Banner image for 2019 Financial Markets Conference: Mapping the Financial Frontier

Policy Session 3: How Has the Housing Finance System Evolved since the Crisis, and Where Is It Headed?

A discussion of topics including mortgage underwriting trends, changes to the current expected credit loss (CECL) accounting standard in banking, and concerns tied to a proposal to release Fannie Mae and Freddie Mac from government control.

Transcript

W. Scott Frame: I'm Scott Frame from the Atlanta Fed, and I'll be moderating this next session—which segues nicely from Don Layton's talk with Raphael. This is called "How has the housing finance system evolved since the crisis, and where is it headed?" As was alluded to by Don and Raphael, there is increasing discussion in Washington related to the recapitalization, and eventual release from conservatorship, for Fannie Mae and Freddie Mac. The FHFA [Federal Housing Financing Agency] director, Mark Calabria, gave a talk yesterday in New York and said that this recapitalization program will begin soon, and that he would like to work with Congress but that he's not going to wait on Congress to act; it's been 10 years since the imposition of the conservatorships, and so a legislative solution to date has not occurred.

But the devil will be in the details. I think as Don mentioned in his remarks that it's technically, legally, and politically a challenging road ahead. So with that said, the U.S.—I have created this schematic here—the U.S. housing finance ecosystem is complex; mortgage functions are performed by a variety of intermediaries, depending on the characteristics of the borrowers or the loans themselves.

For our foreign guests, you'll notice that the system is not bank-centric, but rather relies very heavily on capital markets coupled with government support. While this broad structure schematic is largely, with some exception, what was present prior to the crisis, the distribution of risks and the institutions engaged in this market have evolved quite a bit, and often in response to post-crisis regulations.

Today we have three distinguished panelists, and together we're going to offer some thoughts about four broad, policy-related issues. So, we'll begin with Andrew Davidson, who is the president of Andrew Davidson & Company. He's going to talk about mortgage underwriting trends, with a specific focus on Ability-to-Repay requirements as promulgated by the Consumer Financial Protection Bureau's Qualified Mortgage rule.

So Andy...he's going to focus on the originator side there. Michael Bright will go second, and he's the president and CEO of the Structured Finance Industry Group, which he recently joined, and prior to that, he was the president of Ginnie Mae. He's going to discuss some issues concerning mortgage servicing by nonbank financial intermediaries.

Next, Mark Zandi is the chief economist for Moody's Analytics. He's going to offer some thoughts about the credit risk transfer programs at Fannie Mae and Freddie Mac that Don alluded to earlier as well as the proposed capital regulations for the two GSEs.  And then I'm going to back clean up, and talk a little bit about the evolution of the investor basis for agency mortgage-backed securities over the past two decades, with an eye toward what may happen as the Fed's presence in this market recedes.

So without further ado, I'll turn things over to Andy.

Andrew Davidson: The CFPB established rules for Ability-to-Repay, and the purpose of those was to avoid some of the problems that we had as a result of the financial crisis. And they established this thing called a "qualified mortgage;" and a qualified mortgage basically allows certain legal protections to the originator, and so there is a presumption that you verify that the borrower has the ability to repay.

And it can be achieved in two different ways. There's a set of rules that apply to most people, and the dominant part of that rule is a debt-to-income ratio of 43 percent. If you don't meet that rule, if Fannie Mae and Freddie Mac agree to originate a loan—so if it's approved by them; actually, they don't have to originate it, they just approve that loan—then that loan is also considered a qualified mortgage.

So the advantage of qualified mortgage is that it creates a rebuttable presumption that you've met the Ability-to-Repay rule—sorry, if it's not QM, then it's rebuttable, if it is QM, then it's constructive. So we'd need a lawyer to spell out what the detailed risks are to the originator and the other people in the process as a result of that, but for the most part most of the loans that have been originated since this rule went into effect are qualified mortgages, either under the "43 percent DTI" rule for non-Fannie and Freddie approved loans, or any loan approved by Fannie and Freddie.

The chart that we don't see here shows that as you go from very low debt-to-income ratios, from like 30 percent up to about 43 percent, you can see that the new delinquency rates rise pretty rapidly. And so you can say, "Oh, it makes (pretty much) sense, that's why we need to cut off the debt-to-income ratio so that we don't create loans that have too many risks." That's not quite the one, but we'll see if we can find the other one. [laughter]

And so you say, "Oh, it makes sense to cut off there." Well, it turns out that about 25 percent or so of the loans now being made are being made with DTIs higher than the 43 percent cut-off rate under the Fannie Mae and Freddie Mac rule—which they call the "QM patch"—and that rule actually expires either in 2021, or if Fannie and Freddie leave conservatorship.

So now we have about 25 percent of the loans being made above this patch, so it would be pretty clear that that's a pretty risky thing that's going on. But if we look at the slides, which we don't have, it turns out—as you can see clearly [laughter]—DTI is only one risk factor that affects mortgages. Some of the other important ones are the loan-to-value ratio, and the credit score, and then a third, very large area, of "what is the type of documentation that you used to demonstrate either your down payment, or your income, or the value of the home?"

And so it turns out that if you put the proper constraints on these other factors, that debt-to-income ratio isn't the thing that predicts delinquencies and losses—that you really need a combination of those factors. And it turns out, actually, if we look at the Freddie Mac loans—maybe we'll get you the charts afterward—but if you look at the Freddie Mac loans originated in 2016 that have the DTI higher than this 43 percent, when you look at those loans, they actually have lower delinquencies than most of the other loans that Freddie Mac was guaranteeing during that time period, because they balanced the other risk factors. And so what this really means is that going forward, we really need to find a system that says, "Post-patch, or even without the patch, get rid of a strict DTI limit and let's come up with a more balanced system of determining which loans do demonstrate borrower ability to repay."

If we don't do that—we said about 25 percent of the loans being originated now just wouldn't qualify, and it's a big risk to the system. So we're hoping that over the next year or two, we can find a solution to that problem.

So, I want to wrap up there; I don't know if anyone else has any comments they want to add to that?

Frame: Mark, do you have any thoughts about the QM rule?

Mark Zandi: Well, I think it would be very disruptive if we go get rid of the QM patch, and we go to the hard "43 percent DTI" threshold. I think you mentioned 25 percent of originations today are over that 43 percent threshold, and I think 20 percent of all Fannie/Freddie loans outstanding are over that 43 percent threshold, so actually not addressing this and reverting back to 43 percent would be very, very disruptive to the housing market and thus to the broader economy. So clearly, something has to be done here.

Frame: Michael?

Michael Bright:  Can you put the first slide up? Because I'm going to talk—the one that you had when you were talking—because I'm going to borrow from that, since Andy's particular...

Frame: I can't, but guys, could you put that schematic back up on the screen?

Bright: So, I guess there are a few things worth doing here. I know this isn't really entirely a mortgage crew, so it probably makes sense to take a little bit of a step back and explain certain things, because we use a lot of acronyms in our market. I find that if you ever touch the U.S. mortgage market, you find yourself apologizing a lot to a lot of people, so I have to apologize to you all for the use of acronyms in our market—and it is a very weedy place. Although I did sit through three hours of bitcoin conversations yesterday, so...[laughter] now you're going to hear about mortgages.

So to level set, in terms of the QM, qualified mortgage: Dodd-Frank established something called the ability-to-repay rule, and it basically said that if you make a loan to a borrower in the United States, you need to demonstrate affirmatively that the borrower has the ability to repay this loan. And it put in place a series of criteria that must be checked in order for the lender to be able to certify this ability to repay has been met. And those criteria include that documentation is there that the loan doesn't have a balloon payment, that whatever qualifications you use, use the fully amortizing rate—basically you can't rely on NegAm products or an IO product, you need to make sure that the borrower has—as the rule implies—an ability to repay.

