This site has moved to
The posts below are backup copies from the new site.

July 22, 2009

Economist's View - 5 new articles

"Ten Myths about the Subprime Crisis"

The Cleveland Fed's Yuliya Demyanyk says "most popular explanations for the subprime crisis turn out to be myths." I disagree on Myth 8, perhaps the crisis wasn't "totally" 100% unexpected, but it was generally unexpected and very few people got this right. As for Myth 10, I don't think anyone still believes that the "subprime mortgage market was too small to cause big problems," though that was believed at one time. Also, I'm not completely convinced that Myth 4 that "Declines in mortgage underwriting standards triggered the subprime crisis" is a myth, though that seems to be partly acknowledged in the discussion:

Ten Myths about Subprime Mortgages, by Yuliya Demyanyk: Subprime mortgages have been getting a lot of attention in the United States since 2000, when the number of subprime loans being originated and refinanced shot up rapidly. The attention intensified in 2007, when defaults on subprime loans began to skyrocket. Researchers, policymakers, and the public have tried to identify the factors that explained these defaults.

Unfortunately, many of the most popular explanations that have emerged for the subprime crisis are, to a large extent, myths. On close inspection, these explanations are not supported by empirical research.

Myth 1: Subprime mortgages went only to borrowers with impaired credit

Subprime mortgages went to all kinds of borrowers, not only to those with impaired credit. A loan can be labeled subprime not only because of the characteristics of the borrower it was originated for, but also because of the type of lender that originated it, features of the mortgage product itself, or how it was securitized.

Specifically, if a loan was given to a borrower with a low credit score or a history of delinquency or bankruptcy, lenders would most likely label it subprime. But mortgages could also be labeled subprime if they were originated by a lender specializing in high-cost loans—although not all high-cost loans are subprime. Also, unusual types of mortgages generally not available in the prime market, such as "2/28 hybrids," which switch to an adjustable interest rate after only two years of a fixed rate, would be labeled subprime even if they were given to borrowers with credit scores that were sufficiently high to qualify for prime mortgage loans.

The process of securitizing a loan could also affect its subprime designation. Many subprime mortgages were securitized and sold on the secondary market. Securitizers rank ordered pools of mortgages from the most to the least risky at the time of securitization, basing the ranking on a combination of several risk factors, such as credit score, loan-to-value and debt-to-income ratios, etc. The most risky pools would become a part of a subprime security. All the loans in that security would be labeled subprime, regardless of the borrowers' credit scores.

The myth that subprime loans went only to those with bad credit arises from overlooking the complexity of the subprime mortgage market and the fact that subprime mortgages are defined in a number of ways—not just by the credit quality of borrowers. One of the myth's byproducts is that examples of borrowers with good credit and subprime loans have been seen as evidence of foul play, generating accusations that such borrowers must have been steered unfairly and sometimes fraudulently into the subprime market.

Myth 2: Subprime mortgages promoted homeownership

The availability of subprime mortgages in the United States did not facilitate increased homeownership. Between 2000 and 2006, approximately one million borrowers took subprime mortgages to finance the purchase of their first home. These subprime loans did contribute to an increased level of homeownership in the country—at the time of mortgage origination. Unfortunately, many homebuyers with subprime loans defaulted within a couple of years of origination. The number of such defaults outweighs the number of first-time homebuyers with subprime mortgages.

Given that there were more defaults among all (not just first-time) homebuyers with subprime loans than there were first-time homebuyers with subprime loans, it is impossible to conclude that subprime mortgages promoted homeownership.

Myth 3: Declines in home values caused the subprime crisis in the United States

Researchers, policymakers, and the general public have noticed that a large number of mortgage defaults and foreclosures followed the decline in house prices. This observation resulted in a general belief that the crisis occurred because of declining home values.

The decline in home values only revealed the problems with subprime mortgages; it did not cause the defaults. Research shows that the quality of newly originated mortgages was worsening every year between 2001 and 2007; the crisis was brewing for many years before house prices even started slowing down. But because the housing boom allowed homeowners to refinance even the worst mortgages, we did not see this negative trend in loan quality for years preceding the crisis.

Myth 4: Declines in mortgage underwriting standards triggered the subprime crisis

An analysis of subprime mortgages shows that within the first year of origination, approximately 10 percent of the mortgages originated between 2001 and 2005 were delinquent or in default, and approximately 20 percent of the mortgages originated in 2006 and 2007 were delinquent or in default. This rapid jump in default rates was among the first signs of the beginning crisis.

