Wharton's Benjamin Keys looks at why the systems to protect taxpayers from the kind of mortgage market losses suffered in the financial crisis are still not strong enough.

Eleven years after the real estate bubble popped, leading to the financial crisis, there remain huge, unresolved issues connected with mortgage loans and mortgage insurance. Notes Benjamin Keys, a Wharton real estate professor: “… We have a nearly nationalized mortgage system. Between the FHA, Fannie Mae, and Freddie Mac, they cover well over the majority of the exposure to mortgage losses, and that doesn’t seem very consistent with broader pushes to involve private capital … to make the mortgage … respond to supply and demand in traditional ways.” What’s more, the private mortgage insurance market, which failed spectacularly in its supposed role as an early warning signal of housing troubles during the mid- to late 2000s, is still not necessarily up to the task despite reforms, Keys adds. He looks at why these issues are so critical to the economy — and what fixes could be made — with co-author and Fed economist Neil Bhutta, in a new research paper titled: “Eyes Wide Shut? The Moral Hazard of Mortgage Insurers During the Housing Boom.” In this Knowledge at Wharton podcast, Keys discusses the highlights of the paper. (Listen to the full podcast above).

 An edited transcript of the conversation follows.

Knowledge at Wharton: Your paper takes a fascinating look at mortgage lending practices, specifically mortgage insurance practices in the lead-up to the financial crisis and the utter collapse of the real estate bubble. Looking at mortgage insurance practices is a bit of a proxy for measuring default risk, correct?

Benjamin Keys: That’s right. The mortgage insurance industry is exposed to mortgage losses. They’re in a first loss position if there’s a downturn and there’s a foreclosure wave. And so the volume of business that they’re doing is right on the edge of assessing how much risk there is in the system.

Knowledge at Wharton: You found that while insurers are meant to be a check on risky behavior by lenders, the private mortgage insurance companies actually took on greater risks without charging higher premiums. This is all tied to lending guarantees by Fannie Mae and Freddie Mac, and it’s all wrapped up in a huge “moral hazard,” as the title of your paper suggests. Could you start with a refresher on what “moral hazard” is.

Keys: The idea here is that there is a party that’s taking on risks but is not necessarily being compensated for those risks. And it’s being exposed to additional risk-taking. It’s kind of a complicated context, but Fannie Mae and Freddie Mac are these enormous players in the mortgage market. They are federally guaranteed, and at the moment they are in conservatorship. But prior to the crisis, they were a public/private amalgamation — they were essentially private in the sense that they had shareholders, and they were run sort of like a private company. But at the same time, they had a public guarantee that was really implicit — this “too big to fail” guarantee that basically investors knew at the end of the day, the government would step in and insure them.

 “We had … all mortgage insurance applications over a 20-year period. And with that data, we could see just how much the firms relaxed their lending standards during the housing boom.”

And one way in which there was a layer of protection built into the system — or it was perceived to be protection — was a private mortgage insurance requirement. This was intended to bring private capital into the mortgage market, in ways that it might not otherwise have been there, for any loans that Fannie and Freddie wanted to make and guarantee that had small down payments. So when people are putting down less than a 20% down payment, they were a required in their charter to attach a private mortgage insurance policy to it.

The issue is the risk-taking these mortgage insurers took. These were private firms with exposure of first-dollar losses, and the original intention was to serve as gatekeepers to the housing market. They would protect the housing market from taking unreasonable risk, and in particular protect Fannie and Freddie from taking unreasonable risk.

What we show is that these firms were taking on a huge amount of risk during the boom years, especially in 2006-2007. And they were doing so in ways that exposed Fannie Mae and Freddie Mac to a great deal of losses. These insurance policies [are supposed to] protect Fannie Mae and Freddie Mac from the first set of losses up to the coverage amount of the insurance policy.

Knowledge at Wharton: Would you briefly explain mortgage insurance?

Keys: Mortgage insurance is an insurance contract that the beneficiary receives a payout if a borrower goes through foreclosure and there are losses accrued to the holder of this contract. Mortgage insurance comes in a few flavors and a few forms. The most common form, and the version that we look at in the paper, is a loan-by-loan ex-ante insurance contract, where the mortgage insurers can review the paperwork, just as the lender would review the paperwork, and decide whether or not to grant an insurance policy on the loan.

