Are Lemons Sold First? A Discussion of the Mortgage Market

8/18/2016

Tom Heintjes: Hello and welcome back to another Economy Matters podcast. I'm Tom Heintjes, managing editor of the Atlanta Fed's Economy Matters magazine, and today I'm speaking with Kris Gerardi, a financial economist and policy adviser in the Atlanta Fed's research department. We recently published a working paper by Kris about the home mortgage market, and he's agreed to discuss his research with us today. Welcome to the podcast, Kris.

Kris Gerardi: Thanks, Tom—I'm very happy to be here.

Heintjes: Kris, the title of your working paper is "Are Lemons Sold First? Dynamic Signaling in the Mortgage Market," and it discusses the quality of mortgages and how that quality affects its subsequent resale. Just to help me and the listeners out there understand what your research found, can you talk a bit about what distinguishes a high-quality mortgage from a lesser-quality mortgage?

Gerardi: Sure, Tom. Default risk, which is sometime referred to as credit risk in the market, and prepayment risk are the two dimensions on which we typically judge a mortgage's quality. By default risk, I mean the likelihood that a borrower does not repay his or her loan in full, which is typically a bad outcome for a lender because the lender is often unable to recoup the full amount of the loan balance through the foreclosure process. Prepayment risk refers to the likelihood that a loan is paid off before it reaches full maturity, typically due either to the borrower moving and selling her house or refinancing into a new loan with better terms. Prepayment risk is really only an important concern for fixed-rate mortgages.

Economist Kris Gerardi at the recording of a podcast episode.
Kris Gerardi, financial economist and policy adviser

While both dimensions are important from a mortgage investor's perspective, default risk has received the most media and research interest due of course to the subprime mortgage foreclosure crisis a few years ago that precipitated the global financial crisis. Our paper is mostly focused on default risk.

Heintjes: I see. Kris, before we delve into some of the details of the paper, I want to touch for a moment on the two broadest forms of mortgage lending: one is originate-to-hold, or OTH, and the other is originate-to-distribute, or OTD. Could you briefly explain to us what is meant by originate-to-hold?

Gerardi: Sure, Tom. Originate-to-hold refers to what most people think of as the traditional form of mortgage lending. Basically, it's when the same bank or mortgage company that originates a loan to a borrower also retains the loan on its balance sheet, funding it typically with deposits or some other source of capital.

Heintjes: And then we have originate-to-distribute.

Gerardi: Right. Originate-to-distribute refers to the scenario in which the entity that originates the mortgage does not retain the loan on its own balance sheet, but instead sells the loan either to a different institution—typically, another bank or mortgage company—or sells the loan into the secondary market, where it's pooled with many other mortgages to create a mortgage-backed security, or what we call an MBS in the market. This latter process, which is referred to as securitization, over time has come to be the dominant form of mortgage lending in the United States.

Heintjes: OK, so we have OTD and OTH. Are there differences in default rates between those types of mortgages?

Gerardi: That's a good question, Tom, and it's been the subject of a significant amount of research, especially over the past decade or so. The answer is a bit complicated in that it really seems to depend on the time period and the specific sample of mortgages that are being analyzed. Originators base their decision to either retain or sell a mortgage on lots of different factors, which creates systematic differences in the types of loans that are retained and sold. In addition, there is a lot of dispersion in the preferences of different investors in the market, which has created many different secondary market segments that are characterized by loans with different levels of credit and prepayment risk. So in short, it really depends. There are higher-quality securitized loans, and there are lower-quality securitized loans.

Heintjes: I should have known there was no short answer. Speaking of different market segments, as you just were, some mortgages are securitized and sold to investors by Fannie Mae and Freddie Mac, with whom we're all familiar. But other mortgages are securitized and sold by private financial institutions, and these mortgages are known as private-label securitization. These are the loans that you focus on in your paper, are they not?

Gerardi: Well, we actually perform analysis on both types of securitized mortgages in the paper, but we do focus more on the private-label securitized mortgages, or what we often refer to as PLS loans. So we do a little bit of analysis on what we call the GSE [government sponsored enterprise] loans, which are the Fannie Mae and Freddie Mac, mortgages, but the bulk of our analysis was on the PLS mortgages.

Heintjes: What is the role of private-label securitization in the mortgage market, and what types of mortgages make up the PLS market?

