The Fantasy Testimony Continues
By Arnold Kling
Below is the longest blog post ever. It contains the “written remarks” that extend my fantasy testimony.
Update: I will have to revise this to take a more conservative view of the proportion of investor loans. A reader sent me data on subprime loans showing that investor loans are only foreclosing at a rate slightly higher than owner-occupied. This still leaves open the possibility that investor loans are doing much worse than owner-occupied in the conventional mortgage category, but in any case I need to re-do my math on that one.Written Remarks
1.The History of Mortgage Securitization
Suppose that you were a bank executive and were offered a choice between two methods for holding mortgage loans. Call them Method A and Method B.
Under Method A, you employ the mortgage originators. You give them rules and instructions. You set their incentives. You can choose whether to pay them only for accepting loans or to also pay them for reviewing and rejecting loan applications where appropriate. They originate loans in your local community. The loan terms are set by your policies. You receive the borrowers’ payments and deal with loan delinquencies according to your procedures. Sometimes, your procedures call for rapid foreclosure. In other instances, they dictate some sort of loan workout.
Under Method B, you hold a security interest in a pool of mortgages. These loans were originated by people unknown to you, whose only incentive is to maximize “production,” without regard to quality or risk. They are paid when they originate a loan, never for rejecting a loan application. The loans come from far and wide, including many places with which you are not familiar. If borrowers are delinquent in their payments, you have no control over how this delinquency is addressed.
Most executives, if offered this choice, would prefer method A. Compared with Method A, Method B does not pass a simple sanity check. Using economic jargon, method B has much higher “agency costs.” Under Method A, the people acting as the agents of the executive work for the bank. The executive has the power to align the incentives of the agents with those of the shareholders of the bank. Under Method B, the agents who originate home mortgages are working against you, not for you. They are trying to slip as many bad loans through the door as they can. It is up to someone else, not you, to stop them–assuming that they care..
Today, three-fourths of mortgage debt in the United States is held using Method B, also known as securitization. How did this happen?
Method B has some potential advantages. A pool of loans from different regions has more geographic diversity than a set of loans from a single community. Also, a highly specialized mortgage intermediary, such as Freddie Mac or Fannie Mae, can develop risk management and quality control procedures to manage third-party originators, and with economies of scale they can hold down agency costs. Securities markets make it possible to place mortgage debt with mutual funds, pension funds, and insurance companies, thus broadening the market.
However, with a level regulatory playing field, depository institutions could obtain all of the advantages of Method B with none of the disadvantages. A multi-state bank holding company could have a portfolio of mortgage loans that is geographically diversified. A well-known bank holding company could issue debt instruments that could be held by any investor who now holds mortgage securities.
The mortgage securities market exists because of regulations that prevent depository institutions from using Method A economically at sufficient scale. The balance of advantages shifts to securitization only because of the way that government puts its thumb on the scale.
The secondary mortgage market began in 1968. In that year, President Lyndon Johnson was besieged. His war in Vietnam was unpopular. Along with his cherished War on Poverty, it was raising the need for government borrowing. Each time the President came to Congress to request an increase in the ceiling of the national debt, he faced embarrassment and attacks. To forestall this, his Administration looked for ways to get government housing programs off the books.
One solution was to sell the Federal National Mortgage Association (Fannie Mae) to private investors. Fannie Mae had been set up in 1938 as a national purchaser of mortgage loans originated by third parties, called mortgage bankers. Fannie Mae did not securitize loans, at least to this point. Instead, it held mortgage loans in its portfolio, like a giant Method A lender with the mortgage bankers as its agents. Selling Fannie Mae took Fannie Mae’s debt off the government books, which made the national debt appear smaller.
Another tactic for controlling the national debt was the introduction of the Government National Mortgage Association, GNMA, which introduced the first mortgage-backed securities. GNMA created and sold securities backed by loans guaranteed under programs of the Federal Housing Administration (FHA) and the Veterans’ Administration (VA). The risk of mortgage defaults still rested entirely with taxpayers under FHA/VA. However, selling the GNMA pools took the loans off the government’s books for accounting purposes.