Now these policies are generally issued if a borrower doesn’t put down 20%, and there’s an equivalent insurance policy that the FHA requires, as well, but that is a government-provided insurance policy.

Knowledge at Wharton: What were the key findings in the paper?

Keys: Our findings are pretty striking. This is a market that’s under-studied, and my co-author, Neil Bhutta at the Federal Reserve did some fantastic data-digging around the Fed. You can think about the data that we used in this paper as collecting the equivalent of electronic dust. No one was using this data. We had millions upon millions of records of mortgage insurance applications — the universe of all mortgage insurance applications over a 20-year period. And with that data, we could see just how much the firms relaxed their lending standards during the housing boom.

We were able to show a huge increase in the number of applications approved, and a corresponding decrease in the number of applications denied, leading to a dramatic increase in the overall volume of these policies in 2006-2007. And the timing of this is really interesting, because this is exactly when the rest of the market is starting to say, “Hey, wait a minute. Some of these markets are looking very frothy — we’re thinking Las Vegas or Phoenix — after prices have doubled in the last three years. Maybe this isn’t the right time to be placing big bets on the housing market.” This is exactly when we show that the private insurers placed exactly these kinds of large bets.

Knowledge at Wharton: Why did they?

Keys: It’s hard to establish exactly why, and that’s one of the things that we grapple with in the paper. So first we were just trying to document a set of facts.

Getting at the “why” is harder. There may have been an expectation of a quid pro quo from Fannie Mae and Freddie Mac, but in terms of a direct bailout, that seems very unlikely. It seems like these were privately run firms, many of them publicly traded, so you could buy shares in these firms. You still can buy shares in these firms. They’re making independent decisions on pricing and approval and denial decisions, so they had quite a bit of freedom and quite a bit of control. And maybe they anticipated that Fannie Mae and Freddie Mac would somehow bail them out, but we weren’t able to find any direct evidence of that.

Knowledge at Wharton: But there were other, you call them “misplaced incentives,” where the mortgage insurer is going to get a lot of their premiums up front. And if there is a default, they don’t pay that until later.

Keys: That’s exactly right.

Knowledge at Wharton: Maybe they weren’t as worried about the later as they should have been, since the money was pouring in.

Keys: Exactly. That’s where the moral hazard story comes into play. They’re able to issue new policies and collect those premiums up front and not worry as much about what’s going to happen down the line. And in finance terms, this is sometimes called “gambling for resurrection,” where you’ll place big bets because you anticipate going out of business.

We show that they are already placing these bets well before they may have anticipated insolvency. They are decreasing their denial rates in the early 2000s, and in the mid-2000s — well before there was any sort of expectation that the whole market would collapse — but by late 2006 and early 2007, it’s a different story. The market is turning. Consensus forecasts are for a national downturn in house prices, and yet the mortgage insurers seem to be courting new business fairly aggressively, as their main competitors have disappeared.

So the main competitor to mortgage insurance like this is to take out a second lien. And usually these second liens would be taken out at the time of origination. This is a market that for most of history was just unavailable. No one would let you borrow for the down payment. The whole point was to save up for a down payment, and it’s only 2003-2006 that you see this market establish itself, when there’s a willingness of investors to bear this kind of risk.

“The big-picture conclusion is that putting private capital in a first loss position does not always discipline markets.”

During this period, you see a huge increase in the number of loans, especially in those high-cost markets where prices are rising sharply, of using a second lien at the time of origination. And this is known as a “piggy-back lien,” because you put it on the back of the other first lien. And this is a market that the banks were very involved in and private securitization markets were involved in. And they seemed — for at least a period of time — to be willing to bear that very high risk. Again, they are now in the first loss position.

By mid to late 2006, they’re no longer willing to write those kinds of loans. And so you see the banks withdraw from that market, and you see the private label securitization market withdraw from writing second liens. And so the folks who are looking for an affordable product, who don’t have a 20% down payment — especially in the expensive markets — they immediately turn to the private mortgage insurers.

Now if you thought that the private mortgage insurers were assessing risk correctly, there are two things that they would do. Number one, they would jack up prices immediately. They’d say, “Wow, we see the same risks that the other players in this market see.” And the other thing is that you would see a sharp increase in denials, because now all of the folks who had previously been applying for a second lien are coming to them — including some of the riskiest folks. And instead, we see a decrease in denials, and we see no change in their pricing.