Well, the mortgage industry—probably rightfully—said, "Without specificity on this, there needs to be some legal safe harbor from liability for the ability to repay." So if I check a series of boxes, you need to say, ‘I deem the ability to repay as being met,' and what they came up with was something called the qualified mortgage. So if you originate a loan that's a qualified mortgage, it automatically is a legal safe harbor from liability to the ability to repay, so you can feel confident in making these loans and continuing to stretch your credit quality.

Now, they put in place this temporary mechanism called a patch, which said, "We've got these blunt rules that describe ability-to-repay documentation, and there's supposed to be something about DTI, and we came up with 43," and as Andy pointed out, that's probably not a particularly good metric. But right now, the GSEs are in conservatorship, and they will be for probably another year—this is 2009. And so for a little period of time, if it goes through their automated underwriting system while they're in conservatorship, that'll be deemed as a qualified mortgage, too. And they have compensating factors for the 43 DTI; that means that, different LTVs, different credit scores, different DTIs can be approved by their engine (and some series of magic voodoo they do behind their engine that approves or disapproves a loan).

And so currently, that patch has just kind of rolled forward with this perpetual conservatorship. So now you have a market that's basically used to running everything through Fannie or Freddie's rules engine to get it approved, and then they say, "OK, now you have deemed a qualified mortgage." One of two things is going to happen in the next two years: that patch disappears in 2021, so the CFPB has to figure out what to do with it, or the GSEs may come out of conservatorship, at which point the patch expires as well. And so we're engaged in a big debate over how to roll forward this qualified mortgage patch for the ability to repay. And I know I'm supposed to do servicing; I'll do it in two minutes.

Zandi: And one other point—oh, did you have something else you wanted to say?

Bright: No, go ahead, before I go to servicing.

Zandi: No, I was going to say, on the QM patch: the one thing that makes this such a thorny issue is that if you do the modeling, trying to relate mortgage credit risk default to different loan characteristics like LTV and score, and then DTI, it's very difficult to pick up a statistical relationship between DTI and default risk—not because, I think, that DTI doesn't matter, that debt-to-income doesn't matter; it's intuitive that it would, particularly in a stressed environment. But the income data is very, very poor, so it's very difficult to measure DTI. So when you're actually building a model and trying to estimate, you can't pick up the relationship.

So it's confusing the whole discussion, and it's causing some people to argue there should be no DTI threshold in the construction of what should be considered a QM.

Davidson: I want to correct that a little bit, because I think, since the start of CFPB rules and better definitions of what income is used in the DTI calculation—actually, if we look over the last few years, like this beautiful 2016 data here, that we actually see that there is a correlation between DTI and losses, even once you address the LTV and credit score.  But an important part of making that work was having very good, defined DTI limits. So at the time when the patch was put in place, it is true that DTI meant almost nothing, because most people just reported the exact amount of income they needed to qualify for the loan. So you had most of the loans bucketed at the actual required DTI level; it was like 36 percent front-end, or 50 percent back-end.

It was an almost useless measure. But under the new rules I think we're getting to a point where there is better data, but the definition of the "I" from DTI is extremely important to everything.

Zandi: Well, I think you'd better do the modeling pretty quick, because the debate is happening and people are using the old data today, so hopefully we get to the modeling...

Frame: I want to ask one question, and then push a little bit, which is: We have the government insurance programs—FHA, VA, Rural Housing—we have the GSEs, so you're talking about two-thirds, three-quarters of originations (you guys may have the exact numbers in hand); but does it make sense then to come up with a DTI limit in the ability-to-repay rule, and then basically exempt three-quarters of the market from it?

Bright:  No, no, and that's in my view; and I think that's one of the lingering issues, is...there are a myriad of factors as to why so much of our market runs through a GSE or a Fannie—I'm sorry, or a Ginnie—but this is a big one, because you just can't get the legal safe harbor from liability without running it through the rules engine.

Frame: Yes, and I think related to this, as Andy—it's unfortunate about his slides, but he correctly points out that in the GSE programs, there are these compensating factors, which helps to balance out the risk. In the FHA program, they not only have...I believe last year about half of the loans had DTIs over 50, and we're talking about people with effectively zero down payment. So these are very risky loans, and I think that people should be thinking about that issue.

Davidson: Once again, just because it's a zero down payment doesn't necessarily make it a risky loan. There are loans done through housing finance agencies at the state level with very low down payments which have very, very low delinquencies. But it's a risk factor, and so it's always a question of, "How did you document the other factors in addition to the risk factor?" And if it is a risk factor, make sure that you really got it right so that if it is a low down payment loan you really understand the person's income. If it's an income-generated item, and it's a high DTI, you really verify the source of the down payment.

Frame: Well, what if I added that the FICO was 630?

Zandi: Credit score.

Davidson: Right. Once again, it all depends on how you layer in the risks, and at some point you've got to control the layering of the risks.

Frame: Yes, fair enough.

Zandi: This gets to the solution—well, one solution, a likely solution to the QM patch problem is you use an AUS, an underwriting system that allows you to account for all of these characteristics—including DTI, but score and LTV and other characteristics.  And then you can have a default threshold beyond which you would not get safe harbor. That would be another approach to it, and probably would get to the—oh, there you go.

Bright: So, Andy, what are your views on the QM patch? [laughter]

Frame: Apologies to Andy on this, but why don't we go to a couple of the key points here...

Davidson: Yes, let's just go to this last...so this just shows, on the X-axis is the debt-to-income ratio. The bars represent the volume of loans done at each of those debt-to-income ratios. The black line shows the rate of new delinquencies on those. And so the darker-colored bars to the right are the above 43 percent DTI, and then you can see, as I said, as you move to the very high DTI loans, the actual delinquency rate (shown by the black line) actually falls pretty substantially, due to the compensating factors. Of course, you could do DTI lending that... high-DTI lending looks bad, so it really is a question of putting into the proper compensating factors in an overall system.

But also, if you want to make this generally available to everyone, you have to make sure it's a system that's not easily gamed. So there's a lot of work that needs to go into this, but it's really crucial to our system. Probably fine to move on from there.

Frame: OK, I think we will try to pivot here little bit, and Michael would you like to talk—we're going to go back to our schematic, and talk a little bit about the servicing industry. So, not all mortgage originators necessarily hold on to the servicing rights... Michael, why don't you go ahead?

Bright: If you have that box in there; yes, there you go. By the way, trying to explain box and arrow charts to U.S. senators is a very difficult process. That's one of the reasons there's never going to be legislation, is the number of times you have to sit down—and it's understandable, these folks came from a briefing on Syria and then a briefing on health care, and then a briefing on this, and you sit down you're like, "Here's the system now"—with these boxes and arrows, "and we're going to move these boxes with arrows in them," and it's like—goes away quickly.

But this group, I think we feel confident that we can do the boxes and arrows. So the process that I want to talk a little bit about is the arrow from servicer to this green triangle in the middle, which is an intermediary that then distributes the principal and interest out to the MBS investors. For a very long time, there's been a mix of nonbanks and banks, but banks have played a very big role in that process. And when banks play a role in the collection as either the servicer or the intermediary, and ultimately the person responsible for remitting the P&I to an investor—when banks do it, then you have a company that has a little bit of a strong balance sheet. They certainly have prudential safety and soundness oversight, and presumably they have emergency borrowing ability from the Federal Reserve (they can use the discount window).