If deteriorating underwriting standards explain this phenomenon, we would be able to observe a substantial loosening of the underwriting criteria between 2001–2005 and 2006–2007, periods between which the default rates doubled. The data, however, show no such change in standards.

Actually, the criteria that are associated with larger default rates, such as debt-to-income or loan-to-value ratios, were, on average, worsening a bit every year from 2001 to 2007, but the changes between the 2001–2005 and 2006–2007 periods were not sufficiently high to explain the near 100 percent increase in default rates for loans originated in these years.

Myth 5: Subprime mortgages failed because people used homes as ATMs

Rising house prices and falling mortgage interest rates before 2006 gave many homeowners an opportunity to refinance their mortgages and extract cash. The cash extracted from home equity could be spent for home improvements, bill payments, or general goods and services. Among subprime mortgages that were securitized, more than half were originated to refinance existing mortgages into larger ones and to take cash out of home equity.

While this option was popular throughout the subprime years (2001–2007), it was not a primary factor in causing the massive defaults and foreclosures that occurred after both home prices and interest rates reversed their paths. Mortgages that were originated for refinancing actually performed better than mortgages originated solely to buy a home (comparing mortgages of the same age and origination year). The rates of default for cash-out refinance mortgages within one year of origination were 17 percent for mortgages originated in 2006 and 20 percent for those originated in 2007. In contrast, the rates of default within one year of origination for mortgages originated to buy a home were 23 percent and 27 percent for the origination years 2006 and 2007, respectively.

Myth 6: Subprime mortgages failed because of mortgage rate resets

Among subprime loans, the most popular type of adjustable rate mortgage (ARM) is a hybrid, a loan whose interest rate is reset after an initial two- or three-year period of fixed rates. A fixed-rate mortgage (FRM), on the other hand, never has its rate reset. The belief that rate resets caused many subprime defaults has its origin in the statistical analyses of loan performance that were done on these two types of loans soon after the problems with subprime mortgages were coming to light. Those analyses compared loan performance in a way that was conventional at the time, but which turned out to be inappropriate for these loans.

To ascertain whether ARMs or FRMs were experiencing different levels of default, analysts compared the proportion of outstanding FRMs that were delinquent to the proportion of outstanding ARMs that were delinquent. Based on that comparison, the proportion of delinquent hybrid loans had begun to skyrocket after 2006, while that of fixed-rate loans looked as if it was fairly stable.

The problem with this type of analysis is that it hid problems with FRMs because it considered all outstanding loans; that is, it combined loans that had been originated in different years. Combining old with more recent loans influences the results, first, because older loans tend to perform better. Second, FRM loans were losing their popularity from 2001 to 2007, so fewer loans of this type were being originated every year. When newer loans were defaulting more than the older loans, any newer FRM defaults were hidden inside the large stock of older FRMs. By contrast, the ARM defaults were more visible inside the younger ARM stock.

To illustrate the problem, consider the following example. Suppose there are 1,000 FRMs and 100 ARMs outstanding in the market. In the current year, 100 new FRMs and 100 new ARMs are originated. Suppose the default rate for both types of new loans is 100 percent within a year and that old loans do not default. The observed default rate for FRMs is 100 out of 1,100 outstanding loans (9.1 percent), and the default rate for ARMs is 100 out of 200 outstanding loans (50 percent). Even though the level of default is the same for all new originations, the FRM pool looks much healthier.

If we compare the performance of adjustable- and fixed-rate loans by year of origination (which keeps new and old loans separate), we find that FRMs originated in 2006 and 2007 had 2.6 and 3.5 times more delinquent loans within one year of origination, respectively, than those originated in 2003. Likewise, ARMs originated in 2006 and 2007 had 2.3 times and 2.7 times more delinquent loans one year after origination, respectively, than those originated in 2003. In short, FRMs showed as many signs of distress as did ARMs. These signs for both types of mortgage were there at the same time; it is not correct to conclude that FRMs started facing larger foreclosure rates after the crisis was initiated by the ARMs.

Myth 7: Subprime borrowers with hybrid mortgages were offered (low) "teaser rates"

By design, a hybrid mortgage contract offers a fixed mortgage rate for a couple of years; after that, the rate is scheduled to reset once or twice a year to the current market rate plus a margin that is prespecified in the contract. A market rate combined with the margin may be lower or higher than the initial fixed mortgage rate, as it largely depends on the market rate that prevails at the reset time.