Gerardi: During the housing boom in the early-to-mid 2000s PLS mortgages became quite popular. They're basically loans that do not meet the underwriting criteria of Fannie Mae and Freddie Mac. The PLS market is split into broadly three segments, according to the degree of credit risk inherent in the mortgages. The three segments are referred to often as subprime, Alt-A, and prime jumbo or jumbo prime. Prime jumbo consists of larger loans to borrowers with typically very good credit scores that exceed the conforming loan limit, which is the maximum loan value that the GSEs are allowed to purchase. The Alt-A PLS segment is also commonly referred to as near prime, and this segment of the market is typically characterized by loans to borrowers with slightly lower average credit scores than prime, but comparable to average credit scores in the GSE market, and in which borrower income and/or assets are less than fully documented. The Alt-A market is commonly thought of as a low-documentation mortgage market. Also, Alt-A loans were also more likely to finance investor or vacation home properties. And then, finally, subprime is the riskiest segment of the PLS market, as it's made up of loans that are usually below the conforming loan limit to borrowers with very low credit scores.

Heintjes: PLS mortgages played a fairly central role in the mortgage crisis, didn't they?

Gerardi: Yes, they certainly did, Tom. Subprime and Alt-A PLS were the first types of loans to default en masse back in 2006 and 2007, which triggered the subsequent global financial crisis in 2008.

Heintjes: Kris, at least before the housing crisis, investors who bought riskier mortgage bonds—made up of subprime mortgages, NINJA mortgages, and that ilk—got greater yields in return for assuming correspondingly greater risk. Is that dynamic still at work today, perhaps with less of the Wild West mentality that we saw in, for example, The Big Short?

Gerardi: In the summer of 2007, the entire PLS market actually shut down, with new originations completely drying up. Since then, the PLS market has yet to come back. Today, there is still some originate-to-hold activity performed by the portfolio lenders, mostly banks. But the majority of originated mortgages today are securitized by the Fannie and Freddie in the agency market. Unlike in the PLS market, investors in agency MBS do not take on credit risk, as the GSEs provide full insurance against credit losses. The vast majority of mortgages originated today are actually insured by the government through the GSEs.

Heintjes: You just mentioned Alt-A mortgages and low-documentation loans, and I have to ask: after the housing crisis, is this is still a thing?

Gerardi: Low-documentation mortgage originations abruptly ceased when the PLS market shutdown in 2007 and—due in part to new regulations in the aftermath of the crisis—have not come back. There are a few lenders that are willing to originate mortgages with slightly looser documentation requirements, such as accepting a W-2 form instead of fully documenting multiple years of detailed income flows, but these loans are really only available to borrowers with high credit scores and significant cash reserves.

Heintjes: In your paper, you discuss how the time to sale of a mortgage tells us something about the quality of that mortgage. Just to be clear, are we talking about the bank that lent the money to the homebuyer being the seller of the mortgage?

Gerardi: Yes, in the paper we focus on the originate-to-distribute segment of the market exclusively, where the seller of the mortgage is the institution that originated the loan to the borrower.

Heintjes: And then, we're talking about investors being the purchasers of these mortgages?

Gerardi: We're actually referring to the institutions that buy the loans from the originators and pool them together to create cash flows for the mortgage-backed securities [MBS]. These entities are usually called issuers or sometimes securitizers, and they then sell the MBS to investors in capital markets. So in a sense, we can think of the MBS investors as the ultimate holders of the credit risk associated with the underlying mortgage pools, and the issuers as intermediaries who choose the specific mortgages to buy, but who do not bear any of the credit risk themselves. Our analysis captures mortgage transactions between the originators and the issuers of the mortgage-backed securities.

Heintjes: So now that we have a firmer understanding of some of the components of the mortgage industry and the roles they play, let's get to the crux of your findings: why are some mortgages held back from sale longer? Why are the "lemons" put up for sale first? Is it a mortgage-bond version of hot potato?

Gerardi: I'm not sure that the hot potato analogy is the right one. The crucial mechanism at work in the paper is called signaling in the economics literature, and it arises due to the presence of asymmetric information. In our context, it's asymmetric information between the seller of the mortgage, which is the originator, and the buyer of the mortgage, which is our case is the securitizer or issuer. The idea is actually quite simple. Let's assume that the originator of a mortgage gleans some piece of information that the borrower is especially unlikely to default in the future. However, the originator is not able to credibly transfer this piece of information to the company that wants to purchase the mortgage—the securitizer in this scenario.

Heintjes: What would be an example of such a situation occurring?