The secondary mortgage market thus began as an accounting gimmick to hide liabilities. It has been the same ever since. The economic costs and benefits of securitization are beside the point. Over time, various accounting gimmicks and regulatory anomalies have driven securitization forward.
The Federal Home Loan Mortgage Corporation, Freddie Mac, was chartered in 1970 as an agency under the Federal Home Loan Bank Board. in order to promote Method B lending in conventional mortgage loans, meaning loans that fund middle-class home ownership. An individual mortgage that is above a certain ceiling, which is indexed to house prices unless altered by Congress, is not eligible for purchase by Freddie Mac or Fannie Mae. Freddie Mac also was chartered to stick to “investment-quality” mortgage loans, meaning that it was not to deal in loans with low down payments or what we would now call the sub-prime market.
As of 1970, mortgage lending in the United States was predominantly Method A, conducted mostly by the savings and loan industry (S&Ls). However, inflation and interest rate regulations were soon to wreak havoc on the S&Ls. At that time, interest rates on deposits were limited by government edict, known as regulation Q. As inflation soared, market interest rates rose above regulation Q ceilings. Competitors, notably money market funds, lured consumers away from S&L’s, limiting their ability to serve the needs of mortgage borrowers. Furthermore, restrictions on interstate activity created a shortage of mortgage funds in California even though there were ample savings deposits in the East.
By securitizing loans with Freddie Mac, the S&L’s, particularly in California, were able to expand their mortgage lending beyond the limits of their deposit base. However, it is important to recognize that S&L’s only needed securitization because of the regulatory constraints that prevented them from taking other steps to obtain funds, such as raising interest rates on deposits or issuing debt collateralized by mortgage assets.
In the latter part of the 1970’s, Robert Dall and Lew Ranieri, two executives at the bond-trading investment banker Salomon Brothers, decided that it was time to take the mortgage market away from Method A and the S&L industry. They saw the potential for huge trading profits if America’s mortgage debt could be securitized. With the S&L industry weakened by the combination of higher inflation and interest rates and stifling regulation, Salomon Brothers heavily promoted mortgage securitization, primarily through Freddie Mac and Fannie Mae. In the process, they learned how to manipulate Congress and government regulators to help implement their vision.
For example, in the early 1980’s Freddie Mac introduced a program called Guarantor. Fannie Mae soon followed with a program called Swap. The purpose of these programs was to perpetrate an accounting hoax on behalf of insolvent S&Ls.
At that point in history, the S&L’s had issued fixed-rate mortgage loans which had declined in market value because interest rates had risen during the interim. Had they marked their assets down to their true market values, they would have had to go out of business. As long as they hung onto the loans, the accounting standards then in force allowed them to postpone recognizing any losses.
At the same time, the S&L’s needed to raise cash, and the only assets they had to sell were the underwater mortgage loans. Not selling the mortgages meant running out of cash. On the other hand, selling the loans meant having to recognize losses.
What Guarantor and Swap did was allow the S&L’s to exchange their mortgages for securities backed by those same mortgages. They could use the securities as collateral for borrowing, in order to raise cash. The key to the whole transaction was an accounting treatment that allowed the S&L’s to defer recognizing losses on the securities in the same way that they could defer losses on the underlying mortgages. This accounting ruling was granted by the Federal Home Loan Bank Board, thanks in large part to heavy lobbying by Wall Street and S&L executives.
The result of Guarantor and Swap was to worsen the bleeding of the S&L’s, with large fees collected by Freddie, Fannie, and Wall Street firms. Meanwhile, securitization allowed the S&L’s to avoid recognizing insolvency, so that they could keep gambling even when they had negative net worth. When the S&L’s finally went out of business, the taxpayers took the loss.
By the 1990’s, the depository institutions that had been the mainstays of Method A lending were defunct. The FDIC and the Federal Reserve Board issued new risk-based capital regulations that were intended to prevent a repeat of the S&L debacle. Meanwhile, Freddie and Fannie received their own special regulator, who issued different capital regulations.