Hybrid mortgages were available both in prime and subprime mortgage markets, but at significantly different terms. Those in the prime market offered significantly lower introductory fixed rates, known as "teaser rates," compared to rates following the resets. People assumed that the initial rates for subprime loans were also just as low and they applied the same label to them—"teaser rates." We need to understand, though, that the initial rates offered to subprime hybrid borrowers may have been lower than they most likely would have been for the same borrowers had they taken a fixed-rate subprime mortgage, but they were definitely not low in absolute terms.

The average subprime hybrid mortgage rates at origination were in the 7.3–9.7 percent range for the years 2001–2007, compared to average prime hybrid mortgage rates at origination of around 2–3 percent. The subprime figures are hardly "teaser rates."

Myth 8: The subprime mortgage crisis in the United States was totally unexpected

Observing the extent of the subprime mortgage crisis in the United States and the global financial crisis that followed, it is hard to tell that this turmoil and its magnitude were anticipated by anyone. The data suggest, though, that some market participants were likely aware of an impending market correction.

In a market with rapidly rising prices, mortgage contracts that cannot be sustained can be terminated through prepayment or refinancing. Borrowers can change houses and mortgage contracts easily in a booming environment, and defaults do not occur as frequently as they would without the boom. Because of this ability to dispose of unsustainable mortgages, signs of the crisis brewing between 2001 and 2005 were hidden behind a "mask" of rising house prices. Using a statistical model to control for rising housing prices, Otto Van Hemert and I determined that default rates were increasing every year for six consecutive years before the crisis had shown any signs. This deterioration is observable now, with the help of hindsight and research findings, but it was also known to some extent to those who were securitizing subprime mortgages in those years. Securitizers seemed to have been adjusting mortgage interest rates to reflect this deterioration in loan quality. In short, lenders' expectations of the increasing risk of massive defaults among subprime borrowers were forming for years before the crisis; most likely, it was not the crisis that was unexpected, it was its timing and magnitude.

Myth 9: The subprime mortgage crisis in the United States is unique in its origins

The mortgage crisis in the United States is large and devastating, and it has led to global financial turmoil. In this sense, it is certainly unique. However, neither the origin of this crisis or the way it has played out was unique at all. In fact, it seems to have followed the classic lending boom-and-bust scenario that has been observed historically in many countries. In this scenario, a lending boom of a sizable magnitude leads to a lending-market collapse if it is associated with a deterioration in lending standards, an increase in the riskiness of loans, and a decrease in the price markup of said risk. Argentina in 1980, Chile in 1982, Sweden, Norway, and Finland in 1992, Mexico in 1994, and Thailand, Indonesia, and Korea in 1997 all experienced a pattern similar to the U.S. subprime boom-and-bust cycle. The United Stated has had similar episodes, though on a smaller scale, as well: a crisis with farm loans in the 1980s and one with commercial real estate loans in the 1990s.

Myth 10: The subprime mortgage market was too small to cause big problems

Before the crisis, there was a conventional belief that a market as relatively small as the U.S. subprime mortgage market (about 16 percent of all U.S. mortgage debt in 2008) could not cause significant problems in wider arenas even if it were to crash completely. However, we now see a severe ongoing crisis—a crisis that has affected the real economies of many countries in the world, causing recessions, banking and financial turmoil, and a credit crunch—radiating out from failures in the subprime market. Why is it so?

The answer lies in the complexity of the market for the securities that were derived from subprime mortgages. Not only were the securities traded directly, they were also repackaged to create more complicated financial instruments (derivatives), such as collateralized debt obligations. The derivatives were again split into various tranches, repackaged, re-split and repackaged again many times over. This, most likely, was one of the mechanisms that amplified problems in the subprime securitized market, and the subsequent subprime-related losses. Each stage of the securitization process increased the leverage financial institutions were taking on (as they were purchasing the securities and derivatives with borrowed money) and made it more difficult to value their holdings of those financial instruments. With the growing leverage and inability to value the securities, uncertainty about the solvency of a number of large financial firms grew.