Today's dynamic is that that servicing in the United States is done very, very, very much by institutions that are outside of the banking sphere. So in Ginnie's space, it's about 75 percent of all P&I gets remitted through a nonbank servicer. In conventional space, it's a little bit lower, but it's still well north of 50 percent—I want to say it's maybe 60.  And so you have a dynamic right now where this specialty function—the collection of principal and interest from a homeowner, and then the remittance of that to the investor of record through a GSE or through Ginnie Mae's platform—is done by entities that are outside of the Fed's purview in the regulatory sphere, entirely.

The magnitude of this is quite significant; so at Ginnie Mae, $40 billion (roughly) every month of principal and interest flowed through on the 20th of the month. Fannie and Freddie—I think it varies of course, by month, but just for simplicity let's say it's about 80 billion. So you've got about one $120 billion of P&I every month that comes from borrowers, through this intermediary, out to the MBS investor; and in today's market, not only is that entire process being done by nonbank servicers who are outside of regulatory purview and outside of access to emergency funding, but the MBS investor buys an explicitly mortgage-backed security—either a Fannie and Freddie, who are two companies that are in conservatorship, and so the investors get favorable Basel risk treatment, et cetera, or a Ginnie, which gets really favorable Basel treatment but does have a legislative explicit guarantee.

That is to say that a disruption in that payment system would be highly, highly problematic. Conceivably, especially in Ginnie Mae's space, if P&I were missed on the 20th of the month, it would be a miss on the part of the United States government on a sovereign guarantee to the MBS investor. Karen Pence of the Fed has written about this as a liquidity risk in our system, and I recommend everybody reads this paper. I recommend very heavily that the Federal Reserve staff please read this paper. I think desperately what's needed is a Federal Reserve Board view on the risks that she raises, and on this issue at large.

The risks of a disruption to that come in a myriad of forms. One is, of course, a liquidity run, so that these nonbank servicers—think of Quicken, PennyMac, Ocwen; just, an institution that is not regulated—they rely very heavily on warehouse lines of credit to continue the flow of new business. They rely on principal and interest coming from borrowers who are paying their monthly mortgage on time, and they rely to some degree on the capital that they have.

At Ginnie, we had them run some stress tests where we said if delinquencies were to go up by just a couple percent, and mortgage volume were to drop—which it has—so that you don't have that monthly inflow of new origination volume, how high can delinquencies go before you are insolvent and you can't make P&I? It wasn't great; OK? These numbers weren't like, "Oh, we can show you we can withstand"—they weren't stress test numbers. These nonbanks, it's not like they can make it through an 8 percent unemployment rate environment and continue to remit P&I safely, so you have this macro prudential risk that someone has to get their arms around.

And the other risk that we don't appreciate enough, and that I really think the Fed needs to get its arms around, is an operational hiccup. So I won't give any names, but about my third month running Ginnie, what we'd do (and what Ginnie's job) is to really make sure that that arrow and the arrow to the MBS investor happens. That's Ginnie Mae's primary raison d'être, is to at all times ensure on behalf of the U.S. government that that cash flow is working. And if there was a problem with the servicer, we'd dip into our funds at Treasury and we'd make it on behalf of them.

So about three months into my job at Ginnie, we track the growth of these P&I in the custodial account; so every time borrowers pay their monthly principal and interest, it goes into a custodial bank. Ginnie Mae uses Bank of New York Mellon as its custodial bank. So you monitor the growth of these things, and then on the 20th of the month there's an ACH debit process where Fedwire comes in, pulls money out, and gives it to the investor of record.

Twice in 18 months on the job, on the 20th of the month we went in to debit an account. The money was there, and there was an ACH debit block, a fraud protection block that was put in place. And the fraud protection block, in this particular example, was set up because new fraud protections had gone into place and the issuer had to affirmatively say that Ginnie Mae is an approved debitor of their ACH account. And they had forgotten to do so, which just happens; some low-level operation staffer just thought they were switching fraud protection, didn't think they had to go through their list of approved debitors, forgot to put Ginnie, and we couldn't get the funds.

And so these were very stressful days, because we were running around trying to figure out how to get—in both cases, we came up with a patchwork solution. One time they wired the funds, another time they were able to convince the ACH, in Qatar, at like 4 in the morning; and so they were able to convince their fraud protector, which was a large bank, to remove the debit protections. But it was risky, so we started a process of going through every single issuer's fraud protection, et cetera, et cetera.

The other thing I found is, I said, "OK, well, let's go see Treasury and make sure we have a process, so that if we get to a point where either the money actually isn't in the account, or we're just not going to be able to get it, there needs to be a red phone where we can call Treasury and say, ‘hey, knock knock, two trillion dollars here; you have an explicit guarantee for us, we need our money.'" And there was no one at Treasury who had that red phone, so that was surprising, and so we had to establish the red phone network.

But you know Ginnie had grown to two trillion dollars in size, Fannie and Freddie are five. So collectively, you've got seven trillion dollars of explicit taxpayer, government-guaranteed MBS that are out there, and the entire process—from the collection and dissemination of the monthly P&I payment—is done with no Fed oversight, because almost 80 percent of it is outside of the banking sphere.

So I am in Florida today at this Fed conference, mostly to say, to implore the Fed—and probably in conjunction with the FSOC—to come up with a view on this as systemic risk, and figure out what to do about it. If we're migrating away from tagging institutions as SIFIs and instead looking at activities as being systemically important, please look at mortgage servicing as a systemically important function in our economy. Because I don't want to have a situation where on the 20th of the month, something is missed, either operational or market-driven or otherwise, and that the Fed says, "Wait, what's going on here?" So that's my primary message on that space.

Frame: So let me ask you a question here, Michael: so you have these operational issues; the Fed could have a view on that, you had the payments issue, but the full faith and credit is ultimately the question—but that's an issue for Treasury, correct?

Bright: Yes. Look, definitely having the account at Treasury would be something that's accessible, and having some clarity on exactly how the emergency cash funds would be disseminated on the part of Treasury—that's important, and we did that. You can do that bilaterally between an agency and the U.S. Treasury. My view is that, given the unique nature of mortgage servicing in the United States, given the unique role that it plays in our economy, given the fact that, during the financial crisis I think where the Fed's interest was really piqued was the failure of Countrywide, when they realized that eight million borrowers weren't going to know where to send their P&I, and that no one knew what investor to send that to.

So the collection and dissemination of this P&I is such a critical function in our economy that having a central bank that could potentially do emergency facilitation of funds, that can make sure that it's tracking this—that gives an imprimatur to everybody involved that there's an adult watching it, and ultimately that is looking under the hood at nonbanks to make sure that they have the wherewithal, both on a capital basis and a technical sophistication basis, to make sure they do this without disruption. To me, it's just sitting there for us to talk about, and so last night's discussion was brilliant; it was wonderful and I'm glad that the Fed is looking at leveraged lending and the risks that that poses. I very much hope that in the next six months, at one of these type of events, Chairman Powell gives the exact same conversation but about nonbank liquidity risk on servicing.

Davidson: It's good to step back and say, "How did we get to this point?" So traditionally, as Don pointed out earlier, we used to have a separate system for the mortgage world; and so between the savings and loans thrifts and their regulator, and the home loan banks, and Fannie and Freddie, there was sort of like a unified system that provided for funding of those institutions, so they had access to funding in various forms.  As a result of some of the changes in bank capital requirements—basically making servicing a penalized asset rather than a rewarded asset—and some other issues with liability for loans guaranteed by FHA, a lot of banks...and also, the way the banks felt that they were punished after the crisis, there was that old joke that if a plane crashes on the border between the U.S. and Canada, where are the survivors buried? And, you know...you don't bury survivors. Well, it turned out this wasn't the case—right? The banks who didn't commit the fraudulent loans were the ones who were punished the most, because they were the ones who were the survivors—right?