The bank capital regulations were crude, creating different classes of assets with different risk weights. For example, low-risk mortgages (with down payments of 20 to 40 percent) were given a risk weight of 0.35, on a scale that essentially goes from 0 at the lowest to 1.0 at the highest.
The Freddie/Fannie capital regulations were based on a stress test that simulated a pattern of declining house prices. The stress test penalized mortgages with low down payments, which historically had been outside of the two companies’ charters all along. However, for low-risk mortgages, Freddie and Fannie were required to hold much less capital than banks. This gave Freddie and Fannie a significant cost advantage, and over the decade of the 1990’s the two firms essentially took over the market for conforming mortgage loans (loans falling underneath the loan-limit ceiling and with sufficient down payment and borrower credit history to meet the “investment quality” standard.)
Early in the 21st century, private securitization (meaning mortgages securitized by Wall Street firms, not by Freddie or Fannie) emerged as a major force in the mortgage market. This phenomenon developed for a variety of reasons.
First, credit scoring had emerged as a way of assessing a borrower’s credit history. This in turn lowered some of the agency costs associated with Method B lending. Instead of having to trust the mortgage broker to examine the credit report by hand, an investor could treat the credit score as hard data.
Second, Wall Street needed to find a substitute for the guarantee supplied by Freddie Mac or Fannie Mae. If you buy a Freddie Mac mortgage security and one of the borrowers defaults, that loan is pulled from the pool by Freddie Mac. Freddie Mac immediately pays into the pool the unpaid principal balance on the defaulted loan, and then tries to recover for itself what it can from selling the house. The investor is insulated from the default loss. (Fannie Mae offers the same protection.)
What Wall Street came up with to back private mortgage securities was the idea of a credit risk tranche. If you bought a senior tranche, then it was guaranteed to absorb no losses until at least, say, 10 percent of borrowers defaulted. Since this was a rare event, the senior tranche was considered safe. Junior tranches, which absorbed a disproportionately large share of losses, could be protected with credit default swaps, in which a large insurance company or other intermediary agreed to absorb the losses.
The market in private mortgage securities expanded during a period of rising home prices. In this environment, it was profitable to reach into segments of the market where it was not possible to make investment-quality loans. This included borrowers with blemished credit histories and borrowers who could not afford even a 5 percent down payment. The private securitizers made riskier and riskier loans, but as long as home prices kept increasing, defaults were rare and market participants enjoyed nice profits.
The securitization process so bamboozled the regulators that banks were holding securities backed by high-risk, low-down-payment mortgages that had been rated AAA, and thus had a risk capital weight of 0.20, which is significantly less than the 0.35 risk weight given to a low-risk mortgage with a 40 percent down payment originated using Method A. The regulators were telling banks to treat Method B mortgage loans with low down payments as safer than Method A loans with high down payments.
As home prices rose, it became more and more difficult for borrowers to come up with 20 percent down payments, causing the proportion of conventional mortgages to shrink. As private securitization increased in importance, Freddie and Fannie saw their market shares, which had peaked at 50 percent in 2003, start to decline. Moreover, the segment of the market they were left with was increasingly upscale, so that political leaders began to berate Freddie and Fannie over their lack of support for “affordable housing.” The pressure built on Freddie and Fannie to join in the high-risk lending frenzy, and they caved into that pressure.
News reports show that at both Freddie and Fannie, warnings were issued by staff about high-risk lending. The stress test methodology required the firms to dedicate large amounts of capital for these loans in order to protect the firms in case of a downturn. Not wishing to abandon the high-risk market or to dilute earnings by raising the capital called for by the stress tests, the CEO’s at the two firms simply over-rode staff objections and dove into the market, without raising the requisite capital.
Freddie and Fannie were never the dominant high-risk lenders. Nonetheless, they took on more risk than they should have, with less capital than was prudent. Had they maintained a focus on safety and soundness and stayed out of high-risk lending, the firms would done less to inflate the house price bubble. Freddie and Fannie would be in good shape now to pick up the pieces of the faltering private securitization market. Instead, the two firms themselves required a taxpayer bailout.
