Archive for June, 2008

Credit Market Problems and Banks and Investment Banks

Wednesday, June 25th, 2008

As long as the real estate market was doing well (home prices rising, borrowers paying their loans) the banks and investment banks that securitized the subprime mortgages did quite well. But when prices of homes stopped rising problems started to arise.

Not all mortgage bonds are alike, so there are many varieties of stories about how the problems occurred and how they spread through the system. For the purposes of this essay we will analyze the story of a mortgage bond that was more or less legitimate, meaning that most of the parties to the transactions that produced the bond were relatively honest. There were numerous cases of improper behavior in the bubble period: buyers who misrepresented their financial position, mortgage brokers who did not fully explain the nature of the mortgage and so forth. While I know of few careful measurements of the degree of these problems, there are many anecdotes that indicate that these problems were quite serious. But these phony mortgages do not change the way that the problem unfolds, they merely change the ultimate cost of the problem, so I will set them aside for now.

To start, we can examine a specific bond, the Wells Fargo Home Equity Asset-Backed Certificates, Series 2005-1 (see here for information, registration may be required).

This security represents (see Appendix A, page A-1 and following pages) 7,512 subprime mortgages for a total of $1.2 Billion, from all 50 states and the District of Columbia, with interest rates ranging from 4% to 12%, 15 to 30 years time to maturity, in single family homes and condominiums, loan to value ratios from 50% to 100%, principal from $50,000 to $900,000, both original loans and refinancing, primary home and secondary home and with and without prepayment penalty. To properly analyze this security one would have to understand the real estate market in all the states for virtually every kind of loan imaginable; it is said that estimating the value of securities like this takes a specialist (of whom there are very few) a full week.

There were several types of credit enhancement, including overcollateralization and excess interest. The mortgage pool was divided up into a number of slices or tranches, with around $1.05 Billion in three A tranches, and a number of subordinate tranches from M1 to M14.

The A tranches were rated AAA by all three credit agencies (see page S5) and the other tranches received lower ratings (an article describing the Fitch credit ratings on June 29, 2005 can be seen here). More recently, Fitch maintained the AAA ratings on the A tranches as well as the ratings on tranches down to M5; but the news was not so good for tranches M6 and below (see here).

The news for holders of this particular bond was not that bad, at least for holders of the more senior tranches (the A tranches were recently trading at around 90% of face value). But many other similar bonds (as well as the more junior tranches of this bond) had problems, and these problems have spread throughout the banking and investment banking world.

To better understand these problems you may want to spend a few hours reading this summary [pdf] from UBS to explain how they lost $18.7 Billion during 2007, despite the best efforts of management and the existence of a risk control system. I have selected UBS partially as a result of their own public self-analysis, not because it is somehow worse than other banks (although the Chairman himself recently noted [pdf] that “that there are banks which were more cautious and did some things better than UBS.”

If you do not have a few hours, here is a brief summary with a little commentary.

Short version: UBS jumped into the subprime business at the worst possible time, lost a lot of money and now has a new chairman and CEO.

Slightly longer version: in June 2005, UBS set up a new group to invest in hedge funds, DRCM (Dillon Read Capital Management, see an article in the FT here). UBS had been losing both employees and clients to hedge funds and wanted to have a hedge fund of their own. At first, this fund invested UBS’s own money, but it was opened to outside investors in November 2006.

The reason for this was that UBS had been lagging behind their competitors; an external consultant recommended that UBS develop, among other areas, its MBS Subprime business. Their new division, DRCM, quickly started losing money on several aspects of this business, notably risky subprime debt whose value fell more than they had hedged; unfortunately, at that point “[t]he assets could not be sold given the illiquidity in the market and UBS was exposed to further deterioration.” (page 12).

UBS also lost money on (1) their CDO warehouse; UBS would buy mortgages and hold them until they could be packaged and resold; (2) their Super Senior Positions, the AAA tranches of some of the CDOs they sold to clients. The AAA tranches seemed to yield more than other AAA securities with little additional risk; unfortunately, not all these assets were adequately hedged, so when the value of these assets fell UBS suffered large losses compared to the other activities listed here; (3) trading; UBS bet on short-term price moves in mortgage securities and also lost (relatively little) here.

Six months after it opened to outside investors (May, 3 2007), UBS announced the closure of DRCM (see another FT article here). As I write (almost a year later) UBS is still calculating the losses for 2007 and trying to figure out how to move forward.

The problems at UBS are probably typical of those at other large financial institutions. Money was being made in the subprime market and banks tried to get a larger share. When the housing market slowed down, and the profits turned to losses, the first instinct of a number of some banks was to exit the business as soon as possible. The exit of a number of participants in the market exacerbated the problem. As was noted above, valuing a specific subprime bond is not like valuing a share of stock or a government bond. If you want to know what a share of Microsoft is worth (or a specific Treasury Bond) you can see how Microsoft or the Bond is trading today. If you have a security that is similar (but not identical) to another, more heavily traded security, you can estimate the price of your security from the more frequently traded securities. But the subprime bonds were not always similar to other traded securities. As we saw above for the Well Fargo bond, each bond is a claim against a specific portfolio of mortgages.