Knowledge at Wharton: What are the conclusions?

Keys: The big-picture conclusion is that putting private capital in a first loss position does not always discipline markets. Disciplining risk is actually more complicated than simply saying, “Well, as long as there is an investor on the other side, everything will be fine.” And that’s a lesson that we learned the hard way from the housing crisis.

More broadly, there’s a deeper issue related to how we reform the U.S. mortgage finance system. And this is a debate that’s ongoing right now, as we have a new FHFA director, and we have a treasury secretary who’s planning on releasing a white paper proposing a reform of the mortgage market in the coming months. It has now been over 10 years since Fannie Mae and Freddie Mac were taken into conservatorship, and I think there’s still some ambiguity about what role the private market should play in the housing market.

We’re showing there should be a role for private capital in the mortgage finance system, but the incentives between those private players and any government entity needs to be very carefully explored, to make sure that incentives are aligned to actually keep a close eye on the amount of risk that’s percolating through the system.

Compare this to, say, auto insurance. If you’re an auto insurer, most car accidents are going to be idiosyncratic. There may be an accident on this highway this day, and a different highway the next day. So it’s fairly easy to think about building a portfolio of drivers — and some may be riskier, and some may be less risky. You charge them different rates based on their risk, and you assume, “Well, on any given day, every car in the country isn’t going to get into a car accident.”

Mortgage insurers didn’t fully appreciate that they’re not in the auto insurance business, and in fact prices throughout the entire country can fall at the same time. But the problem with that story of whether they were just ill-informed sort of cuts against what they knew by 2006.

By late 2006, the other players in the market had already left. Moody’s was already predicting a national downturn in prices. And so it wasn’t that they were completely clueless about the risks that they were facing. More generally, we need to spend a lot more time thinking about the overall risk exposure and the risk exposure that taxpayers ended up bearing in this space. Fannie and Freddie needed $187 billion of taxpayer funds to bail them out, and it looks like a lot of those funds went towards loans that had PMI attached in terms of the direct losses that Fannie and Freddie suffered.

“The bottom line is that the systems in place thus far to protect taxpayers from losses in the mortgage market are insufficient because of the incentive misalignment between the different players….”

Knowledge at Wharton: What would happen today under the same circumstances, given new regulations?

Keys: It’s hard to say. There have been a few changes that have affected the private mortgage insurers. And the biggest one is Fannie and Freddie’s private mortgage insurance eligibility requirements. And that’s a mouthful, but what that’s getting at is if you want to be on the approved insurer list that Fannie and Freddie use, you have to meet a certain set of capital requirements.

Now they’ve revised these twice since the downturn, and the most recent one didn’t seem to have much bite. So most of the insurers said, “We’re already in compliance with the capital requirements.” It’s hard to know if that’s just a function of the housing market doing well right now. We won’t know the answer to that until the housing market is tested again.

Knowledge at Wharton: What would be a stronger way to reduce this risk?

Keys: There are a few things that would be useful. One, encourage the private mortgage insurers to make publicly available stress tests and understand the risks that they’re bearing at the moment. That’s not something that they’ve made public at this point — at least to my knowledge.

The ones that are publicly traded reveal their financials to a degree, but they’re not doing the same type of stress testing that’s required under Dodd-Frank for the large banks.

There’s a sort of broader set of questions to be had about how they’re going to make decisions in a downturn. How do you provide credit during a downturn? When the market wants to freeze up, that’s exactly when you need to have some sort of lender of last resort. In the Great Recession, that was the FHA. So the Federal Housing Administration stepped in, dramatically expanded the amount of business they were doing, and they did that when the PMI firms were struggling severely under the weight of all the losses that they had taken on in 2006-2007.

If we do reform Fannie Mae and Freddie Mac, what’s the relationship going to be to the private mortgage insurance market? Are we going to have the same sort of mandate in place, and is there a way to stabilize that system such that it provides liquidity through a downturn?

Knowledge at Wharton: You wrote about a couple of other potential “fixes.”