Finally, it is important to bear in mind that Freddie Mac and Fannie Mae were part of the Method B mortgage lending process. If capital requirements had been rationally tied to risk and applied equally to all institutions, Method A lending would have driven method B lending out of the market. Freddie and Fannie would not have grown to dominate the market. Instead, my conjecture is that they would not have been able to gain even a toehold in a free and fair market. Mortgage securitization is entirely a product of regulatory distortions.
The regulatory distortions were by no means accidental. Often, the regulatory loopholes were pried open by lobbyists working for Freddie Mac, Fannie Mae, or Wall Street firms that profited from securitization. Key members of Congress were generously plied with campaign contributions, in return for which they championed securitization and emasculated the regulators who oversaw Freddie Mac and Fannie Mae.
2.Speculation and the Housing Market
I credit other economists with warning early and strongly about a speculative bubble in housing. Prominent examples include Dean Baker, Paul Krugman, and Robert Shiller. Shiller, in particular, has given a very eloquent description of the bubble in the housing market in his recent book, The Subprime Solution.
As of 2004, I was one of those who thought that high home prices reflected unsustainably low interest rates. Today, some people continue to blame the run-up in home prices on Federal Reserve policy and low interest rates. However, I have changed my position on that, and I now believe that the bubble was speculative.
A major contributing factor to the speculative bubble was the explosion in lending for home purchase with little or no money down. When the down payment is small, the buyer’s equity consists almost entirely of price appreciation. When prices are rising, anyone can buy a home with a low down payment, and any mortgage loan is safe. Low-down-payment lending helps foster a speculative frenzy on the way up. Of course, prices cannot go up forever. Once prices stop rising, the low-down-payment loans tend to go sour rather quickly.
There is considerable evidence that many homes were bought purely for speculative purposes. William Wheaton, a professor of urban economics at MIT, estimates that in recent years the growth of housing units exceeded the increase of household formation by six percentage points. These excess houses were bought by speculators.
Further evidence of speculative excess can be seen from an article that appeared in the December 2007 issue of the Federal Reserve Bulletin. It showed that loans for non-owner-occupied homes (also known as investor loans) grew from roughly 5 percent of total mortgage originations a decade ago to more than 15 percent of originations in 2005 and 2006. This may in fact understate the amount of speculative buying, because it is common for speculators to lie on their loan applications by claiming that they intend to occupy the home.
When I was looking at mortgage credit risk, we expected investor loans to default at three to ten times the rate of normal loans. If 85 percent of loans ((the ones issued to real homeowners) default at rate X, and 15 percent of loans (issued to speculators) default at rate 3X, then loans to speculators would account for about 35 percent of all defaults. If you use 10X, then speculators would account for about 85 percent of all defaults. Thus, it seems reasonable to suppose that between 35 percent and 85 percent of troubled loans are for non-owner-occupied housing. This makes attempts to “protect homeowners” or do a “bailout from the bottom” highly problematic.
One issue that is worth studying is the fact that there were price bubbles in real estate markets in some foreign countries, also. It is not likely that those bubbles were the result of monetary policy of the United States, and even less likely that they were the result of regulatory changes in the United States. Did other countries also encourage higher leverage in real estate, or were the bubbles able to inflate without any such encouragement? I wish that I knew the answer to that question.
3. The Foster-Van Order Model of Mortgage Default
When I was at Freddie Mac, we adopted an approach to dealing with mortgage credit risk that was developed by Chet Foster and Robert Van Order, two economists who came to Freddie Mac from the Department of Housing and Urban Development. The idea is that a borrower has trouble making the payments on a loan has two choices. One choice is to sell the house. The other choice is to allow the lender to foreclose on the house.
Allowing the lender to foreclose is what we call the option to default. If the house can be sold for more than the outstanding balance on the loan, the borrower will sell the house. We say that the default option is out of the money. On the other hand, if the value of the outstanding balance on the loan exceeds the market price of the house, the borrower might as well default. We say that the default option is in the money.
Speculators are more likely to exercise the default option than are owner-occupants. The owner-occupant takes into account the cost of relocating his or her family.