So we punished the survivors for the acts of the nonsurvivors, and so therefore the banks have backed out a lot—substantially—from the mortgage business, bringing in more of these nonbanks. And so, while there have always been some nonbanks and there has always been some of this risk, we now have a tremendous amount of the mortgage market with people who don't have access to the liquidity and the regulatory structure that you would find in the banking system. And that's what's making this a growing problem that has always been there to some extent, but is now a more important problem.

Frame: Do you have something to say on that?

Zandi: Yes; I think Michael's making a really good point that the liquidity issues in the Ginnie issuer system highlight a broader risk to the entire financial system: that much of the risk is now leaving the regulated part of the system, the depository institutions, the banking system—the part of the system that the Fed pays a lot of attention to—and being pushed into the so-called "shadow system." In the case of Ginnie—correct me if I'm wrong, Michael, but—in the case of the Ginnie issuer system, almost 85 percent of all the loans now are made by nonbanks. I think for Fannie and Freddie it's close to 50 percent, and for the entire mortgage market, it's 60 to 70 percent.

But this also goes to the leveraged loan market; this is one of the things I think that is not talked about with regard to the risk posed by the leveraged loan market, and that is: if you look at the entire leveraged loan market, it's probably 1.2, 1.3 trillion. I'm not sure where Jay got his numbers on CLOs; we rate all the CLOs, and it's closer to 600 billion.  Maybe he's including foreign CLOs, overseas CLOs—I'm not sure. But let's go with 1.2 trillion in leveraged loans, 600 billion in CLOs; that leaves another 600 billion that's going out there, kind of in the netherworld, that we know everything that's going on with regard to the loans that are in the CLOs, because we rate them—not that that should give you any comfort, [laughter] but we rate them and we know a lot about those loans.

But we know nothing about the other 600 billion that are out there in the netherworld, with all kinds of institutions that are very unregulated, very opaque; we have no idea, and we have no idea what the underwriting is like for those loans. So what's going on in the Ginnie issuer system is an issue for the entire financial system, and when Jay talks about, "Everything feels okay," I think it feels like he's just focused on the part that is in his remit. It's really...the concern is what's not in his remit, and that is where I think the concern should be.

Bright:  And that's a problem with regulators, and it's a frustrating thing. But I went regulator by regulator at Ginnie to explain this challenge, and say, "Look, can you guys—you're on the FSOC, can you at least raise this topic?"  More often than not, I got, "Oh, I'm glad that's not in my risk area." And I'm thinking, "Oh, my gosh; this is the failure of macro prudential supervision, where everybody does talk about these macro risks—and I still am hopeful that the FSOC will rise to the challenge of figuring out what to do with the fact that so much banking doesn't take place at banks anymore, and where those risks lie.

And I think—you call him Jay, I'll just call him Chairman Powell since I didn't get to have dinner with him.

Zandi:  Well, I had dinner with him last night, yes.  [laughter] I think you can call him Jay, too.

Bright:  So if Jay and Donald can sit down and talk about this...[laughter] they could work something out, no problem; Chuck and Nancy...

Everybody talks a big game about macro prudential supervision, but then when you get to see them on a one-on-one basis, maybe the chair of the Fed would be the exception—probably the chair of the Fed would be the exception, maybe the Treasury secretary as well—but the prudential supervisors themselves are trained in silos. I mean, they're trained to think in silos, they operate in silos. And I had high hopes that the FSOC was going to be a mechanism for really breaking down those silos, and it's done a little, but I don't know that it's done enough.

Frame: So thank you guys. I'm going to turn over to Mark here to talk a little bit about these credit risk transfer securities, and the new proposed capital rule for Fannie and Freddie, that Don Layton referred to earlier, and maybe you can give us a little bit more color on that.

Zandi: Sure; thanks, Scott—and I want to thank you for the opportunity to be here. I should also say, just for sake of disclosure, I'm on the board of directors of an MI company, MGIC, and I'm a lead director of the CDFI, so we deal with a lot of affordable housing lending. So I wear a few hats in the mortgage space.

I think it's fair to say that the housing finance system has evolved substantially since the financial crisis—particularly in the space that the GSEs reside in, Fannie Mae and Freddie Mac.  The rest of the system doesn't feel like it's changed nearly as much, in the 10 years since; but for Fannie and Freddie it's been a sea change, in lots of different ways.  You mentioned the—Don mentioned...is it okay if I call you Don?  [laughter] OK. Yes, not Donald. Don and I actually go way back; in fact, I'm the guy who came up with the single utility, which Don—I think you characterized that as far left; is that what you'd consider "far left?"  [laughter] That's disturbing to me, but OK, yes, OK.

There have been a lot of examples of this evolution, in a real sense; and I think to echo what Don said, we already have had GSE reforms—a significant amount of GSE reform; the common securitization platform and single security are good examples of that. But I would argue that nowhere has the evolution in the GSE part of the system been most significant as in the management of mortgage credit risk. And I suppose that shouldn't be too much of a surprise, right? Because that was ground zero for the financial crisis; seven million people, seven million households lost their homes during the crisis. For context, before the crisis there were 52 million mortgage loans outstanding; seven million of them defaulted during this period—so clearly, a very significant failure.

The good news is a lot has changed to make sure that doesn't happen again; underwriting is a good example. The QM rule, actually, I think has been very therapeutic, it's a debate as to how to make it work better and make it work in the context of Fannie Mae and Freddie Mac no longer being in conservatorship. But I think we'll find a solution to that, and I think it means that the mortgage loans that are being made today are substantially better than the ones...the most important thing about the QM rule isn't the DTI patch, it's that you just can't do certain types of mortgage products, right? I mean, that's the most fundamental change in QM, and that really has made a big difference in underwriting.

And right now, the credit quality mortgage space is pristine; it's as good as I've ever seen it. We get data from Equifax; I get every credit file in the country, every month—so it's a census of all the mortgage loans. And the delinquency rate, the all-in delinquency rate, is—and I've got data back prior to the crisis, even in the boom times—the delinquency rate today as of April is the lowest it's ever been in history, ever been in history. Some erosion in recent vintages, but it's really minor; but in aggregate, beautiful.

And also in terms of managing credit risks: some really big changes—systemwide, financial systemwide, I mean the bank stress tests are a really good example of that, stressing of a bank portfolio's. Fannie and Freddie have a stress test, and it's very therapeutic in making sure that they have enough capital.

CECL is coming; CECL is a big deal. A lot of debate about it; CECL is the accounting change in the way that financial institutions, including Freddie and Fannie, have to account for loan losses. Right now it's an incurred loss system; it's moving over to expected loss. A lot of debate about this; banks hate it, I might be the only person in the world who thinks it's a great idea. I think it's going to be therapeutic, make the system a lot less procyclical going forward, in my view.

But also a lot of changes with regard to capital, and the capital framework that folks that take credit risks have to adhere to. The best example of that is PMIERs, that's the acronym for the capital framework that Fannie, Freddie, and FHA put in place for the MIs, the private mortgage insurers. And even though the MIs don't like it, I think it was great; I think it really was very useful. It caused great changes in the MI business; now the MIs are all risk-based pricing, which they weren't doing before—but in my view that's probably a good thing, not a bad thing, in terms of credit risk.

