Arnold King argues that mortgage securitization should be allowed to die in order to keep taxpayers from inheriting more risk

The Humpty-Dumpty of mortgage securitization.

Like Humpty-Dumpty, mortgage securitization has taken a big fall. There is a widespread presumption that government policy, if not all the king’s horses and all the king’s men, should be aimed at putting securitization together again. The purpose of this essay is to question that presumption.

The first section of this paper will describe how securitization worked at Freddie Mac in the late 1980s, when I worked there. This will allow me to introduce and explain the concepts of interest rate risk and credit risk in mortgage finance.

The second section of this paper will describedevelopments in the mortgage industry from the mid-1980s through the 1990s, when Freddie Mac and FannieMae took on more interest rate risk. The third sectionlooks at what evolved over the past ten years, when theprocess for allocating and managing credit risk changed,with “private-label” securitization and the growth ofsubprime mortgages.

If the profits of the securities business would be fully privatized, and the credit risk would be fully socialized, the systematic risk would be borne by the taxpayers again.

The fourth section of this paper describes various options for reviving mortgage securitization. The final section steps back and looks at interest rate risk and credit risk from a public policy standpoint. Government policy influences the allocation of credit risk and interest rate risk in capital markets. What are the social costs and benefits of various allocations? I suggest that policymakers might consider reverting to the housing finance system that preceded the emergence of securitization, in which depository institutions were responsible for managing both credit risk and interest rate risk for mortgages.

Freddie Mac around 1989

I joined Freddie Mac as an economist in December 1986. About eighteen months later, I was promoted to Director of Pricing/Cost Analysis, under the Vice-President for Financial Research. My job was to oversee the models used to manage interest rate risk and credit risk.

At the time, my primary focus was credit risk. Freddie Mac bundled loans into securities, and then it sold the securities. If a mortgage loan defaulted, Freddie Mac would pay the entire unpaid balance on that loan to the security holder and then try to recover as much as it could from foreclosure proceedings on the property. Thus, Freddie Mac insulated security holders from the credit risk. This was known as the guarantee business, because Freddie Mac would guarantee that investors in its mortgage securities would not have to worry about individual mortgage defaults.

A change in market interest rates could affect the value of the cash flows due to the investor in a mortgage. Because Freddie Mac packed nearly all of the mortgages it guaranteed into pass-through securities, Freddie Mac in the late 1980s had very little interest rate risk. The interest rate risk was borne by the investors who bought Freddie Mac securities and relied on the cash flows that were passed through. Note that at that time there was a difference between Freddie Mac and Fannie Mae. Fannie Mae had traditionally financed most of its mortgage purchases with its own debt, rather than with pass-through securities. Thus, Fannie Mae had taken on interest rate risk.

Interest rate risk arises because the typical mortgage in the United States is a thirty-year fixed-rate mortgage with a prepayment option. This means that the firm receiving the cash flows of the mortgage (call this the mortgage holder) faces a difficult problem in matching funding with the cash flows from the mortgage.

Suppose that the interest rate on the mortgage is 8%,and suppose that the mortgage holder finances its positionby issuing a five-year bond at 7%. If interest rates remainunchanged, then after five years the holder can issueanother five-year bond at 7%. If this environment persistsfor thirty years, then the holder clearly makes a profit.However, suppose that after two years, market interestrates drop by two percentage points. The borrower takesadvantage of this to obtain a new mortgage at 6%, usingthe proceeds from this refinance to pay off the originalmortgage. The holder is still stuck with having to payinterest on the five-year bond for three more years at 7%.However, the holder cannot find investments that yieldmore than 6 percent, so the holder takes a loss. This isknown as prepayment risk, or the cost of the prepaymentoption.

On the other hand, suppose that after two yearsmarket interest rates rise by two percentage points andthat this rise is permanent. Now, the mortgage borrowerwill try to retain the loan as long as possible, while themortgage holder’s financing will run out after five years.At the end of the fifth year, the holder is going to haveto obtain new funding, which will cost 9%, so that theholder is going to be suffering a loss. This might be calledduration risk, because the cash flows from the mortgagehave a longer duration (the last payment will not bereceived until thirty years from the date of origination)than the holder’s liability (a five-year bond in ourexample).

When I joined Freddie Mac, its major competitors hadrecently been stung by duration risk. In the late 1960sand early 1970s, mortgage interest rates were around6%. By the early 1980s, market interest rates had morethan doubled. Fannie Mae, which at that time relied onmedium-term debt to finance its mortgage holdings, waslosing $1 million a day in 1982. More importantly thesavings and loan industry, which financed its mortgageholdings with short-term deposits, was bankrupt.Thus, by the early 1980s, the approach of fundingmortgages with short-term deposits was discredited.Securitization, which allowed the interest rate risk tobe transferred to institutions such as pension funds andinsurance companies, seemed to be a superior financingmethod.