So the problem for the banks was that the market for these securities was drying up (banks exiting the industry, dumping their portfolios, those remaining unwilling to increase their positions) at a time when the models may have indicated that the bonds were more valuable than other market participants were willing to pay for them. Pressure from investors and regulators caused banks to “mark to market” instead of “mark to model” meaning that the bank would find out what other banks would pay for a bond and carry it on their books at that value, recognizing the loss on their financial reports.

By recognizing these losses some highly leveraged firms came under great pressure. Banks are required by international rules (the Basel II Agreement) to keep a certain level of capital relative to assets; if they experience large losses on subprime debt then they may be forced to sell off other assets; but a number of banks found new investors to add capital (e.g., Citibank and Abu Dhabi) to ensure that they would have sufficient capital.

Some banks with SIV programs (and a mortgage warehouse) had a further complication. By contracting with mortgage originators, they were obligated to buy mortgages for the purposes of reselling them through a SIV. But the demand for the SIV dropped and the costs of financing the SIV in the asset backed commercial paper market increased (see here for ABS commercial paper rates). So SIVs became a burden on banks, being forced to buy mortgages that they could not sell and were expensive to finance.

There were further problems for the end customers who bought some of these securities. Some held this debt in money market mutual funds, that normally fix the value of the fund at $1 per share and provide a return in in the form of interest. When the price of these securities fell, firms were faced with the possibility of “breaking the buck” or allowing the value of the money market fund to drop below $1, which is viewed as breaking an implied contract with people who invest in the fund, and some parent companies had to add extra funds to their money market divisions to preserve the value of the fund.

Foreclosure: What happens when people can’t pay their mortgages

Wednesday, June 25th, 2008

How much does a lender lose when a homeowner stops paying a mortgage? Approximately half the value of the mortgage, assuming there was no fraud. This essay will sketch out how the foreclosure process works in the normal case and in the less common case of fraud. The argument presented below is a simplification of the actual process that I believe captures the important aspects; if you have any questions or find any errors please contact me here.

The typical story (at least in the popular media, see here or just google for foreclosure) of a homeowner who cannot pay his mortgage goes something like this: an otherwise model citizen (let us call him Mr. Smith) goes to the bank to borrow money to buy a house. Mr. Smith does not carefully examine the paperwork that he signs when he gets the mortgage. Perhaps someone at the closing told Mr. Smith what a mortgage lender once told me when I was examining mortgage documents, about the “Golden Rule”: he who has the gold makes the rule. In other words, sign the papers now or cancel the mortgage and come back on another day. (When one reads a bit more about Mr. Smith it might occur that he has just bought several new cars and taken an exotic vacation, or that he was “persuaded” by his banker to exaggerate his income in order to qualify for a mortgage on a much more expensive house than he might have otherwise been able to afford; in many cases, the blame for the poor choice of mortgage can be apportioned across a number of actors.)

In any event, Mr. Smith finds that after a year or two of low, affordable payments, his payments are now much higher. Mr. Smith apparently agreed to an option ARM (such as this one) that allowed him to make small payments but now demands a payment that he is unable to make. Mr. Smith has very little equity in his house (perhaps he took out a 95% mortgage or even more, and his first few payments were interest-only). After Mr. Smith misses each payment the bank sends him a letter, and after three or four missed payments the bank gets a court to declare that Mr. Smith is in default and the bank now has the right to evict Mr. Smith and sell the house. Mr. Smith may try to sell his house, but finds that the amount that he would net from a sale is somewhat less than he would owe the bank; part of this gap may be due to an inflated price paid by Mr. Smith, part due to a drop in prices since Mr. Smith purchased his home).

(Years ago, a local banker might have visited Mr. Smith and tried to work out a program to help him; today, Mr. Smith’s loan might have been sold several times and he may be paying a servicing agent in another state with whom he had no previous relation. Again, a personal anecdote; once I had a mortgage that was resold once or twice; the eventual owner of my loan called up to offer to refinance my loan at a lower rate. But after a few questions is turned out that this lender did not make the kind of loan that I had. I eventually refinanced with another lender, but if I had been having financial difficulties it may have been impossible for that lender, who had bought my loan, to renegotiate the conditions of my loan.)

One possibility (before the bank forecloses) is for Mr. Smith to arrange a “short-sale”, in which he sells the house to a buyer for less than he owes the bank. The bank must establish that Mr. Smith has made a good faith effort to sell his home and that he actually is in a difficult financial position not as a result of his own excessive behavior. A short-sale is difficult if Mr. Smith has multiple mortgages against the house, because the holder of the subordinate claim (the “second” or junior mortgage) will not receive anything until the first mortgage is paid off. If both mortgages were from the same lender, then there is no conflict. But if multiple lenders are involved it is hard to solve this problem.