But the two biggest things that have changed, in terms of making sure that mortgage credit risk is being managed better or appropriately, is the CRTs, the credit risk transfers, and the capital framework that Don mentioned; so I'm going to spend a few minutes on those two things. And so all that, just for a little bit of background.

On the credit risk transfers, you get a sense of how important they have become here; this shows essentially who owns the credit risk, where is the credit mortgage credit risk. This is the share of credit risk going to different parts of the system on originations in each of these years, from 2001 through 2018—I don't have all the data I need for 2018 yet, so that's a bit of an estimate. The blue part of the bar represents the GSEs, that's the credit risk that Fannie and Freddie are taking.

You can see Fannie and Freddie account for 45, maybe up to 50 percent of all origination volume; but by my estimate, they're taking about 25 percent of the credit risk. And the thing you'll immediately note is that the amount of credit risk that they're actually retaining is as low as it's been in the period that I'm showing, even prior to the crisis.

At the peak, during the crisis, Fannie and Freddie obviously played a role filling the void in the marketplace when private-label securitization (PLS) collapsed; and so the share of credit risk that they were taking back in the crisis was close to half, but they've reduced that to about 25 percent. One thing I will say: these are my estimates, so my calculations.  My sense is that I'm probably at this point overestimating the credit risk that Fannie and Freddie are retaining; it's probably a little lower than 25 percent, but just for didactic purposes, I think this is helpful.

The red part of the bar, that's the PLS market, the private-label securitization market. You can see in the boom times leading up to the crisis, that's when this market really took off.  At one point it was accounting for, as you can see, about 30 percent of all the credit risk. It basically shoved FHA—that's the green part of the bar—out of the market altogether, back before the crisis, and also Fannie and Freddie.

One amazing statistic for me is the share of mortgage debt held by Fannie and Freddie fell 10 percentage points—not the mortgage originations, but the share of debt outstanding fell 10 percentage points between 2003-2004 to 2006-2007. That, to me, is the single most important statistic suggesting that the cause for the crisis was not Fannie Mae and Freddie Mac—they were shoved out of the market—it's that red part of the bar that's the PLS market that was at the root of the problem.

Take a look at FHA—that's the green part of the bar—that's Michael's space when he was at Ginnie; here I'm probably underestimating the amount of credit risk being taken there, it's about 20 percent. You can see bank portfolios—that's what the banks are retaining (that's the gray part of the bar); the blue part of the bar is private mortgage insurers—that's MGIC, the company I'm on the board of. Now look at the orange part of the bar; that's the CRTs, the credit risk transfers. You can see they began back in 2013, and they've steadily expanded and now account for at least 20 percent of the market.

So here's the bottom line: if you add up the blue bar and the green bar, that's really what taxpayers are on the hook for; it's about 50 percent of the market, and about 50 percent of the market is now private, what I consider to be purely private. A little bit higher than it was precrisis, but pretty consistent to where we've been historically—under the working assumption that precrisis Fannie and Freddie had an implicit guarantee, essentially they were backed by taxpayers under that working assumption.

CRTs have been, in my view—and I agree with Don—a slam dunk success, an amazing story; I really think, a significant innovation. This includes: about 75 percent of those CRTs are to the securities markets, about 20 percent are insurance and reinsurance; and of the remaining, half is lender recourse and some senior subordinated structures.

I will say, one of the most interesting developments recently is that the MI companies are now doing CRTs, right? So MGIC is just piggybacking off of what Don did at Fannie and Freddie, and we're now offloading risk into the market using CRTs. So you can see the risk distribution here is becoming across the board, across all institutions.

One of the most recent innovations that's been very important is bankruptcy remote trusts. This is a new innovation, which I think is a great change because it takes some of the counterparty risk off the table. So if Fannie and Freddie become private, there's going to be counterparty risk which might cause the cost of CRTs to rise. Right now, if you're a CRT investor, you're backstopped by Fannie and Freddie because Fannie and Freddie are part of the government—you're backstopped by the taxpayer. But if they get privatized, then there's going to be some counterparty risk; and this structure that's just being now put into the marketplace, that helps to address that issue—and it also helps with matching of when credit losses occur and when you actually get the payout from the CRT.  So that helps with the timing issues that could be very important here.

So this is, in my view, a very, very important...of the five trillion outstanding in Fannie and Freddie debt, three trillion is now backed in some way by CRTs—so a very, very significant innovation, a big change. And in a really fundamental sense, this is GSE reform—right? We're taking risk away from the taxpayer, and putting it into the private system; exactly what you want. And you're distributing the risk more broadly throughout the system so it reduces the SIFI risk that Fannie and Freddie pose to the system. So in my view, this has been a fantastic innovation.

Couple of quick things to worry about: one is, CRTs have existed in a pristine credit environment.  So they were put on the planet in 2013 to now, and the capital markets have been just going in one direction, and that is "everything's great." Credit spreads have been coming in because of all the liquidity in the system that has juiced up demand for anything with a yield. So one question would be: How will these CRTs perform in a risk-off environment?

And you could get to a case where the cost of the CRTs in a risk-off environment is greater than the guarantee fees that Fannie and Freddie are now charging. And so then the question is what happens in that case? Do Fannie and Freddie start retaining the risk again on their balance sheet, because they can't offload it into markets? Or do they raise their guarantee fees, which means higher mortgage rates? So the system could become more procyclical, so something to consider, and at some point we are going to get a stress in the capital markets, and we're going to see how well these things work.

One other concern about the CRTs is it might be more costly than entity-based capital. MIs, like the GSEs, go out and raise capital, and they use that capital and they can distribute the risk over time. CRTs, you can't do that; I think that's a minor issue, but something to be conscious of that might be a little bit more costly than would otherwise be the case. But broadly speaking, I think a very significant innovation.

Davidson: So then on that point: I think if you look now, the CRT is definitely much cheaper than capital because what you're doing is you're taking an entity that is

essentially a monoline entity that would have whatever capital charge it has—10 percent, 15 percent times capital—and then instead of holding that as equity, it becomes mezzanine bonds, and those mezzanine bonds have spreads from 50 basis points to 1,000 basis points on the riskiest—but mostly in the 500-600 range, on the upper end.  And so you're ending up with an all-in cost of capital roughly in the like 4 or 5 percent range rather than 10 percent range because it's just better to diversify the source of capital away from monolines—which, to me, is why the MIs are finally waking up and saying, "Oh, this is cheaper than us going out and raising more equity. It's good for Fannie and Freddie because it brings in this other source of capital, and we can't raise equity."

The other system before was just based on, "Well, these people were locked in; we could only use this"—it wasn't part of the competitive economic environment. And so the CRT has really made it so that these are now priced like other financial instruments, and just more efficient than equity.  It may be that there are some problems in some environments, but my guess is that those are small relative to the difficulty of raising new equity if you have legacy losses sitting on your books.

Zandi: Do you want me to turn to the capital framework for a little bit?

Frame: Sure, yes.

Zandi: OK; so the other significant innovation, or change, evolution—the system that I think is important—is the new capital framework. Now, this is more implicit than explicit, right?  I mean, it's not like Fannie Mae and Freddie Mac hold capital—I think they hold a few billion in capital; that may change given what's going on in Washington, and likely will (they'll start to retain capital).  But this is more theoretical, in the sense that they're not holding capital, but their pricing—meaning the guarantee fees they're charging—are consistent with this capital framework. And just to give you a sense of it, if we could go to the next slide, you can see that—this is my, again, my calculation of the system for Fannie and Freddie loans through the cycle for the typical borrower (the borrower that's in the middle of the credit distribution). So just kind of right down the middle.