The big challenge with securitization was managing credit risk. This required pricing policies, capital policies, and risk management policies.

For pricing, we wanted to price mortgages accordingto risk. We specified a probability distribution for houseprices, and we assumed that losses from mortgagedefaults would take place when individual house pricesfell below the outstanding mortgage balance. Becauseof this, the guarantee fee charged on a loan that wasfor 80% of the purchase price would be higher than thefee charged on a loan that was for 60% of the purchaseprice.

Our capital policy was tied to something that we called“the Moody’s scenario,” since it was suggested to us bythat credit rating agency, based on what happened in theGreat Depression. Under this scenario, the average houseprice would fall by 10% per year for four years, andthen remain flat thereafter. Although this was the averageprice path under the Moody’s scenario, we simulated adistribution of house prices, in which some fell by moreand some fell by less. We then measured the total lossesunder this scenario, and we assumed that we wouldneed enough capital to cover those losses. The cost ofthis capital was then priced into the guarantee fee. Thiscapital charge did even more to penalize the relativelyhigh-risk loans, such as loans backed by rental propertiesor loans with a high loan to value ratio.

Pricing for risk is fine, assuming that the loanorigination process is standardized. However, becauseloan origination was not under the control of FreddieMac, we faced principal-agent problems. The incentiveof the originators was to aim for volume, regardlessof quality. To appreciate the challenge that we faced,consider that we might encounter a shady mortgageoriginator, whose intent is to create fraudulent mortgagesand then abscond with the origination fees – or even theentire proceeds of the loan. Freddie Mac might revokethe eligibility of the shady operator, only to find that theoperator moves to another location and does businessunder a different name.

To manage this principal-agent problem, Freddie Mac had a number of devices (and Fannie Mae had very similar measures):

• a qualification system for sellers. To sell loans to Freddie Mac, you had to prove that your staff had experience and you had to show sufficient capital that you could buy back loans that had been improperly originated.
• a seller-servicer guide. This spelled out exactly the procedures and rules that we wanted originators to follow when approving or rejecting loan applications.
• contractual obligations. Sellers were providing contractual representations and warranties to us that they were following our guide.
• quality control. We inspected the loan files of a sample of loans from each originator. Loans that were found to violate the “reps and warrants” were put back to the seller to be repurchased.

All of these risk management processes were costly, and all were imperfect. The principal-agent problems in securitization were difficult, and we were constantly tinkering with our systems to try to improve them.

Freddie Mac and Fannie Mae achieve dominatio

In the mid-1980s, inflation and interest rates began trending down. This made it profitable to finance mortgages with medium-term debt. Indeed, the downward movements in interest rates served to highlight prepayment risk. Mortgage holders had difficulty protecting themselves against sharp declines in interest rates, which caused borrowers to refinance while the holders were still paying interest on medium-term debt.

Fannie Mae found a solution to the prepayment problem by issuing callable debt. For example, it might issue a ten-year bond that it could extinguish at par after five years, if interest rates had fallen. This effectively transferred prepayment risk from Fannie Mae to its debt investors, in return for which Fannie paid a slightly higher interest rate.

The innovation of callable debt ultimately produced a dramatic change in the mortgage market. It undermined the Freddie Mac model of issuing pass-through securities. Investors were more comfortable with the relative simplicity and transparency of callable debt. In the 1990s, Freddie Mac jointed Fannie Mae as a “portfolio lender,” meaning that it held its own mortgage securities and funded them with callable debt.

Funding mortgages with callable debt was so efficient that by 2003 Freddie Mac and Fannie Mae together held half of the mortgage debt outstanding in the United States. This dominant position was undermined by other innovations, discussed below.

If Freddie Mac, Fannie Mae, and AIG all are unable to proceed without
government backing, then there is no way that securitization can come
back without the government acting as a guarantor of last resort.

When I joined Freddie Mac, its major competitors had recently been stung by duration risk.

Securitization goes private-label

The benefits of securitization come from the fact that investors do not have to go to the trouble and expense of examining the underlying mortgages. Investors know the types of mortgages and the interest rates on the mortgages in the pool, which is the information that they need to manage interest rate risk. However, investors assume that they are entirely insulated from credit risk, because of the guarantee provided by Freddie Mac or Fannie Mae.

The guarantees from Freddie Mac and Fannie Mae were credible because of the capital and historical record of those firms. Most of all, however, the guarantees were credible because of the perception that it would be politically unacceptable to allow those firms to fail.

There are mortgages that Freddie and Fannie could not guarantee, because of legislated limits on the size of loan eligible for those agencies. Moreover, ten years ago there were loans that the agencies would not guarantee, because low downpayments or weak borrower credit history made the loans high risk for default.