Gerardi: An example might be that the originator discovers the borrower will inherit a large sum of money from a terminally ill relative in the near future, but the originator is not able to obtain formal documentation to prove that that's the case. So the originator and buyer agree on a price for the mortgage that's a function of the observable characteristics of the loan, such as the borrower's credit score and income. But that, unfortunately for the originator, does not take into account this piece of private information since it can't confirm it. It's unfortunate from the originator's perspective because if that private information was able to be credibly transmitted to the purchaser, then the purchaser would be willing to pay a higher price for the mortgage, since it's much less likely to default than an otherwise similar loan.

And while the originator isn't able to directly transmit its specific private information to the purchaser, economic theory tells us that it may be able to indirectly signal to the buyer that the loan is especially high quality by holding the mortgage on its own portfolio for a longer period of time before attempting to sell it. And by doing this it would be able to obtain a higher price for the loan.

Heintjes: OK, that makes some sense, except isn't it costly for the originator to hold onto mortgages for longer periods of time?

Gerardi: Absolutely, Tom. But the key thing to understand, is that it is less costly for the originator to hold high-quality mortgages compared to lower-quality mortgages because the high-quality loans, by definition, will have a lower probability of defaulting while on the originator's balance sheet. The purchaser understands this and all else equal will be willing to pay a higher price for a more seasoned loan.

Therefore, if private information on loan quality played an important role in the mortgage market, and originators delayed the sale of loans that they believed to be of higher quality based on their private information, then we would expect to see a positive relationship between mortgage quality and time-to-sale in the data.

Heintjes: That seems like a relatively straightforward prediction to test.

Gerardi: Well, it may sound like a pretty simple prediction, but it turns out that it's very difficult to test in the data. The reason is, the prediction is about loan quality based on private information held by the originator, which by definition is unobservable to the purchaser of the loan. Thus, one needs to somehow figure out a way to isolate variation in loan quality that is unobservable from variation that is observable, and that's a really tough task in practice. I don't have time to get into the details of our methodology, so I'll refer the listeners to the paper for more details about how exactly we go about doing this analysis.

Heintjes: And we will link people to your paper. But what exactly do you find?

Gerardi: We find a pretty strong negative correlation between the probability that a loan defaults—which is our measure of loan quality—and the number of months before the loan is sold. In other words, the longer that it takes for a loan to get sold, the less likely it is to default.

Furthermore, we find that this relationship seems to be strongest in the Alt-A segment of the market where, as I said before, the majority of loans are low documentation. We think this is important because there are several studies in the literature that have shown evidence that private information played an especially important role among low-documentation loans.

And then finally, we find that the relationship between loan quality and time-to-sale also depends on whether or not the originator is affiliated with the issuer, the institution buying the mortgages. Consistent with this idea, we find weaker results in those cases compared to when the originator and issuer are unrelated.

Heintjes: Kris, based on what you've observed, what are we able to infer about the mortgage market?

Gerardi: That's a good question. I think there are basically two broad takeaways from the paper. First, I think it provides some additional evidence that private information played an important role in the mortgage market during the run-up to the housing bust and mortgage crisis.

Second, it provides the first piece of evidence that delay of trade may have been being used to signal loan quality in the market. This is important because it suggests that market participants were using mechanisms to try to overcome the problems posed by asymmetric information.

Heintjes: We're all familiar with the idea of "information asymmetry," which is basically the advantage one party has when he knows more about a given situation than the other party. But with all the tools at our disposal now, is information asymmetry still a major factor at work in the mortgage market?

Gerardi: That's an important question, Tom, but unfortunately I don't have a concrete answer. Market participants and regulators certainly learned some valuable lessons from the mortgage crisis, and as a result, there is probably less information asymmetry in the market today compared to, say, a decade ago. In addition, it appears as the evidence suggests that asymmetric information played an especially significant role in the PLS market, which as we've just discussed is pretty much long gone and doesn't appear to be coming back any time soon. From that perspective, I'd say asymmetric information is probably still a factor, but much less so than it was before the foreclosure crisis.

Heintjes: So we can sum it up by saying, RIP PLS.

Gerardi: That's right.

Heintjes: Well, Kris, this has been a fascinating conversation, and I want to thank you for taking the time share your insights on your research with us.

Gerardi: Thanks a lot, Tom.

Heintjes: I should note that you can read Kris's paper on our website, and you can see lots of other interesting content on the Economy Matters web page. Well, we're at the end of another Economy Matters podcast. Again I'm Tom Heintjes, managing editor of Economy Matters, and thanks for spending time with us. Please come back next month for another episode!