It shows that you would expect in an inequilibrium—meaning the economy's at full employment, inflation's at the Fed's target, the economy's growing at its potential, there's no global QE, everything's functioning as you would expect it—the typical 30-year fixed-rate loan should be going for about 5 1/2 percent.

You can see how I break that down. The yield on the mortgage security, that's 4.4 percent, 440 basis points. There are some costs for servicing and origination, I'm estimating 50 basis points, and there's the G fee, that's the guarantee fee that the GSEs are charging, and then I decompose that into the various costs that go into that guarantee fee, the most important being the cost of capital, expected credit loss, some administrative costs, and then—the payroll, we have a payroll tax surcharge; I think that ends in 2022 or something.

But I kept that in there only because my guess is in some future system, that payroll tax surcharge is never going away. It's probably going to be there in one form or another, either it's still going to go into the Treasury, or it's going to be used for some kind of affordable housing plan—use the 10 basis point fee to generate some revenue that you can use for affordable housing. So I just kept it in there.

Then you can see how I come up with the cost of capital, and here's the key point: look at the equity position of the GSEs—about three and a quarter percentage point. This comes from the capital framework, and that is based on looking at all the different risks that the GSEs are facing, including credit risk, operational risk, liquidity risk, there's kind of a SIFI add-on there for going concern, all those kinds of things.

That's based on their current book of business. If you go back and you run the capital framework, this theoretical capital framework, through the types of loans that Fannie and Freddie had before the crisis—they were obviously lower quality loans—the amount of capital that they would have to hold was something closer to five and a quarter or five and a half percent, so it gives you a sense of how the capital framework would work in an environment where underwriting standards are being lowered. By the way, that's precisely the amount of capital Fannie Mae and Freddie Mac would have needed to absorb all of their losses in the Great Recession. They would have survived, they wouldn't have been put into conservatorship.

So the capital framework is very much built around the experience of the Great Recession. Also, one concern that people have had about a new capital framework—particularly if you're on the far left, it would be: What happens to people on the edges of the credit distribution—you know, the folks at lower income, lower quality? And this kind of gives you a sense of that. This shows the mortgage rate, my calculation of the mortgage rate, for different types of borrowers based on credit score—that's 620, the first bar is 620-700. For context, the average credit score in the country is 700, with above 80 percent LTV; so this is your low-risk borrower. If it was full risk-based pricing, then you'd get something close to eight percent, but under the capital framework, you see some subsidization here, their mortgage rate comes down about seven and a half.

If you go to the high quality bars, go to the bar at the bottom of the table; that's high score, above 740, low LTV (below 60 percent). You can see that those borrowers are paying more than the risk-based price under this framework.  So in my view, the capital framework is pretty good—consistent with current pricing, so it would not disrupt the market if in fact this was adopted. I think it's at least a good starting point for where the capital framework should ultimately end up when it's actually implemented.

A couple of things to note; one is this result here does depend on the GSEs continuing to deliver the cross subsidy to the system in the future in the same way that they're delivering it today. And the way they do that is they use different returns on equity, targeted ROEs, based on where the borrower is in the credit distribution. So if you're a low-quality borrower, they charge a lower ROE. If you're a high-quality borrower, they charge as if you want a higher ROE so you can cross-subsidize in the system.

So to get this kind of cross subsidy in the system, you're going to need the future GSEs and their regulator to continue to manage in the same ways this cross subsidy, in the future as they do right now (if this capital framework is adopted). That is a little bit of faith, or maybe it can be codified in some way; but that's something that needs to be addressed, I think, if there's going to be enough political support for GSE reform actually to get the GSEs out of conservatorship, and for reform to actually get across the finish line.

The final thing I'll say is that this may be procyclical in the sense that if you start to see losses on your mortgage book because of the way the framework is established, the GSEs have to raise more capital against that and it could result in higher costs and higher G fees—in a time when that's probably, if you were a prudential regulator, you wouldn't want to see that, you would want to do just the opposite. But on the other hand, the MIs are under that kind of a capital framework, too. This is the way the system has gone, so I'd say "what's good for the goose is good for the gander" seems like a pretty good approach.

So a lot more work to be done here, but I think we've made a lot of progress, and I think this is a pretty good place to begin the discussion about the future system.

Frame: I totally agree, Mark. In the interest of time: Michael or Andy, do you want to add something?

Davidson: On here...one of the things we'll probably talk about is the competition—should you have multiple guarantors—and this chart's actually a good place to look at that. So right now we do have this cross subsidization that's provided by Fannie and Freddie basically overcharging the high credit borrowers relative to their credit risk and undercharging the lower credit quality borrowers. So imagine if you had a new GSE, you could come in and basically concentrate only at the bottom of this chart. Well, they would clearly be more profitable, they could start cutting their G fees, and then you wouldn't be able to do the cross subsidization.

And so if you are going to allow multiple guarantors, you have to either say there is no cross subsidization, or, everyone has to do cross subsidization. Or you have to have regulated guarantee fees, and even then you'll have a problem because you still have one entity which might be more profitable than another, based on their mix of business.

So this is just another complication in what is the path to creating more guarantors.  Personally, I think two is already plenty, and we shouldn't move to more. But once again, this would just be another obstacle that would have to be overcome.

Bright: Yes, definitely. Look, economics—the way any of us learned economics are like not even in the room at this moment, so the whole structure that we have in the American housing market was politically constructed to achieve political and social goals, and that's just the way we have it in the U.S. So Fannie and Freddie were chartered by Congress, Ginnie Mae was created by Congress; it's meant to achieve an end that is deemed as being politically better than a pure, Pareto-efficient economic outcome.

And so when we talk about reform, it is difficult to square "we're going to bring in these economic forces, and they're going to do these economic goods," but at the same time we have these outcomes that we want to have happen, and that's a tough boundary condition to operate within, and it explains why 10 years later, the government owns the entire mortgage market. It's not an outcome that makes intellectual sense on paper, but it is an intellectually honest outcome in the sense that it does very much map to what the political economy of the United States would accept.

Frame: Thanks, Michael. I wanted to go ahead and turn to the funding side of the market. So, going back to our schematic, we've kind of gone from origination to servicing to managing credit risk, and I want to talk a little bit about market risk and basically who funds a large share of U.S. mortgages—and these are going to be the mortgage pools that are guaranteed by the three agencies: Ginnie Mae, Fannie Mae, and Freddie Mac.

Now, as part of its policy normalization, the Federal Reserve has limited its reinvestment of principal received on its agency MBS portfolio, which has slowly led the portfolio balance to decline. And so as a result, our holdings of agency mortgage-backed securities have gone down by roughly 10 percent over the last year or so, from roughly 1.8 trillion to 1.6 trillion. Now, principal redemptions—coupled with growth in agency MBS outstanding—has then had the effect of reducing our investment share of this market.

And so the primary things that I wanted to talk about today—the first is really: What has been the evolution of the clienteles for agency mortgage-backed securities over the last couple of decades? And I pulled some data here—these are estimates from Inside Mortgage Finance, but it gives you a flavor of how the investor bases have evolved. So in 2002, Fannie and Freddie, which are the purple bars, actually held 33 percent of all agency mortgage-backed securities outstanding. Commercial banks held about 20 percent, and foreign investors 6 percent (the foreign investors would be the orange slice of that bar).

So at this time, Fannie and Freddie were not only guaranteeing mortgages, and creating MBS, but then they were retaining the MBS and funding these MBS on their own balance sheets by issuing agency debt. And so it's at this time that you start to hear concerns in Washington about systemic risk emanating from the GSEs. And so those conversations happen, there were accounting scandals at both Fannie and Freddie in the early 2000s.  The regulator imposed capital surcharges, and it sort of changed the dynamics of this market.