Keys: The current model is subject to the same set of concerns we raise in our paper: The  mortgage insurers don’t necessarily have incentives that are aligned with that of protecting Fannie Mae and Freddie Mac, or protecting the taxpayer.

One way to deal with this — rather than providing ex-ante insurance up front when a loan is being originated — which is the system that we are focused on here, is to provide insurance or an insurance-like policy on the securities that Fannie Mae and Freddie Mac issue ex-post. That has been known as a credit risk transfer system or CRT. It’s something Fannie and Freddie have piloted in the wake of the crisis.

Knowledge at Wharton: Is it a kind of derivative?

Keys: Exactly. It is a derivative contract that functions like an insurance contract, and it provides a payout if a pool of loans doesn’t perform at a certain level. That seems to be a very popular product being purchased by hedge funds and other investors, but again, there’s a question of countercyclical liquidity. In a downturn, are those hedge funds going to want to invest in the housing market in that way? It seems very unlikely that we would have necessarily the same players being interested to buy that type of derivative at the going rates.

“It’s surprising that the industry hasn’t had to address the choices that they made in those years. There hasn’t been a mea culpa.”

One of the challenges for this other system — this sort of ex-post insurance, as opposed to ex-ante – is, how do you keep the players at the table even when the market turns downward.

Knowledge at Wharton: What is the bottom line of your research?

Keys: The bottom line is that the systems in place thus far to protect taxpayers from losses in the mortgage market are insufficient because of the incentive misalignment between the different players in the mortgage and insurance markets.

You have the lenders, on one hand, who don’t have much of a stake in the exposure of Fannie and Freddie to risk. You have the insurance companies that had misaligned incentives. They could generate policies very quickly and not necessarily worry about the consequences as much as they should have.

And you have Fannie Mae and Freddie Mac who perceive this government bailout on the horizon, and in that case were not very disciplined about their credit risk or their counterparty risk. So they weren’t monitoring the PMIs closely.

All this made for a system that was much more vulnerable to systemic risk than people understood. The big takeaway for us is that a lot of these tensions aren’t resolved.

As we look at some of the reform proposals on the table and that will be coming out in coming months, I’ll be very curious to see how they address these kinds of incentive misalignments.

At the end, you have the FHA serving as the backstop, which has exposed the government to a huge amount of potential losses. Right now we have a nearly nationalized mortgage system. Between the FHA, Fannie Mae, and Freddie Mac, they cover well over the majority of the exposure to mortgage losses, and that doesn’t seem very consistent with broader pushes to involve private capital, broader pushes to make the mortgage market act like a market that responds to supply and demand in traditional ways.

So bringing private capital into the system is tricky, and that’s what needs to be sorted out.

Knowledge at Wharton: It sounds like you were surprised to find some of the things that you discovered while doing this paper. Is that right?

Keys: Oh, absolutely. It jumped off the page at us. We were stunned just looking at some of the basic facts that we show in the paper — just to show the growth of mortgage insurance in 2007. This was a time when there wasn’t ambiguity about where the housing market was going. The folks within the industry were talking about the market cooling, and yet we see the mortgage insurance industry grow by 50% in 2007. It was the kind of thing where we had to double and triple-check our numbers to make sure that we weren’t doing something wrong. So yes, we were very surprised to see the volume of insurance policies being written in 2007, and that they were disproportionately being written in the riskiest markets like Phoenix and Las Vegas.

“It has now been over 10 years since Fannie Mae and Freddie Mac were taken into conservatorship, and I think there’s still some ambiguity about what role the private market should play in the housing market.”

It’s really surprising that the industry hasn’t had to address the choices that they made in those years. There hasn’t really been a mea culpa, and it’ll be interesting to see with some of the reform conversations going forward, they place smaller or larger opportunities on mortgage insurers.

There’s a version where Fannie and Freddie almost compete with the mortgage insurers for a certain type of business. There are other versions — a sort of government utility model, which almost removes mortgage insurers from the equation. There may be some scenarios where they’re going to have to fight pretty hard to keep a foothold in the mortgage finance system.

My hope is that they’ll spend a bit more time addressing why they made the choices that they did — in part because then we’ll have less ambiguity in our research, but also because I think for policy-makers going forward, we’ll have a much clearer sense of what exactly were the misaligned incentives and how can we fix them.