The Birth of the Mortgage-Backed Security, 1968–1986

A savings and loan in Ohio in 1974 held its mortgages in a filing cabinet. Each was a thirty-year promise from a named household at a fixed rate, each legally distinct from the others, and if the institution needed cash it could not sell them. No buyer existed, because buying meant hiring someone to read several hundred credit files, verify several hundred appraisals, and then collect several hundred monthly payments by post. A mortgage was an asset you originated and then lived with until it matured, prepaid or defaulted.

Within about a decade that stopped being true, and the change was not a discovery so much as a sequence of legal permissions. A federal guarantee arrived in 1970, a tax provision in 1981, a structuring technique in 1983, a statute overriding state investment law in 1984, and a tax code amendment in 1986. Each removed one reason an institutional investor could not own a piece of somebody else's mortgage. By the end, the thirty-year American home loan had been converted into a bond that a Japanese insurer could buy on a screen.

Splitting Fannie Mae

Fannie Mae had existed since 1938 as a government body that bought mortgages insured by the Federal Housing Administration, holding them on a balance sheet that counted against the federal budget. By the late 1960s, with Vietnam and the Great Society competing for the same accounting space, that balance sheet had become politically inconvenient.

The Housing and Urban Development Act of 1968 solved the problem by division. Fannie Mae was reconstituted as a shareholder-owned corporation, chartered by Congress but off the federal books, licensed to buy conventional mortgages meeting stated underwriting standards. The government functions were transferred to a new entity inside the Department of Housing and Urban Development: the Government National Mortgage Association, which nobody has ever called that twice.

What Ginnie Mae received in the split was narrow and, as it turned out, decisive. It could not buy mortgages at scale or hold a large portfolio. It could guarantee, with the full faith and credit of the United States Treasury, securities issued by approved private lenders and backed by pools of FHA-insured and Veterans Administration-guaranteed loans. The agency was not a buyer. It was a credit substitution machine.

In 1970 it guaranteed the first such security. The structure was a pass-through: a lender assembled a pool of government-insured mortgages, sold undivided interests in the pool to investors, and forwarded the principal and interest collected from borrowers each month to the holders of those interests, less a servicing fee. Ginnie Mae guaranteed that the payment would arrive on time whether or not the borrowers paid, which meant a holder did not need to know anything about the households in the pool, or about the lender, or about the neighbourhood.

That is the whole of the innovation, and it is worth stating plainly because it is so often described as financial engineering. There was no engineering. A government agency agreed to absorb the credit risk of a defined pool, and in doing so converted several hundred separate acts of due diligence into one act of trust in the Treasury.

The Certificate Nobody Could Buy

Conventional mortgages, the ones without a federal insurance wrapper, were a larger market and they had no such guarantor. The Emergency Home Finance Act of 1970 created one: the Federal Home Loan Mortgage Corporation, chartered to serve the savings and loan industry and known from birth as Freddie Mac. In 1971 it issued its first Mortgage Participation Certificate, the first pass-through security assembled from ordinary uninsured home loans.

Volume stayed modest through the 1970s, and the obstacle was not investor appetite but law. American institutional investors do not choose their holdings freely. A state-chartered insurance company, a savings bank, a public pension fund and a trustee each operate under legal-investment statutes that enumerate permitted assets, and those statutes had been drafted when the permitted assets were government bonds, corporate bonds and real property. A participation certificate in a pool of mortgages was none of these. Securities laws in individual states, the blue sky laws, added registration requirements on top.

Richard Green and Susan Wachter's account of the period treats this as the central fact about the American mortgage: its form is a product of legal and regulatory choices rather than of borrower preference, and the long fixed-rate prepayable loan that dominates the United States exists nowhere else at comparable scale (Green and Wachter, 2005). The instrument had to be made legal before it could be made liquid.

Salomon Brothers Builds a Desk

Lewis Ranieri had joined Salomon Brothers in 1968 in the mailroom, part-time, and by the middle 1970s was working on an unfashionable product. Mortgages were regarded on Wall Street as a sleepy utility business belonging to thrifts. Ranieri's insistence was that the product was analytically tractable rather than sentimental. "Mortgages are about maths," he told colleagues, and he built a research department on that premise.

In 1977 Salomon and Bank of America issued the first private-label residential mortgage-backed security, a pool of conventional loans with no government guarantee behind it. Ranieri coined the word securitising during the same period, which is the rare case of a piece of financial jargon whose author is known. His own retrospective account describes the early market as one where the product existed and the buyers were legally barred, a problem no amount of salesmanship resolves (Ranieri, 1996).

What resolved it was interest rates. A savings and loan funded thirty-year fixed-rate mortgages with deposits that repriced continuously, a maturity mismatch that is profitable while short rates sit below the mortgage coupon and lethal when they do not. By 1981 the mismatch had inverted violently, and institutions across the industry held loans written at 8 or 9 per cent while paying depositors far more, a squeeze examined in the savings and loan crisis and produced directly by the Volcker disinflation.