So you'll see there, by 2007 Fannie and Freddie's share of this market had declined from that 33 percent down to 15 percent. Banks were roughly steady at 16 percent, but foreign investors—again, that orange slice—grew their share dramatically. So besides the capital surcharges at Fannie and Freddie, certainly what we see at that point in time is consistent with the savings glut narrative at that time, where there was a huge foreign demand for safe, U.S. dollar-denominated assets.

So in 2008, we have the financial crisis, and the imposition of the conservatorship. So then I wanted to look at 2009, which was only two years later. There, Fannie and Freddie were still holding about 15 percent of the market, banks 18 percent, but you can see how dramatically foreign investment declined. Again, it went from 27 percent to 12 percent in just 2 years. So these foreign investors, they don't seem to like this idea of this implied guarantee. If you read Hank Paulson's book from this period of time, he goes to China for the Olympics and the Chinese were not having any of this. They wanted to get some certainty that they were going to get their money back.

But it's at this time that the Fed then steps in with QE1, and that's going to be represented by the blue slice of that bar, and so the Fed very quickly emerges and holds 20 percent of the market then at that time. So in a lot of ways, what the Fed did was basically step in for the Chinese and the Japanese, in terms of this market. By 2014, the Fed had expanded its footprint and itself held 31 percent of all agency MBS at that time.  Banks also increased to 23 percent, some of that driven by the new liquidity rules. Foreign investors did not come back, and Fannie and Freddie—as part of the conservatorship agreements with the Treasury Department—they were required to allow their retained portfolios to run off over time.

So you can see really between 2002 and 2014, what sort of amounts to the substitution: The Fannie and Freddie portfolios recede, the Fed's portfolio really substitutes for it. In 2018, the Fed still holds about a quarter of the market, but this really then raises the question, as part of policy normalization: If the Fed is committed to reducing its footprint in this market in a very significant way, who's going to pick up this 25 percent of the market—or at least, at this time, a trillion and a half dollars of assets?

We've talked a little bit about this. Now, the proposals for administrative reform of Fannie and Freddie, at least at this time, do not envision large, retained portfolios. The past legislative discussions, I think certainly in the context of the administrative constructs that have been floated around over the past decade, there seems to be a broad consensus that these retained portfolios were unnecessarily large, and the social risks associated with them were outsized.  So Fannie and Freddie aren't going to come and backfill this, and so to me, the next logical investor group to look at would be, well, who holds U.S. Treasury securities? Specifically, foreign investors.

So I have two pie charts here. So the one on the left basically breaks down, as of the end of last year, the holders of marketable U.S. Treasury debt (of which there was about 6.2 trillion outstanding), and as you can see, 40 percent of that is held by foreign investors; the Fed holds about 20 percent, money market and mutual funds about 13 percent. So of that 40 percent held by foreign investors, where are those investors located? So China, including Hong Kong and Taiwan, is roughly a quarter; Europe, roughly a quarter; and Japan, 16-17 percent. So that seemed to be a natural place to look.

In talking about money market and mutual funds, one of the institutions that I have been studying recently are mortgage REITs. I found them to be interesting because they're basically like leveraged mutual funds. And there's a subclass of these institutions that only holds agency MBS, but it seemed like, out of the nonsovereign investors, that this seem like a fairly efficient vehicle to potentially pick up some of the slack here because they don't pay any corporate income taxes—all the income flows through to the investors.  And also, our experience during quantitative easing suggested that they behaved collectively like marginal investors in this market.

And I just wanted to end with a question for my colleagues here on the panel—but I think it may also relate to the last session on monetary policy, which is: If there is a recapitalization and release of Fannie and Freddie without legislation providing a full faith and credit guarantee behind these agency MBS, how might that change the Federal Reserve's thinking about the current stock of holdings of agency securities, but also the way we look at any possible future investment if there were to be another downturn?

So I'll turn it over to my colleagues and see if they have any thoughts on these...

Zandi: You mean recapitalize with no government backstop?

Frame: Correct.

Bright: So, I can give some perspectives from my time at Ginnie. So, half the Ginnie Mae job is being terrified of that thing that we talked about earlier, and then the other half was going around the world and selling the bonds to all the investors. So you get a pretty good perspective on what is on foreign central banks' minds, and pension funds globally, and what their concern is. So that pie chart that's on my right—yes, the breakdown of the foreign investors—that's ambitious for mortgages, even with an explicit government guarantee.

So for example, I had multiple meetings in Germany where an insurance company would say, "U.S. mortgage-backed securities? No thank you. I'm not taking the meeting. My boss got fired, he bought a bunch of mortgage-backed securities; we don't do it anymore." And you're like, "No, no, no. See, this is an official government passport, we're here representing a government agency. Here's a statute that shows that we have explicit guarantee," and then they maybe would politely listen to you, before then also asking you to leave. [laughter]

And so Europe, even with an explicit government guarantee, they're just totally burned out on U.S. mortgages. It's like a generational thing for a lot of folks there, before they come back. London, maybe—although they may not be in Europe much longer, so it'll be a different pie chart.

The Chinese actually are the most sophisticated, in my experience, at this, meaning: actually at SAFE—the State Administration of Foreign Exchange there, serves at the PBOC—one of their foreign exchange investors, I think she's worked at Fannie. And so they have a pretty good understanding of the subtleties of an implied guarantee, and prepayment differences. They'll yell at you like they did Hank Paulson if they're worried about not getting their principal back. It turns out they'll also yell at you if they think that they're getting their principal back faster than they expected—so you get yelled at by them over prepay speeds on occasion, but they know what they're doing.

So the Chinese can figure out most systems that we have. The Japanese are very conservative investors; their investor guidelines explicitly say Ginnie—we had a bunch of MOUs with Japan Post, Nippon Life, and the Bank of Japan—but, a lot of concern over what would replace conservatorship for Fannie and Freddie.

So broadly speaking, taking a step back, this is a community of folks who understand, on some level—and even though they understand it, won't be willing to engage in, but let's say understand on some level—two states. They understand an explicit government guarantee that's created by Congress. They generally know what that means; they know Ginnie, they know who to call if they're concerned over getting principal back or upset about getting it back too quickly—so that, they understand. They've come to largely understand conservatorship, and what that means. We are likely on the precipice of asking this community to understand something different with the five trillion dollars, and I'm not confident that we're all going to like the answer that we get.

There is trepidation over U.S. mortgage-backed securities; it just is still there. The Japanese kind of get it; like I said, the Chinese, a little. So there are sovereign wealth funds with U.S. dollars that they have to invest in, and they will match to the Barclay's index, so there's going to be some investors. But to quickly ask the world to buy Fannie and Freddie debt—which is a big part of this, that we always forget about talking about, because it's used to finance delinquent loans being pulled out of pools—or to buy Fannie and Freddie issued mortgage-backed securities without it being in conservatorship or without there being a congressionally-created, explicit guarantee, to me is a really big risk and I'm not entirely sure I understand what we get for it.

Now, that's mitigated a little if we do it with the existing Preferred Stock Purchase Agreement, or PSPA—another acronym I'm throwing at you. But basically some amount of capital that is really Treasury-pledged, to step in on the enterprise's behalf, but I just think it's a risky experiment. And when I was on the Hill and we were working on this—and most legislative formats, even Jeb Hensarling's legislative format, did have and contemplate Congress explicitly creating a defined role for the government that was paid for and that you could then go and tell these investors, "Here's the way this thing's going to work." If we're going to start pulling at the thread of the sweater without that, I'm not convinced it necessarily works.

