percent first-loss threshold very few credit losses—well under $1 billion—would have been transferred. CRTs issued on these loans would have been almost completely ineffective. Fannie would have received a modest economic benefit from M-2 tranches issued against its 2005 book, paying $3 billion in interest and transferring $4 ½ billion in losses. The only significant benefit from CRTs would have come from Fannie’s 2006 and 2007 books: on those it would have paid just $4 billion in interest, and been able to transfer $20 billion in losses before all of the investors’ principal was wiped out.

But I’m sure you can guess what the problem here is. Fannie easily would have been able to sell CRTs against its pre-2005 loans, just as it’s been able to sell them to cover the even better books of business it’s put on over the last four years. And it probably still could have sold CRTs in 2005. But how likely is it that capital markets investors would have kept putting money into M-2 tranches throughout 2006 and 2007, and lost 80 percent of their investment? These are opportunistic investors, looking to make money on CRTs, not lose it. They can withdraw from that market whenever they want, and there are reliable indicators that signal when credit risks are about to spike. A slowing rate of increase in home prices is a classic flashing yellow light for rising delinquencies and eventual defaults, and falling home prices are a fire alarm for the same things.

U.S. home prices began to fall in the second quarter of 2006, so that means there is zero chance that investors would have continued to buy Fannie’s CRTs in the second half of 2006 and in 2007. If Fannie had relied on CRTs during the last recession for credit loss protection, it would have lost some $10 billion on the CRTs it did issue, and then been stuck with all of the credit losses from the loans it acquired after the middle of 2006. Private mortgage insurers wouldn’t have stepped in to replace the lost CRT investors, because by then the high winds were blowing, and the MIs were scrambling for capital to cover the risks they already had.

There literally is no good argument for the CRT model compared with the equity model. With the equity model, you go into a credit downturn with capital on your balance sheet; with the CRT model, you go in hoping investors will voluntarily supply that capital by buying CRT securities that let you burn up all their principal. And when they don’t, you’re out of luck. By the time you know you need the equity the capital markets won’t give it to you, and you’re very likely to fail.

The causes of the 2008 mortgage market meltdown were the lack of regulation of origination and financing practices, and a lack of capital. Dodd-Frank addressed the regulation of originations. Now we need to address the lack of capital by updating guarantor capital requirements to meet current standards of taxpayer protection, and then, to ensure that our secondary market financing system remains strong and resilient, insist that guarantors use actual shareholders equity, not contingent risk-transfer transactions, to satisfy those capital requirements.

CRTs still can play a role in a guarantor’s capital structure under the equity model. The guarantor would determine when and how to use each type of CRT, based on its assessment of their economics, while the regulator would determine the amount of capital relief to grant, based on the standard that the CRT transaction or mechanism be at least as effective in absorbing credit losses as the dollar amount of equity the guarantor is permitted to forego. This combination of an economic evaluation by the guarantor and an “equity equivalency” assessment by the regulator would ensure that CRTs used in the future do not weaken the mortgage finance system’s ability to withstand stress, as today’s CRTs so obviously do.