About ten years ago, a number of innovations emerged that substituted for an agency guarantee, allowing “private-label” securities to compete with those of the agencies. Borrower credit scores provided a simple, quantitative measure of the borrower’s credit. Structured securities allowed credit risk to be reallocated, with subordinated holders taking most of the risk and senior holders only taking what was left over. The various tranches were evaluated by credit rating agencies, so that investors could treat AA private-label securities as equivalent to agency securities (a practice which was formally ratified by bank regulators in a policy that took effect on January 1, 2002). For extra comfort, the holder of a security could purchase a credit default swap, which would pay off in the event that the security’s principal repayment was jeopardized.

With all of these layers of protection, holders did not have to examine the underlying mortgages. In fact, it is not clear where that responsibility lay. With agency securitization, it was clearly the responsibility of Freddie Mac and Fannie Mae for managing, measuring, and bearing credit risk. However, with private-label securitization, those functions were diffused. The Wall Street firms that packaged securities had no experience with the risk management functions needed to ensure quality standards in mortgage origination. The credit rating agencies had most of the responsibility for credit risk measurement, but they bore none of the risk. In retrospect, the incentive to be overly generous in rating securities was far too high.

Toward the latter part of the housing boom, the risk management process also broke down at Freddie Mac and Fannie Mae. Private-label securitization and the growth of subprime mortgages were leading to a sharp fall in the market share at the two agencies. In retrospect, the agencies should simply have held on to their capital standards and risk management controls. However, at the time, they suffered from doubts about whether their traditional approach was still valid. They began to loosen standards for mortgage quality, to maintain insufficient capital relative to credit risk, and to purchase subprime mortgage securities based on agency ratings rather than on an assessment of the risks of the underlying loans. As a result, when the crisis hit, the agencies were not in a position to survive the losses that they incurred.

Fix securitization?

On April 30, 2009, Gillian Tett lectured at the London School of Economics.1 Tett, the author of perhaps the best book published so far on the origins of the financial crisis,2 was asked a question about the impact of the failure of Lehman Brothers. In her response, she said that having the securities market break down was the equivalent of waking up one morning and finding that the Internet and cell phones had broken down.

The consensus in the financial community is that securitization simply must be fixed. The question is how this can be done.

Securitization worked because the holders of securities assumed that they were not bearing any credit risk. Securitization broke down when mortgage defaults reached a level where holders of securities were no longer confident that they were insulated from credit risk. For mortgage securitization to work again, the credit risk will have to be absorbed in a credible way by someone other than the holder of the securities.

Mortgage credit risk includes both systemic risk and idiosyncratic risk. Systemic risk is the risk that conditions in the overall housing market will take a sharp turn for the worse. Idiosyncratic risk is the risk that mortgage originators will deliver faulty or fraudulent loans into the securitization process.

The private-label securities operations did not seem to have strong mechanisms for dealing with idiosyncratic risk. Recently, the U.S. Treasury published recommendations for financial reform that included requiring mortgage originators to retain a 5% interest in the mortgage loans that they deliver for securitization. This strikes me as a crude approach. Five percent is too high for an honest but capital-strapped mortgage broker. At the same time, it is too low to deter serious fraud: retaining a 5% interest is no penalty at all if your intent all along is to abscond with 100% of the funds.

A basic problem in private-label securitization is that the functions for managing idiosyncratic risk (procedures for qualifying sellers, establishing and enforcing guidelines, and so forth) are no one’s responsibility. Some party must take on those functions in order to address idiosyncratic risk.

Another problem with all forms of mortgage securitization is that of systemic risk. At this point, there is no private sector firm that can credibly insulate security holders from systemic risk. If Freddie Mac, Fannie Mae, and AIG all are unable to proceed without government backing, then there is no way that securitization can come back without the government acting as a guarantor of last resort.

One suggestion that I have heard is that government should provide support along the lines of “the GNMA model.” This strikes me as nonsense. The Government National Mortgage Association (GNMA or Ginnie Mae) packages loans that are guaranteed by the Federal Housing Administration (FHA) and the Department of Veterans Affairs (VA). GNMA is not taking any credit risk. Instead, losses are absorbed by the agencies from which it obtains the loans.

The only way that the “GNMA model” could be used for the entire mortgage market would be if the FHA were to guarantee every mortgage. However, the FHA is not even capable of properly pricing the credit risk within its own niche. The FHA currently is suffering significant losses, creating large liabilities for taxpayers.

Another proposal is what I refer to as “Wall Street’s Wet Dream.”3 The idea is that a government agency would behave like a late 1980s Freddie Mac. This agency would take all of the credit risk in the securitization process, but it would not hold any securities in portfolio. This would be ideal from Wall Street’s point of view, because it would maximize the circulation of mortgage securities, rather than keep them stuck inside Freddie Mac or Fannie Mae in their own portfolios. The result would be an agency that bears no interest rate risk (which Freddie and Fannie were able to manage successfully), but which bears all of the credit risk (which brought them down).