The advantage (for the bank) of a short-sale is that the home is likely to remain in good condition and the bank will likely recover something like 70% or more of the mortgage. However, if a short-sale is not possible, the bank must take ownership of the property (foreclosure) and sell it.

Once the bank owns the house it must pay the relevant expenses (taxes, condo fees, lawn care fees and so forth), but it may take some time for the bank to remove Mr. Smith from the house. If Mr. Smith does not leave voluntarily, the bank can offer him (through an intermediary) “cash for keys” (see a Wall St. Journal article here) a modest payment (perhaps $1000 or so) in return for voluntarily leaving the property in good condition. The property often must be cleaned up (and sometimes be repaired or have certain fixtures replaced) in order to be sold. There will be additional time to market the property and complete the transaction.

Depending upon the local market conditions and the nature of the home the bank should receive around 50 to 60% of the mortgage around a year after the first missed payment.

As noted above, the case of fraud can further reduce the return for the bank. In a typical fraud case the property will be flipped several times between buyers, often with the cooperation of an appraiser, a bank loan officer or an underwriter. Mr. Jones may buy a house for $100,000, sell it to Ms. Johnson for $200,000, who in turn sells it again to Mr. Jones for $300,000 (without much in the way of involvement of outside parties).

Mr. Jones will then go to the bank finance this purchase and borrow $250,000 which would appear to be a conservative loan against a $300,000 house. The money will be divided between Jones, Johnson and their confederates. No payment on the mortgage is ever made. The bank forecloses and winds up with a $100,000 home (if even that price was a fair value) that must be sold and will not be sufficient to cover the 250,000 mortgage, so the bank’s return could be as low as 10% or 20% of the mortgage amount.

Why GE stock fell so much on Friday April 11

Tuesday, June 24th, 2008

Last Friday (April 11), General Electric Corp (GE) announced that quarterly earnings were $0.43 per share, or about $4.3 billion. This would seem to be fairly good news; the number was down a bit from profits a year ago, when quarterly earnings were $0.44. Yet the \stock price dropped by almost 13%; largely in response to this news, the rest of the market dropped by 2% (see here for a typical newspaper story).

Why does a stock fall so sharply on what would seem to be not so bad news? The answer has to do with the market’s expectations for GE, and GE’s past aggressive use of a tactic known as “earnings management”. (Disclaimer: I am not an accountant and have never actually analyzed GE’s accounting statements. I own a bit of GE stock through investments in broad market indices. This essay, like all essays on this site is not to be construed as a recommendation to buy or sell any particular security but is intended to increase your understanding of markets to help you make better investment decisions.)

The market expected GE to earn around $0.51 per share, and GE had indicated that they expected to earn between $0.50 and $0.53 per share. From this perspective, the result of $0.43 was a big disappointment to the market, especially for GE, which rarely disappointed the market with earnings announcements.

GE has long been accused of “earnings management”, that is arranging its business transactions in a way to produce continuously improving accounting results (see here for more on the topic). There are perfectly legal methods that a firm can use to ensure that earnings meet a particular target. For example, a firm whose income is a bit less than expected can delay some expenses until the start of the next quarter. If income is unexpectedly high in another quarter, the firm might realize some expenses earlier than necessary. Both these transactions will smooth out quarterly earnings.
As long as the amounts involved are relatively small, and the company is growing profitably, there is little cost associated with such transactions.

But once in a while a firm cannot meet the earnings target; perhaps all the news toward the end of the quarter was bad. In this case the company has to make a decision: should they make a heroic effort and still miss the market’s expectation by a small amount? Or should they admit as much negative news as possible, missing the expectation by a large amount, but setting the stage for positive developments in future quarters?

GE announced a number far below the market’s expectation and the market reacted strongly, because of GE’s history of meeting (or beating) the quarterly earnings expectations. But analysts more familiar with GE than I am will be able to decode the accounting statements and decide whether GE made all possible efforts and announced the best possible result (in which case the outlook for the future is very gloomy), made minimal efforts and announced the worst possible result (in which case the outlook for coming quarters should be fairly bright) or was somewhere in between.

The general lesson is that news does not have any particular significance outside the context of market expectations. Most people outside financial markets would have thought that earning $4.3 billion in three months represented a reasonable performance; only when you understand the specific history of GE’s alleged practices can you begin to understand why the market might react so strongly to such a number.

Why Eliot Spitzer was investigated

Tuesday, June 24th, 2008

Eliot Spitzer, the governor of the state of New York (and former New York State Attorney General), resigned on March 13 as a result of his involvement with an expensive prostitution service. Many commentators noted the hypocrisy of a former prosecutor, who investigated prostitution services, losing his job as a result of patronizing such a service. But the reason that the governor was investigated was not because he was found to be visiting prostitutes, but because of the way that he paid for the service.