When the borrower makes a down payment of 20 percent, the default option is far out of the money. House prices have to fall by 20 percent before the default option begins to make sense. On the other hand, with a down payment of 2 or 3 percent, the default option can be in the money if house prices decline only slightly. That is what makes low-down-payment mortgages dramatically more risky than mortgages with a down payment of 20 percent or more.
4. Credit Default Swaps and Systemic Risk
I will define systemic risk as follows: whenever individuals make contingency plans that can only be executed if others are not trying to execute similar contingency plans, there is systemic risk. For example, suppose that many of us have money in a bank where deposits are not insured. I form a contingency plan which says that if the bank gets into trouble, I will run down to the bank and withdraw my deposit before the bank runs out of money. If everyone else forms the same contingency plan, we cannot execute our plans at the same time. Instead, the result is a bank run.
Another example of systemic risk was involved in the stock market crash of October 19, 1987. At that time, many institutional investors had purchased “portfolio insurance,” in order to protect against a downturn. Portfolio insurance was supposed to act like a stop-loss order for a diversified stock portfolio, giving owners a guarantee that the value of their investments would not fall below a given floor. Each company that sold portfolio insurance had a contingency plan that consisted of executing a computer program to sell shares of stock in the event that stock prices started to decline. Collectively, these contingency plans were incompatible, because the computer programs were all trying to sell at the same time.
Finally, we come to credit default swaps. The buyer of a credit default swap pays a fee to a seller, in exchange for which the seller promises to pay a large sum to the buyer in the event of a default on a mortgage security or corporate bond. A seller of a credit default swap is in same position as a property and casualty insurance company. Ordinarily, the insurer simply collects premiums, and only rarely must it pay claims.
Faced with the prospect that they might have to pay significant claims, some sellers of credit default swaps probably had formed contingency plans that involved selling short bonds and stock related to their guarantees. Thus, if you had sold a credit default swap on debt issued by Lehman, you could hedge by shorting Lehman debt, Lehman stock, or by shorting the securities of similar financial institutions.
Of course, these individual contingency plans, when executed collectively, resulted in waves of short-selling. Credit default swaps apparently were another example of systemic risk, in which individual contingency plans were not mutually compatible.
Note that eliminating counterparty risk, by moving credit default swaps from the over-the-counter market to an organized exchange, would not solve the problem. Even when traded on an organized exchange, credit default swaps would have to be hedged using short-selling strategies that create the equivalent of bank runs.
5. Suits vs. Geeks
Financial innovation has outpaced the ability of financial executives and regulatory agency heads to remain current. The financial engineers (the geeks) create products that behave quite unlike ordinary bonds. The executives (the suits) rely on intuition that applies to simpler securities. The results are catastrophic.
When I was at Freddie Mac, there was hardly any gap between the suits and the geeks. The Foster-Van Order model of mortgage default was ingrained in the corporate culture. The CEO, CFO, and other key executives understood this model and its implication that mortgage defaults would be much higher for mortgages with low down payments. Moreover, the suits bought into the idea of using a stress test to set capital requirements. Using a stress test methodology, in which mortgages are evaluated according to how well they would survive a downturn in house prices, the capital required to back mortgages with low down payments is prohibitively high.
When a new CEO came to Freddie Mac in 2003 (several years after I had left), a gap apparently opened up between the suits and the geeks. Warnings issued by the Chief Risk Officer and others about low down payment mortgages were ignored by the CEO.
For decades, Wall Street traders have taken advantage of depository institution executives’ inability to keep up with financial innovation. Securities with very unfavorable risk characteristics often are pawned off on unwitting banks or savings and loans. As part of this process, the Wall Street firms obtained crude and misleading risk measures from the credit rating agencies. Industry geeks knew that these ratings were inappropriate, but the suits at banks relied on the ratings, in part because that is what the suits at the regulatory agencies were telling them to do, particularly with regard to capital requirements.