Many of the myths presented here single out some characteristic of subprime loans, subprime borrowers, or the economic circumstances in which those loans were made as the cause of the crisis. All these factors are certainly important for borrowers with subprime mortgages in terms of their ability to keep their homes and make regular mortgage payments. A borrower with better credit characteristics, a steady job, a loan with a low interest rate, and a home whose value keeps increasing is much less likely to default on a mortgage than a borrower with everything in reverse.

But the causes of the subprime mortgage crisis and its magnitude were more complicated than mortgage interest rate resets, declining underwriting standards, or declining home values. The crisis had been building for years before showing any signs. It was feeding off the lending, securitization, leveraging, and housing booms.

Recommended Reading

"Understanding the Subprime Mortgage Crisis," by Yuliya Demyanyk and Otto Van Hemert. Forthcoming 2008. The Review of Financial Studies.

"Quick Exits of Subprime Mortgages," by Yuliya Demyanyk. 2009. St. Louis Review 91:2 (March/April), pp. 79–93.

As a user, you are authorized to reproduce Federal Reserve Bank of Cleveland publications, or portions thereof, appearing on this website, without limitation as to number...

Commercial Real Estate "Does Appear to be Headed Further South"

Calculated Risk has me convinced that this is something we ought to be worried about:

Bernanke: CRE May Pose Risk, by Calculated Risk: From Bloomberg: Bernanke Says Commercial Property May Pose Risk for Economy

Federal Reserve Chairman Ben S. Bernanke said a potential wave of defaults in commercial real estate may present a "difficult" challenge for the economy, without committing to additional steps to aid the market. ...

It "may be appropriate" for the government and Congress to consider "fiscal" steps to support the industry, Bernanke said today. Ideas for fresh support for the market could include government guarantees for commercial mortgages, Bernanke also said today ...

"As the recession's gotten worse in the last six months or so, we're seeing increased vacancy, declining rents, falling prices -- and so, more pressure on commercial real estate," Bernanke said yesterday. "We are somewhat concerned about that sector and are paying very close attention to it. We're taking the steps that we can through the banking system and through the securitization markets to try to address it."

A few key CRE stories this month...[list of news items] ...And a comment from the USG (building materials supplier) conference call this morning:

"Nonresidential construction does appear to be headed further south, perhaps significantly so."

No kidding.

Fed Independence

Continuing the recent discussion here and elsewhere on Fed independence, this is not the first time audits and other threats to independence have been seriously considered:

On the mend, The Economist: It has been a long time since comments on the economy by an official of America's Federal Reserve comments could be described as cheerful. Yet there was no denying the upbeat tone of Ben Bernanke's testimony to Congress on Tuesday... His fingers may be crossed but it is clear that Mr. Bernanke thinks the recession, if not over now, soon will be.

That is a far cry, though, from seeing a threat from inflation and Mr. Bernanke made it clear that the federal funds target rate, now near zero, will remain there for a long time. On Wall Street, most reckon that means until well into 2010 at least.

Yet the Fed is already under pressure to explain how it intends to tighten monetary policy, even by congressmen who usually want nothing of the sort. ... Whether inflation [occurs] depends on if the Fed raises interest rates in time and thereby curbs the appetite for credit. Mr. Bernanke spent much of his testimony explaining how he can do just that. ...

Politics could ... interfere with the Fed's willingness to tighten monetary policy in time. Congress's nonpartisan investigative arm, the Government Accountability Office, can now audit the Fed with the exception of its monetary policy, lending programs or relations with foreign central banks. A bill in Congress would lift those prohibitions. Mr. Bernanke argued that the threat of such audits would lead investors to question the Fed's willingness to do unpopular things, like tighten monetary policy, unsettling them and driving up long-term interest rates.

This is not idle speculation. Anti-Fed sentiment was also strong in the 1970s, when Congress first sought to have the GAO audit the central bank. Arthur Burns, chairman at the time, fought back, and a compromise was struck to allow audits, but with the current prohibitions. Mr. Burns later reflected that the effort of "warding off legislation that could destroy any hope of ending inflation" involved "political judgments" that may have weakened his anti-inflationary resolve.

For all the discussion, any tightening of policy is a long way off. ...