An institution holding an 8.85 per cent mortgage from 1977 in a year when the same loan was being written at 16.63 per cent was holding an asset worth considerably less than its face value. Selling meant booking the loss, and booking the loss meant reporting capital the institution did not wish to report.

The Tax Break That Made a Market

Congress removed that obstacle on 30 September 1981. The provision allowed a thrift selling mortgage loans to amortise the resulting loss over the remaining life of the loans rather than recognising it at once, converting an immediate hole into a slow drip. To claim the benefit an institution had to actually sell. Selling and then buying somebody else's loans at similar prices left a thrift with much the same economic position, a refreshed tax posture and a realised loss on paper.

The consequence was turnover in the hundreds of billions of dollars, almost none of it driven by any view on housing. Thrifts sold because the tax code paid them to, and bought because they wanted the assets back. Salomon Brothers stood in the middle. Michael Lewis, who arrived at the firm later in the decade, recorded that traders on the mortgage desk estimated somewhere between 50 and 90 per cent of their profits came from simply taking the other side of the thrifts' trades (Lewis, 1989).

That is an unflattering description of a market-making franchise, and it is also how the secondary mortgage market acquired the depth that made it useful. A dealer who transacts constantly develops inventory, pricing conventions, analytics and a client list. By the time the tax-driven churn subsided, mortgage pass-throughs had become an asset class with a trading infrastructure rather than a product in search of one.

Fannie Mae, which had spent the 1970s as a portfolio lender borrowing in its own name, issued its first mortgage-backed security in 1981 and adopted the same guarantee-and-pass-through model its sibling had pioneered. Ginnie Mae issuance reached $46 billion in 1985, and the agency accounted for 54 per cent of all MBS outstanding that year.

Cutting the Pool Into Tranches

A pass-through had one defect that no guarantee could fix. American borrowers may prepay a mortgage at any time without penalty, and they do so most enthusiastically when rates fall, which is exactly when an investor would prefer to keep a high coupon. Every holder of a pass-through shared that prepayment risk pro rata. Pension funds wanting long duration and money funds wanting short duration were offered the same uncertain average life, and neither got what it needed.

In 1983 First Boston and Salomon Brothers built a structure for Freddie Mac that solved it by redistribution. The collateralised mortgage obligation held the same pool but issued several classes of bonds against it, ordered by seniority in receiving principal. Early principal payments went to the first class until it was retired, then to the second, and so on. The resulting securities had genuinely different maturity profiles: the first class was short whatever borrowers did, the last absorbed the bulk of the uncertainty and was priced for it.

Prepayment risk had not been reduced. It had been sorted, and sold to whoever was least troubled by it. The same logic of tranching a pool by seniority was applied to credit rather than timing a decade later, when JP Morgan built the structures traced in the invention of the credit default swap.

The Law Catches Up

Two statutes then removed the remaining barriers, and both were technical enough to have attracted no public attention whatever.

The Secondary Mortgage Market Enhancement Act of 1984 addressed the legal-investment problem directly. Its Section 106 pre-empted state blue sky registration requirements and state legal-investment restrictions for investment-grade mortgage-related securities, so that state-chartered banks, insurance companies, pension funds and trustees could buy them on the same footing as government bonds. States retained a right to reassert their own limits within seven years; few did. A category of assets that had been legally unavailable to most of the American institutional investor base became available by federal pre-emption.

The Tax Reform Act of 1986 fixed the second problem. A multi-class structure like a CMO sat awkwardly in the tax code, risking treatment as a taxable corporation and therefore a second layer of tax between borrower and investor. The Act authorised the real estate mortgage investment conduit, available from 1987, which held a fixed pool of mortgages, issued multiple classes of interests and paid no entity-level tax.

What the Structure Assumed

Each step had been a narrow fix to a specific impediment, and the aggregate was a system resting on an assumption that nobody had been asked to defend: that the credit quality of the underlying loans was somebody else's settled problem. Ginnie Mae's guarantee made that true by statute. Freddie Mac and Fannie Mae made it true by underwriting standards enforced at purchase. The private-label market made it true by subordination and, later, by rating agency opinion.

The FDIC's retrospective on the decade noted that the thrift industry's troubles had been a problem of interest-rate mismatch long before they became a problem of credit, and that the regulatory response addressed the second while the first had already been solved by the capital markets (FDIC, 1997). Securitisation was the successful part of that story. A thrift that sold its loans and bought securities no longer bore thirty years of duration risk on a five-year funding base.

What the machinery did not contain was a party whose own money depended on whether the loans were any good. An originator who sold every loan within ninety days had no thirty-year interest in the borrower's income being real. That gap was invisible while government agencies or their underwriting standards sat behind nearly every pool, and it became the central mechanism of the 2008 crisis once private-label issuance grew large enough to operate without them.

Ginnie Mae was founded in 1968 and did not pass $1 trillion of securities outstanding until July 2010, forty-two years later. The instrument it invented in 1970 took a decade to become legal for most investors to own and another to become the default form in which American housing debt was held. The filing cabinet in Ohio was emptied not by anyone deciding that mortgages should be bonds, but by a series of officials each removing one sentence from a statute.