The “Wall Street Wet Dream” model would ensure that mortgage securities can be traded safely and profitably. The government agency would be responsible for managing the idiosyncratic risk of dealing with mortgage originators. It also would have to price for the systemic risk that arises from fluctuations in regional and national housing markets. Ultimately, this systematic risk would be borne by the taxpayers. Thus, the profits of the securities business would be fully privatized, and the credit risk would be fully socialized. Given the way Washington and Wall Street relate to one another in our society, it is a good bet that this is the model that will gain the most political support.

At this point, there is no private sector firm that can credibly insulate security holders from systemic risk.

Returning to the savings and loan model

Policymakers should consider returning to the mortgage finance system that we had forty years ago. Mortgages were originated and held by firms that financed their mortgage holdings with deposits. These were the savings and loans (S&L).

The savings and loans did not suffer from the principal-agent problems that plague securitization. The loan originator is controlled by the institution that is going to hold the mortgage. The management of idiosyncratic risk is internalized.

To manage systemic risk, regulators could subject depository institutions to capital regulations and stress tests. Mortgage holders that benefit from deposit insurance would have to hold enough capital to withstand a severe drop in house prices.

The biggest flaw with the savings and loan model was that the savings and loans could not manage interest rate risk. When inflation and interest rates leaped higher in the 1970s, the savings and loans were stuck holding mortgages with interest rates that were well below the new prevailing market rates.

There are various ways to make the S&L model work better than it did in the past. One is to conduct monetary policy in a way that stabilizes the inflation rate. In fact, since the early 1980s the Fed has been able to do that.

Another approach would be to encourage variable rate mortgages. These do not have to be the sort of high-risk, “teaser” loans that became notorious in recent years. Instead, they could work more like Canadian rollover mortgages, in which the interest rate is renegotiated every five years. The loans would be on a thirty-year amortization schedule, and they would start out at a market interest rate, rather than an artificially low rate. If interest rates were to rise sharply over any given fiveyear period, we would expect that the borrower’s income would have risen as well, so that the burden of the loan would not be significantly larger.

Defenders of securitization will argue that it is more efficient to fund mortgages in the capital market. I would make two counter-arguments. The first counterargument I would make is that the alleged efficiency of securitization has not been demonstrated in the market. Mortgage securities are the artificial creation of government, starting with GNMA and Freddie Mac. The second counter-argument is that adding efficiency to mortgage securitization serves to divert capital from other uses, and it is not necessarily the case that this capital diversion is best for society. More recently, bank capital regulations severely distorted the cost of holding mortgages relative to holding securities, strongly favoring the latter. In a completely free market, it is doubtful that securitization would emerge, particularly in light of recent experience.

At its best, securitization appeared to lower mortgage rates about one quarter of one percentage point relative to loans that could not be placed into securities. Suppose that government-backed securitization could be put in place to achieve a comparable reduction in mortgage interest rates. Is that an outcome for which we should aim?

Compared with the risks that the government must assume in order to make securitization work, it seems to me that lowering mortgage interest rates by one quarter of one percent is not a commensurate benefit. This is particularly so given the alternative uses of capital. If mortgage rates are a bit higher than they could be under the most efficient system, then some capital will go toward other uses – interest rates charged to businesses or to consumers for other loans will be slightly lower. It is not clear that mortgage loans are better for society than other uses of capital. If it turns out that the “originate and hold” model is not as efficient as securitization for delivering low mortgage interest rates, that would not be a tragedy.

In order to revive securitization, taxpayers would have to absorb large risk. The social gains would be small, or perhaps even nonexistent. The best thing to do with the shattered Humpty-Dumpty of mortgage securitization would be to toss the broken pieces into the garbage.

Arnold Kling is an economist and member of the Financial Markets Working Group of the Mercatus Center at George Mason University. In the 1980s and 1990s he was an economist with the Federal Reserve Board and then with Freddie Mac.

First published in finreg21.com, Reforming Financial Services Regulation in the 21st Century.

1. The recording of this lecture is available at http://www. lse.ac.uk/collections/LSEPublicLecturesAndEvents/ events/2009/20090311t1935z001.htm or http://www.creditwritedowns.com/2009/05/fool’s-gold-gillian-tettlectures- on-shadowy-world-of-derivatives.html
2. Gillian Tett, Fool’s Gold: How the Bold Dream of a Small Tribe atJ.P. Morgan Was Corrupted by Wall Street Greed and Unleashed aCatastrophe (Free Press, 2009).
3. For an example of the type of proposal I am describing, see Harley S.Bassman, “GSE’s: The Denouement.” http://www.zerohedge.com/sites/default/files/RateLab GSE Denouement.pdf