Zandi:  Let's just say, recapping, releasing Fannie Mae and Freddie Mac with no government backstop would be a catastrophic error on every level, because the cost of a 30-year fixed-rate mortgage loan would rise very substantially, and the 30-year fixed-rate loan—or long-term, prepayable, fixed-rate loans—would become a shadow of what they are today. It would be a massive change to our system.

Now, we can debate the merits of long-term, prepayable mortgages, and whether that's a good thing or not, but getting from where we are today to that is going to be incredibly disruptive, and it's hard to even imagine going through that process. And you're just not going to find investors—and this is evident across the globe. Go look everywhere else on the planet (there might be a couple of exceptions, like Denmark because of their unique mortgage finance system). But the long-term, prepayable, fixed-rate loan is not a feature of other mortgage systems for a good reason: because you can't find investors who are going to take credit risk and interest rate risk at the same time without charging a very high rate to do it.

So this idea that we can go down this path is highly misplaced. The other thing I'd say in that regard is putting Fannie Mae and Freddie Mac back into the system, into the wild without a government backstop—and therefore without the government fetters that come with that backstop—means that Fannie Mae and Freddie Mac will be very large, systemically important financial institutions with tremendous market power, and will dominate the system, just like they did precrisis. And that is going to lead to outcomes that are not good.

So going down that path would be a grievous error, in my view.

Davidson: I just have one more point on the GSE reform. So it may be that the PSPA thing could work—I have severe doubts that it's stable enough—but what you're really doing is putting a gun in the hand of the GSEs that they can point at themselves and say, "Give me these powers or we're not going to be making enough money, we're not going be able to raise enough capital; if we don't have enough capital, this risk is on you."  And so you're playing around a little bit in a dangerous area where you're depending on individuals to take action.

So like I say, it could work.

Frame: I think we call that "moral hazard" in the economics business. [laughs]

Davidson: Right, and so you're recreating some of the moral hazard.

Bright: And it's worth repeating that if you look on paper and you say, "OK, well, what have we gotten for that?" Well, the retained earnings, that's money that was going to go into the economy anyway; I mean, that was going to go into the general fund. So instead of going into the general fund, it sits on the balance sheet of two companies that are in conservatorship. So you haven't really gotten anything there per se, right? Yes, if you do an IPO you would be bringing in new capital, so that's I guess what we're getting, but the magnitude of the dollars we're talking about—you're talking about 30, 40 billion of new, in exchange for asking the global community to trust this 5 trillion-dollar thing, with an unclear outcome? And what happens; Is it going to inflate the quality of these spreads, blow out, and do they rally? Can the Fed continue to buy it?

I mean, quite a number of sort of dangerous questions, that to me—it's unclear that it's worth it.

Frame:  Gentlemen, we're running a little bit low on time here; we've had a great conversation, but I did want to turn to two or three of the audience questions. So here's one: Mark had raised the issue of CECL, which is the new loan loss requirement. So, how do you all see the new CECL framework impacting the mortgage market, specifically the banks and the GSEs? And when we talk about impact here, you're thinking about what may happen to interest rates for mortgages.

Davidson: So, for the high-quality loans that are mostly being made, the lifetime loss just isn't really that large of an amount. My understanding is that looking at CECL has encouraged the GSEs to lower the attachment points on their credit risk transfer, so that they have even a greater offset, which even further limits what their CECL charge could be, especially in a down housing rate environment. For most of the other mortgage players, it's a lot of work to move toward CECL, but it's not their biggest CECL area because, once again, mortgage credit risk isn't the biggest area.

It is unfortunate that the CECL rules don't allow you to capitalize your excess earnings against your expected lifetime loss. So there's a little bit of asymmetry, which does mean that on the margin everyone's going to have to hold more capital against mortgages, or basically every asset, than they would have otherwise—which generally increases the cost. But it's not really a unique mortgage problem associated with CECL. Our clients are much more concerned about other CECL calculations than the mortgage calculations.

Zandi: So, there are two broad reasons why you want to do CECL. One is, it's less procyclical than the current incurred loss system—which matters a lot when you're in a boom time and underwriting is declining. Under CECL, financial institutions will have to reserve more against that, because their expected loss (because of the lower underwriting) will be higher. So it makes it less likely that those institutions are going to be aggressively lending in a boom time, and in a bust, because of the tightening in underwriting, they'll have to reserve less.

So net-net, it's still procyclical, but it is much less procyclical than the current system, which is, very clearly, highly procyclical.

The second reason is that it creates transparency—and I'm not so much concerned about transparency in the regulated part of the system, the banking system, but in the shadow system, it's incredibly opaque, we don't know what's going on. And CECL, one of the benefits of it is transparency around what's going on in these institutions, and so we're going to get an insight into what these fintech companies are doing and how much risk they actually do pose—because right now, we don't have a sense of that.

In terms of the mortgage market, this is a legitimate criticism that if regulators do not adapt—if the Fed does not adapt—to CECL, then what it will mean is that on the margin you'll have to hold effectively more capital against longer duration assets like a 30-year, fixed-rate prepayable loan or a lower quality loan. And so that dings lower- and middle-income households; you don't want that to happen.

But I would say, if you do the calculations we're talking basis points, we're not talking tens of basis points, but I would say that it would be very important for the Federal Reserve and other prudential regulators to adapt to the CECL accounting standards and make sure that CECL's not a backdoor way to raise capital. I don't think that's what it was intended for, and it shouldn't be.

Frame:  OK, thanks, Mark. This question was asked a couple of different ways, but I thought I would put up the one that's a little bit more provocative here. So some folks asked, "Why should the government be in the business of subsidizing 30-year fixed-rate mortgages? If you look abroad—Denmark's an exception here, in that they also have 30-year fixed-rate mortgages—but many other countries are still able to attain high homeownership rates with adjustable rate mortgages."

And as I think, I might add on to this in sort of feeding at this from the Fed's perspective. You saw what it took with QE, how big of a stimulus we had to provide to drive down rates; monetary policy in a world with 30-year fixed-rate mortgages seems to be rather muted. Now one might argue, in a world with especially very short-term adjustable rate mortgages, there's a lot of volatility.

But I'd be curious about your own personal thoughts on these issues.

Bright: Pass; next question.

Davidson: So from a monetary standpoint, the question is who bears the brunt of the monetary policy? So in a country with adjustable rate mortgages, the consumer—homeowner—bears the brunt first, right? Your monetary policy, its mechanism is through the homeowner. In a country where we have fixed-rate mortgages, the effective monetary policy is primarily through the banking system and through the corporate sector.

Yes, there are plenty of ways of structuring an economy; our economy has been structured this way—this is basically the deal we've cut. And we could try and cut a different deal, but I think politically it just seems that this is what we've chosen to do.

Bright: Yes; that's right. Sometimes I do think, though, that the homeownership rate is not necessarily the right metric for comparing cross-country. I think that there needs to be some sort of a hedonic adjustment to it, because we do have newer...there's a difference in home quality, I think, but that would be an interesting study. I just sometimes think it's a blunt tool.

But the answer to the question is, it's just a political decision. And it's one of those things that once you do it, and the system gets used to it, to pull the rug out causes a decline in home prices, and then that's a big reason.

So why should the government be in the business? Because it once got in the business, and so now we're in the business.

Zandi: That makes sense, somehow. [laughter]

Frame: That's a good answer.

Zandi: I understand that.

Bright:  Sadly true.

Frame: All right. Well, I'll tell you what: with that, I think we're going to wrap up this panel; everybody needs a cup of coffee. So, thank you very much. [applause]