There are laws in the U.S. that require the reporting of cash withdrawals and deposits of $10,000 or more, designed to discover those who might be earning money from illegal activities. These forms are not very complicated, and pose modest costs for businesses and ordinary citizens who need cash for legitimate reasons. But every law provokes a reaction; in this case, clever people would make transactions of somewhat less than $10,000. So a new crime was created, called “structuring” (see here), which approximately implies that it is also illegal to make repeated transactions less than $10,000 designed to avoid the $10,000 rule (note: I am not a tax attorney and none of the above is intended to represent legal advice; I am merely a market analyst trying to understand the situation and explain it to others).

As a result of the structuring rule, banks are required to report wire transfers that might appear to be “structured” transactions. Unfortunately for ex-Governor Spitzer, he made repeated wire transfers to pay for the services he was purchasing to a company that was somewhat suspicious. The company did not appear to have an office at its official address. Investigators wondered why a governor would be making payments to a dubious corporation…could he be paying off a blackmailer?

By monitoring the governor’s telephone calls and text messages, investigators learned exactly what the governor was buying with his payments, and the investigation shifted to prosecuting the providers of the prostitution service; ultimately the governor was probably only Client-9 in a complicated case against the four people who ran the service.

The governor was ultimately forced to resign as a consequence of his hypocrisy, having prosecuted the businesses of which he was a customer. But if he had chosen a more moderately priced service, and not paid with money transfers, he would likely still be the governor.

How do traders lose so much money?

Tuesday, June 24th, 2008

Every few years there is a story of a trader who loses huge amounts of money; most recently, Jerome Kerviel at Soc Gen, but there were many before him. The magnitude of these losses is a great mystery: how can one person lose Euro 5 billion? Didn’t someone realize that there was a problem when M. Kerviel lost his first billion? I believe that there is an answer to this question that lies at the heart of financial regulation.

Traders are similar to professional football players. The best football players will help their club earn championships and in turn will be paid huge salaries. There is tremendous competition to become a top player, and players will use every available technology (such as equipment and training) to become faster and more powerful. Referees are not so highly rewarded. While a top referee may make a good salary relative to other professions, the reward is nothing like that of a football player. Few spectators decide whether or not to attend a game on the basis of who will be the referee; while many will argue at great length about the quality of the decisions the referee made, referee errors are considered a part of the game, and clever players will exploit a referee’s shortcomings. Often one gets the sense that referees fail to accurately make decisions in part because they are not at the same talent level as the players on the field.

Traders, instead of scoring touchdowns on the field, take risky positions in order to make profits. They adopt complex trading strategies combining computer models and their own intuition. But banks place limits on the traders, to minimize the chance of a large loss. The risk managers can be seen as referees, enforcing the bank’s risk limits. Risk managers, like referees, are seen as a necessary burden. The risk manager is often viewed as a cost-center, bringing no added profit to the firm.

While we are likely never to know the full story of M. Kerviel (nor that of Mr. Leeson, the trader who lost so much money at Barings Bank), the outline is fairly clear. Kerviel knew the “back office” systems fairly well, having started his career in the part of the bank that accounts for traders’ activities. Kerviel was supposed to make small bets, but wanted to make larger bets to show his trading skills to his superiors. He used his knowledge of the bank’s accounting systems to conceal the size of his positions, allegedly avoiding the normal control procedures. Whatever the details, it is quite clear that Kerviel was a step faster than the referees, who were unwilling or unable to restrain his behavior until he had large losses, which were subsequently multiplied as the bank may have had to quickly sell off his positions at a greater loss.

There is no obvious solution to either the referee problem or the trader problem. Football leagues are willing to spend some money to improve the quality of referees, but only just enough to keep fans watching, and not enough to have perfectly refereed games. Every few years an important game will be decided by what appears to be an incorrect decision by a referee and some fans will pay less attention to that sport in the following years.

Banks need to allow creative traders to trade, otherwise they would lose these employees to other firms (note how many traders have already left traditional financial institutions for private equity and hedge funds). The risk is that every few years a financial institution will disappear, a risk that many institutions seem to think is worth taking.

Understanding Real Estate Auctions

Tuesday, June 24th, 2008

One of the consequences of mortgage defaults is that banks (and other lenders) end up owning properties. Once a bank owns a property it is immediately responsible for all payments associated with the property (taxes, maintenance fees and so forth). Banks have a strong incentive to sell the property as soon as possible. One way that such properties are marketed is through real estate auctions (some homeowners may sell their homes at auction for other reasons as well).

There are two major systems for real estate auctions in the US, one government and one private. The government auctions (see here) are designed to sell houses that the government owns (the government owns these houses when people borrow money from government agencies such as the HUD or VA do not pay their mortgages and the government agency forecloses their home).

Homes sold by the U.S. government are listed on their web site. The government encourages buyers to work with a real estate agent who can help the buyer with the bidding process which differs by agency; HUD accepts the highest “reasonable” bid offered during a fixed period, but there are also special discounts for people such as teachers and police officers who buy houses in certain areas. Once the government accepts your offer the transaction is more or less like any other home sale (the seller pays the commission to real estate agents, the usual policy in the U.S.). The government seems to try to make their auctions as much as possible like buying a house from an ordinary seller; in the U.S. normally (perhaps before the current crisis) a house would be listed for sale for a few weeks, the real estate agent would then present the offers by purchasers to the seller, and then the closing (final transfer of the house from buyer to seller) might happen a month later. So a transaction would take about two months from the initial listing of the home until the sale. Of course if the house is viewed as priced too high then it could take much longer and the house might be listed for many months before it is sold.

Private auctions exist in many forms, from online sites such as eBay to traditional auction companies (see, among many others, the site of Elite Auctions in California or the NAA auction site).

The NAA (National Auction Association) estimates that there were about $16 billion in real estate auction proceeds in 2007 (see here). If the average US house price is around $200,000 (a very rough estimate) that probably means that there were around 80,000 or so homes sold by auction last year. This is roughly 1.5% of the 5.5 million sales of existing homes last year.

Auctions, like most advertising, are shifting from newspapers to the internet. If you Google “real estate auction” you will get over 500,000 hits.There is a perception in the market that if a bank acquires very nice houses that they might sell them through real estate agents and only less desirable houses will be sold via auction. To protect sellers, there is often a reserve price, below which the house will not be sold. Buyers are at some risk of paying too much, which tends to keep prices low. There are companies that help you bid at auction (such as this one).

Buying a house at auction is a bit riskier to the buyer than buying a house directly from the seller. When you buy a house in the U.S. you traditionally have the right to have an independent inspector evaluate the home and after the inspection you can negotiate with the seller to fix various problems in the house or get a reduction in the price. If you go to an auction you may be interested in 5 houses and it would be very expensive to have inspections done on all five.

In the U.S. banks tend to pre-qualify home buyers; a home buyer would go to the bank before the auction and be approved for a mortgage of a particular amount at a particular rate (depending on income and credit score and other factors). Banks appear to treat homes purchased at auction more or less like other homes. Home auctions increase when foreclosures are high, and foreclosures are rising rapidly due to the subprime mortgage crisis. As home prices fall and the economy slows, there is increasing pressure on homeowners who have large mortgages relative to the value of their property (remember that in the U.S. normal mortgages are 80% of value and at the height of the housing bubble there were banks who loaned more than 100% of the value of the home). If a homeowner has any serious trouble (loss of a job or bad health) and it is impossible to sell the home for more than the mortgage then there is an increasing chance the home could end up at auction.

How the Housing Credit Bubble got started

Tuesday, June 24th, 2008

The mortgage business was traditionally fairly conservative with a local savings and loan lending to a resident of their community. Bankers typically knew their markets from personal experience and often knew the borrowers personally (see the film It’s a Wonderful Life for the Hollywood version).

The major factor that transformed the housing boom into a credit bubble was disintermediation in the credit markets. Former Fed Reserve Chairman Alan Greenspan has argued (see his article in the Financial Times blog here) that there were real estate bubbles in over 20 countries in the first half of this decade (from 2001 to 2005); the common factor seems to be that long-term interest rates declined in all major countries at the same time. Greenspan argues that the housing price bubble in the U.S. was not different from the housing bubble in the other countries. What was different in the U.S. was the credit bubble, from the securitization of subprime loans.

In this essay I will describe how the mortgage market works, the development of securitization and then how things went wrong.

While the first savings institutions in the U.S. date back to the early 19th century, the savings and loan industry was greatly expanded by the Federal Home Loan Back Act of 1932 (see here for the text of the Act). In the 1950s and 1960s, a borrower would get a mortgage at a savings and loan or a commercial bank, which would then hold the mortgage on its books.

Here is a table derived from the Federal Reserve Flow of Funds (from Table F218) that shows the amount of home mortgages made in selected years and the amount of the change in home mortgage assets of the banking system.

Year Home Mortgages From Banks
1955 12.6 9.1
1960 11.3 7.8
1965 17.1 13.3
1970 13.1 8.1
1975 38.8 26.4
1980 92.6 38.9
1985 178.6 52.7
1990 206.7 1.8
1995 167.6 65.8
2000 427.0 145.9
2005 1060.7 324.0

Billions of Dollars; Banks include commercial banks, savings institutions and credit unions.

In the table you can see that from 1955 to 1975, the largest source of mortgage finance was the banking system. Other holders of mortgages included the household sector (often mortgages provided by the seller of the property), life insurance companies, finance companies and the government sponsored entities (GSEs).

By the 1980s a greater fraction of loans were held outside the banking system, as more mortgages were sold to Fannie Mae and other GSEs. In 1985 Agency and GSE-backed mortgage pools accounted for $77.6 Billion of the $178.6 Billion home mortgages, compared to $52.7 Billion for banks.

GSEs such as Fannie Mae (the Federal National Mortgage Association or FNMA) purchase mortgages from mortgage issuers and then resell them in the form of Mortgage Backed Securities (MBS) that are packaged in ways attractive to institutional investors. There is a limit to the size of the mortgages that they can purchase from mortgage issuers; until recently the maximum mortgage was $417,000 but this has been recently increased to $729,000. While Fannie Mae was established by the government, it is a private company; but investors seem to believe that if Fannie Mae had real trouble they would get some help from the government. There are other GSEs that perform similar functions including GNMA and Freddie Mac.

In recent years the ABS (Asset Backed Securities) issuers became a more important source for home mortgages. In 2005 $507.4 Billion mortgages were acquired by ABS pools compared to $167.6 by Agency and GSE-backed mortgage pools and $324.0 Billion that were held by banks.

Asset Backed Securities (ABS) are pools of securities whose credit rating often exceeds that of the underlying securities. Obligations such as home loans, auto loans or credit card receivables may have a high degree of risk, but can be “packaged” to produce a security with fairly low risk. This is usually done by providing a credit enhancement or dividing a pool of risky debt into various tranches. An example of credit enhancement would be overcollateralization and yield spread, for example issuing a $100 security with a 6% yield backed by $120 of risky debt with a yield of 10%. An example using tranches might be using sequential pay: $500 of mortgages could fund 5 bond issues; holders of bonds 2 through 5 would receive nothing until bond 1 was fully paid off.

With all that in mind, here is what happened in the peak years of the housing boom. Interest rates were low; financial institutions were hunting for high quality securities that would yield a bit more than Treasury securities. ABS or more specifically MBS (mortgage backed securities) seemed to be the answer to the problem. The typical transaction was something like this (and see here for an example with more details):

  1. Mortgage broker makes a somewhat risky home loan.
  2. Mortgage broker sells the home loan immediately to a SIV (Structured Investment Vehicle) that packages the mortgages.
  3. The SIV then creates new securities that represent claims on the mortgages. Some of these securities get AAA ratings since they are overcollateralized and have a positive yield spread.

These new securities were incredibly popular among institutional investors. As Chairman Greenspan notes in his essay, judging from the demand, the mortgage securities appeared to be underpriced. Buyers asked the SIVs for more securities, so the SIVs asked the mortgage brokers for more mortgages. Mortgage brokers faced little risk if they could sell any mortgage immediately. Why carefully examine the loan application if there was a line of buyers waiting to buy the security as soon as the deal was signed? And so the bubble expanded, until it could not and everything fell apart.

Housing prices stopped rising, borrowers \started to default on their mortgages, investors stopped buying mortgage products from SIVs and SIVs stopped buying mortgages from mortgage brokers and people with poor credit got fewer mortgages.

The question to ask today is why these securities seemed so cheap at the time.

  • One serious question is how many securities were real and how many were fraudulent. In a boom period there are always some illegal/unethical participants in the market who create a product that looks like the real thing but is deficient in some way. See here for some mortgage brokers who are going to prison.
  • A related question is how many mortgage brokers made mortgage loans without sufficient attention to details. Undoubtedly the pressure on mortgage brokers (lots of money if they could process mortgages quickly) led some to overlook missing information and ignore possibly disqualifying circumstances.
  • Another question is how many of the SIVs misrepresented exactly what they were selling or the likely returns.

Certainly the low interest rate environment and the hot housing market were necessary underlying conditions. The major risk for someone who buys a mortgage is that the borrower will stop paying. But if home prices are rising, then even if the borrower has a financial problem (loss of a job, poor health) and cannot pay his mortgage the problem is not so serious. As we will see in the essay on “What happens if borrowers do not pay their mortgages?”, banks usually expect to lose about half the value of a defaulted mortgage. But if home prices have risen sufficiently, the borrower may be able to sell the home and pay off the mortgage preventing any loss for the owner of the mortgage.

Mortgages for those with less than perfect credit

Tuesday, June 24th, 2008

The mortgage rate that you pay depends, among other factors, upon your credit score. There are three major credit scoring agencies in the US; the Fair Isaac Corporation ( compiles these scores and allows consumers to view and track their scores over time.

Credit scores range from 300 to 850; the higher your score, the lower the interest rate you will pay on your mortgage. The credit score depends on the payment history (how many times in the past you have missed payments), the amount of debt you owe, the length of your credit history, the amount of credit you have applied for recently and the types of credit that you have used. The median FICO score in the US is 723. As of April 2008, a borrower with a credit score of 760-850 would pay 5.5% for a 30-year fixed rate mortgage, 700-760 5.7%, 660-700 6.0%, 620-660 6.8% and 560-620 9.2%. Below 620 is considered subprime.

People with poor credit histories traditionally had a difficult time obtaining mortgages. Many years ago, when lenders were stricter, home buyers needed to make a 20% down payment, which was probably a tighter requirement than the credit rating. Even borrowers with good credit had difficulty saving such a large amount and could not get financing. With the financial innovation in recent years there were many new ways to borrow.

One innovation is “piggyback” mortgages. The borrower takes out two loans, one normal mortgage for 80% of the value of the house, and a second, riskier mortgage for an additional percentage. This second mortgage tended to be at a higher rate and a shorter maturity than the normal mortgage, but it essentially allowed the homebuyer to borrow some or all of the down payment.

Once borrowers did not need to have a significant amount of savings to buy a home it was only a matter of time until loans were made to those with less than perfect credit. With the official government policy of increasing homeownership rates there was pressure on lenders not to refuse mortgages to borrowers with imperfect credit; numerous suits were brought against banks for the crime of “redlining” (see here for a settlement where a bank was obligated to offer to make mortgages at “the prevailing interest rate to borrowers who would not otherwise qualify for that rate”). With a rising housing market and low global yields on funds after the recession of 2001, the stage was set for the subprime crisis.

Given the high interest rates assigned by the market to subprime borrowers (roughly 9% at a time when borrowers with good credit paid 6%) the great innovation of recent years was creating adjustable rate mortgages (ARMs) that would allow subprime borrowers to pay low rates for a few years and purchase a larger house than they would otherwise be able to afford (see here for a guide for consumers to adjustable rate mortgages; the part relevant to subprime mortgages is at the end where they discuss hybrid ARMs).

The typical subprime loan is a 2/28 or 3/27 hybrid adjustable rate mortgage (see here for a brief description). This loan has a low rate for the first two or three years and a higher rate for the remaining years. The advantage for the borrower is that they can buy a bigger (more expensive) house right away, pay the low rate for a few years and then refinance to a lower rate. The key to the deal is that if a subprime borrower can pay a mortgage for two or three years his credit will improve and he will then be able to refinance the house at a better interest rate.

Some lenders added prepayment penalties to the deal, meaning that if the borrower repaid the mortgage (refinancing the loan) in the first five years the borrower would have to pay an additional fee. Why did borrowers accept such terms?  Some argue that the prepayment penalties were not obvious to borrowers.  My own experience with several mortgage closings is that there are many documents to be signed at a closing, and it is quite possible that unscrupulous lenders could conceal such terms from borrowers who were not represented by good attorneys.  Regulations to compel lenders to make these terms more obvious have been implemented in the recent past.

But during the 2004-2006 housing bubble it seemed like home prices were rising rapidly, so that even if a borrower would not be able to repay the loan he probably thought that he could sell the house at a profit as a last resort.  When housing prices stopped rising, the number of defaults on subprime loans started to rise sharply.

Overbuilding, or are there too many houses in the US?

Tuesday, June 24th, 2008

Story told by an old investor on CNBC (more or less accurate): I was golfing near New York City one day and the caddy knew I was a famous investor. The caddy asked me about the real estate market in San Diego; I asked him why he was so interested. He told me he owned six condominiums and was trying to decide what to do with them. I replied that it depended on where they were and whether the buildings were managed well. He replied that he had never been to San Diego.

After the attacks of September 11, 2001 and the collapse of the NASDAQ, American investors were wary of the stock markets; with low short and long-term interest rates (partly due to the Fed’s rate cuts, partly due to a world wide saving boom that lowered long-term rates around the world), Americans started to invest in something they thought they knew better, housing. As home buyers started to buy larger houses prices for existing homes started to rise, and home builders tried to meet the new demand.

The first attempt to meet growing demand was to convert rental housing into condominiums (with better kitchens and bathrooms). There is a several year lag between planning a new house and the time it is ready for a new occupant (due to the time required to acquire the land, obtain local government permission to build, and the time required to build the house), so the initial rise in prices resulted in increased homebuilding a few years later (see here for price data and here for construction data).

Year Median Price Average price Houses Built
2001 $175,000 $213,000 1.57m
2002 $188,000 $229,000 1.65m
2003 $195,000 $246,000 1.68m
2004 $221,000 $274,000 1.84m
2005 $241,000 $297,000 1.93m
2006 $247,000 $305,000 1.98m
March 2007 (peak price) $262,000 $329,000 1.61m
December 2007 (lowest price) $226,000 $283,000 1.33m
January 2008 $2226,000 $283,000 1.34m
February 2008 $244,000 $296,000 1.25m
March 2008 NA NA 1.22m

Source: Census Department.

A few points on the data before the analysis starts: first, the U.S. housing market is fairly seasonal, with more transactions in spring and summer than during the rest of the year, so it is hard to put too much weight on any specific monthly number. Second, the price data is at best an indication of what is going on; the purchase of a house is a very complex transaction and the price is one number among many that determine the actual cost of the house. But other fees, interest rates (some homebuilders offer financing too) and other attributes of the house can change the deal significantly. There were many reports of homebuilders offering upgrades (i.e., nicer kitchens or bathrooms) in order to stimulate sales.

Was there too much building? One way to examine this question is to look at the number of homes built relative to the demographic demand. A crude estimate is that the demand for new homes in the US equals the number of new households and the desired number of homes per household. The number of new households in turn depends on the population and average household size. The Census Department provides some data here.

Economists and demographers have estimated that given the rate of population growth, the average size of households and the desired number of homes per household (some households own multiple homes, including seasonal or vacation homes) that the demand for homes would be around 1.5 million homes per year in the current decade.

Looking at the table above, you can see that more than 1.5 million houses were build every year from 2001 until 2006 (in 2007, total annual home construction was almost exactly 1.5 million). These numbers imply that there were roughly 2 million “extra” houses built between 2001 and 2006; to return to equilibrium, home building must remain under 1.5 million per year for several years. The adjustment does not occur all at once (home building going to zero) because the extra houses are not distributed equally in all states and cities. A family looking for a new house in Pittsburgh is not affected by the hundreds (or thousands) of extra homes in Miami or San Diego.

The most recent data imply that home construction numbers should continue to decline; the most recent housing starts data reveal that at current rates fewer than 950,000 homes will be built in the next 12 months. Housing starts have been below the 1.5 million level for one year, reducing the overbuilding by around 250,000 homes. It is impossible to know how many houses there should be in the U.S. at any time, but we can say that the gap between demographic demand and the supply of homes has been getting smaller.

How are US Homes Financed?

Tuesday, June 24th, 2008

A house is a long-lived asset, and there is a different story behind each of the 75 million private houses in the U.S. Because the number of homes is so large, even small problems (in terms of the overall economy) involve millions of homeowners. In this section we will look at the financing of U.S. homes from two perspectives that are both imperfect, but they should at least provide an overview of the picture of the entire home market.

The first perspective comes from the Federal Reserve Flow of Funds Accounts. The Flow of Funds provides a fairly good picture of the overall financial position of the balance sheet of U.S. households, publishing data with a lag of around two months (data for the period ending December 31, 2007 were published on March 6, 2008).

The Flow of Funds tables compiled by the Federal Reserve show that at the end of of 2007 (the fourth quarter data), the value of household real estate was $20,150 billion dollars. These homes have been financed with $10,500 billion mortgages, which means that the owners’ equity is around 48% of the total value (debt is around 52% of the total value). So the average homeowner has a mortgage of a little more than half the value of his property.

The shortcoming of the Flow of Funds data is that in presenting average data, it is difficult to know how many households are having serious problems. So the next data set that we will look at is the American Housing Survey, a publication of the Census Bureau. This survey is far more complete but is published with a much bigger lag; the survey is performed in odd-numbered years and released around a year later. The latest data currently available are for 2005 and data for 2007 should be available later this year.

In the American Housing Survey in Table 3-15, we find that in 2005 about 25 million of the 75 million homes have no mortgage at all, being owned outright by homeowners. The approximately 50 million mortgages have a median principle amount of $92,000 (55% loan to value), interest rate of 6%, with 24 years of payments remaining. The median mortgage was made in 2002 for 29 years.

To better focus on the housing crisis, we will focus on the riskier mortgages. About 10 million mortgages have a loan to value ratio greater than 80%, of which 2.5 million mortgages have a loan to value greater than 100%. This number has certainly increased since the 2005 survey, especially with the recent fall in the prices of some homes.

18 million of the mortgages had been refinanced at some point, 15 million to receive a lower rate. Only 2.5 million mortgages had been refinanced to receive cash, that is paying off a smaller mortgage by taking out a larger mortgage and receiving cash as a result.

The typical US mortgage is a 30-year, fixed rate loan for less than $417,000, a so-called “conforming” loan. In late February of 2008 (the most recent available data), the average loan of this type had a coupon of 5.93%; given the lag between the time that a loan is negotiated and the deal is signed, these rates are associated with loans priced in mid January. The Federal Housing Finance Board (FHFB) reports a large amount of data about the types of loans made here. Loans that were taken out in February were on average $218,000 on homes worth $310,000; 26% of the loans had a loan to price ratio less than 70% and 25% were over 90%; only 8% were adjustable rate mortgages.

Other loans include 15 year fixed-rate loans and adjustable rate loans; jumbo loans (for more than $417,000) are counted separately.  Most loans allow borrowers to refinance without a penalty, meaning that when mortgage interest rates drop there is often a wave of refinancing.

Bad Behavior has blocked 628 access attempts in the last 7 days.

FireStats icon Powered by FireStats