Most recently, the suits vs. geeks divide emerged concerning the issue of whether mortgage securities are undervalued. Many suits, including Henry Paulson and Ben Bernanke, took the view that mortgage securities were artificially undervalued. The original Paulson Plan, in which the Treasury would enter market to buy mortgage securities, was based on the assumption that it would be profitable to hold these securities to maturity.
Geeks tend to believe that the market has correctly reduced the values of mortgage securities. Using the Foster-Van Order model, it would seem that the default option on many mortgages is “in the money,” causing a huge loss of value for mortgage-backed securities. The fact that so many loans issued in recent years were investor loans is even more ominous. Finally, the imbalance in the housing market means that prices could fall even more. This would expand the losses on mortgage securities.
From a geek perspective, there is an asymmetry in the possible outcomes from investing in mortgage securities. The likelihood of earning a profit is very high, but the amount of the profit will be small. The profit will come from scenarios in which house prices fall only modestly over the next several years. On the other hand, the likelihood of a severe housing depression is low, but the consequences for mortgage securities would be devastating. The potential for taking a large loss, even with a small probability, would lead a risk-averse investor to be cautious about buying mortgage securities, even though the likelihood of earning a small profit is high.
For an analogy, consider a game in which we role a six-sided die. If the number on the die comes up 1,2,3,4, or 5, I give you one dollar. If the number comes up 6, you give me hundred dollars. You are more likely to win than to lose, but you still would not play that game unless I made it more attractive. The same holds true for investing in mortgage securities.
In other words, the market may be quite rational in pricing mortgage securities. The suits who wish to have Treasury speculate in that market would be forcing American taxpayers to engage in a gamble that professional investors would rather not take.
Michael Lewis’ book, Liar’s Poker, contains an insightful history of the early days of the mortgage securities market. The book is based on Lewis’ experience as a trainee and salesman for Salomon Brothers. He portrays the way that Wall Street exploited the weaknesses of the S&L industry to muscle its way into the mortgage market.
The specific details of risk weightings and capital requirements, which are an important part of the story, are not easy to track down. One helpful FDIC document is here: http://www.fdic.gov/news/news/financial/2008/fil08069a.html
For an analysis of how risk weightings create regulatory arbitrage and artificially boost mortgage securitization, see the paper “Risk-Based Capital Requirements for Mortgage Loans,” by Paul S. Calem and Michael Lacour-Little. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=295633
For documentation of the increase in loans for non-owner-occupied housing in recent years, see “The 2006 HMDA Data,: by Robert B. Avery, Kenneth P. Brevoort, and Glenn B. Canner,” Federal Reserve Bulletin, December 2007. http://www.federalreserve.gov/pubs/bulletin/2007/pdf/hmda06final.pdf
For more on the suits and geeks divide at Freddie Mac and Fannie Mae, see the stories in the New York Times by Charles Duhigg. “At Freddie Mac, Chief Discarded Warning Signs,” August 5, 2008. http://www.nytimes.com/2008/08/05/business/05freddie.html?hp; and “Pressured to Take More Risk, Fannie Reached Tipping Point,” October 5th, 2008. http://www.nytimes.com/2008/10/05/business/05fannie.html?hp
Also, see Susan Woodward, “Rescued by Fannie Mae?” Washington Post, October 14, 2008. http://www.nytimes.com/2008/10/05/business/05fannie.html?hp. She argues that Fannie Mae modelers understand mortgage credit risk, but business executives ignored the modelers.
Three academic webcasts provide useful information and analysis of the crisis. At MIT (http://events.mit.edu/event.html?id=9303415&date=2008/10/08), William Wheaton provides information on the housing glut promoted by speculation. At Harvard (http://video2.harvard.edu:8080/ramgen/AAD-PAN/FinMktsPanel.rm), Ken Rogoff describes how the financial sector in the United States is bloated. Also, Robert Merton talks about what I call the suits vs. geeks divide. Finally, at the University of California, San Diego (http://www.econbrowser.com/archives/2008/10/roundtable_disc.html), James Hamilton has some helpful slides that illuminate the difference between what I call Method A lending and Method B lending, along with a chart showing the evolution over the past decade of the market shares of various forms of mortgage lending.