I think one of the problems that people are trying to get at when they want to take away the Fed's independence is the concentration of power within the Board of Governors (the view by some that the Board represents special rather than public interests, e.g. Wall street, also plays a role), and they see devices such as audits from the GAO as a check on that concentration of power. Here's an edited version of part of an old post:

While the Fed was initially structured to balance competing interests and to share power, the system has evolved into an institution with centralized rather than shared power. The intent to share power and balance competing interests is evident in the structure of the Federal Reserve system. For example, individual district banks are overseen by a board of nine part-time directors. These directors come in three types. Three of the nine are type A and are bankers, and three are type B and represent the business community. Legislation prohibits type B board members from being bankers. In a further attempt to make the process representative, type A and type B directors representing banking and business interests are elected by member banks within each of the twelve Federal Reserve districts. Type C directors are appointed by the Board of governors and are intended to represent the public interest within the district banks.

Thus, the districts themselves provide geographic representation that is population based, while control of the district banks balances public, banking, and business interests. Initially, the district banks functioned as twelve cooperating banks and each district had considerable control of monetary conditions within the district. It was very much a shared power arrangement. As one example of the power district banks had, each bank had full control of the discount rate for its district (the discount rate was the only tool available for controlling the money supply when the Fed was formed, open-market operations were not well understood until later and there was no provision for the Federal Reserve system to control reserve requirements, another way to affect the money supply).

The shared power arrangement within the Federal Reserve system changed after the Great Depression when monetary authorities failed to respond adequately to crisis condition. The problem, or so it seemed, was the shared power nature of the system. The deliberative, democratic nature of the institution prevented it from taking quick, decisive action when it was most needed. Furthermore, the Fed did not have the tools it needed to deal with system-wide disturbances, it was mostly equipped to deal with problems at individual banks (the discount window is well-suited to help individual banks, but not system wide disruptions; on the other hand, open-market operations can inject reserves system-wide and hence is a better tool for systemic problems).

The solution to this was to concentrate power into the hands of the central bank so that should a crisis occur, they can act quickly. There are risks, of course, to concentrating power so narrowly, but in the aftermath of the Depression we were quite willing to take that risk if it helped to avoid another catastrophic outcome (as it may well have done).

Thus, after the Great Depression power was concentrated. For example, banks no longer control the discount rate in their districts. They can propose a change in the discount rate at an FOMC meeting, but the Board of Governors must approve the rate and they will only approve one rate, the rate they decide. So while the rates are still formally set in the districts, they are essentially set by the Board of Governors. When all such changes in the concentration of power over time from the districts to the Central Bank in Washington D.C. are considered, it becomes very clear that the Fed has evolved from a very democratic, shared power arrangement at its inception to one where it functions, for all intents an purposes, as a single bank in Washington, D.C,. with twelve branches spread across the U.S.


I am not at all in favor of lessening the degree of independence that the Fed currently has, but I do think we need to make changes in the way the President and Boards of the district banks are chosen. As it stands, the Board of Governors in Washington has considerable influence over who is appointed to key positions such as the President of the district banks, and those Presidents represent five of the twelve votes at the meetings where monetary policy is set. More independence of the district bank Presidents and other district bank personnel from the Board of Governors would be a healthy change (there is also a question of whether geographic representation through district banks is the best way to capture the public interest, but I'll leave that aside for now).

"Two Ideas for Appraisal Reform"

These seem like reasonable ideas to me:

Two ideas for appraisal reform, by Richard Green: Lawrence Yun of NAR is complaining that appraisals are preventing legitimate real estate transactions from occurring. Because of the way appraisers sometimes choose comparables, I have some sympathy for this view. And as I noted in an earlier post, Rhonda Porter says the Home Value Code of Conduct is nothing more than a way to line the pockets of Appraisal Management Companies. I have some sympathy for this view as well.

But we should not go back to the days when appraisers were basically paid to stay out of the way of the consummation of a deal. So let me suggest two proposals:

(1) Appraisers should not be allowed to see the offer price of a house. This is the only way their valuation will be truly independent.

(2) Appraisers should use valuation techniques that allow them to report a standard deviation of their estimate. Subdivision tract houses will have small standard deviations; architect designed villas will have large standard deviations.

We could then move to a pricing rule where Mortgage Insurance will be required if (1) the LTV based on appraised value is greater than 80 percent or (2) there is a greater than five percent chance that the true value of the house implies an LTV of 95 percent.

Step (1) would be easy to implement, and I think would help a lot. Step (2) will require lots of training (and perhaps different parameters from those that I am suggesting).

We need to stop kidding ourselves that we can measure house prices precisely. We need to start measuring the level of imprecision.

links for 2009-07-22

No comments: