The Paperwork Crisis: When Certificates Broke Wall Street

The Cage

In the back rooms of Wall Street brokerage houses, the department that held customer securities was called the cage — a wire-mesh enclosure, built like a bank teller's window, where clerks handled the engraved certificates that were the legal evidence of who owned what. By December 1968 the cages of lower Manhattan had stopped functioning as filing systems and had become something closer to landfill. Certificates sat in cardboard boxes in corridors. Some firms stored them in bathrooms, in stairwells, on windowsills. At one house the overflow was reportedly kept in wastebaskets, indistinguishable from the rubbish, because the vault had no room left and nobody had time to build shelving.

At the end of that month the securities industry's unresolved transactions — trades where the seller's broker had not delivered the certificate to the buyer's broker within the required five business days — stood at $4.1 billion. These were the "fails to deliver," and the number was not a measure of fraud or of market risk in any conventional sense. It measured a clerical breakdown. Wall Street had sold shares it could not hand over, not because the shares did not exist but because nobody could find the paper.

A System Built for Three Million Shares

American share settlement in the 1960s ran on physical certificates and human messengers, and the design had barely changed since the nineteenth century. When an investor bought 100 shares of General Motors, the transaction was not complete until a specific engraved certificate, bearing a serial number and registered in the seller's name, was physically carried from the selling broker to the buying broker, endorsed by a signed stock power, sent to the issuer's transfer agent, cancelled, and reissued in the new owner's name. Messengers walked bundles of certificates between offices on Broad Street and Wall Street under settlement deadlines. Each step generated its own paper: confirmations, comparison slips, transfer instructions, dividend records.

That machinery had been sized for a market that traded about three million shares on an average day at the start of the decade. It was a market dominated by individual investors making occasional purchases, and the processing load grew slowly enough that firms could add clerks to match it.

Then the volume arrived. The conglomerate boom, the mutual-fund expansion, and the speculative enthusiasm that John Brooks later chronicled as the go-go years pushed daily turnover up by a multiple, not a margin. Average daily volume on the New York Stock Exchange reached roughly five million shares in 1965 and approached thirteen million by 1968, with individual sessions in April of that year running near fifteen million (Brooks, 1973). A processing architecture designed for three million shares a day was being asked to absorb four times that load, using the same certificates, the same messengers, and a clerical workforce that Wall Street could not hire fast enough or train well enough.

Firms responded by hiring in volume and by buying computers they did not know how to use. Back-office headcount at some houses doubled within two years, staffed largely by inexperienced clerks working overtime on records they did not understand. Several firms installed data-processing systems mid-crisis and made things worse: converting a broken manual ledger into a computerised one preserved the errors and removed the clerks who knew where the bodies were buried. Goodbody & Co. lost about $7.5 million in 1969 through a single computer conversion failure. Wells, in the standard scholarly account of the episode, describes the period as one in which the industry's operational capacity and its trading volume diverged faster than any firm's management could reconcile them (Wells, 2000).

Closing on Wednesdays

Regulators and the exchange reached for the only lever that acted directly on the input side: they reduced the number of hours in which trades could be created. Trading sessions were shortened in August 1967 and again in early 1968, with limited effect. In June 1968 the New York Stock Exchange took the more drastic step of shutting entirely one day a week, giving back offices a full business day to process without new transactions arriving.

Wednesday closings ran from 12 June 1968 to the end of December. It remains one of the few occasions in the exchange's history when American equity trading was suspended for reasons that had nothing to do with war, panic, weather, or attack. The market was halted because the paperwork had won.

What finally reduced the fails was not administrative reform. It was the collapse of trading volume that came with the bear market of 1969 and 1970 — a cure that arrived attached to a second and more dangerous disease.

The Certificates That Walked Away

A settlement system that cannot locate its own paper is a settlement system that cannot tell theft from misfiling. As the backlog grew, so did the disappearance of negotiable securities from brokerage vaults, mail rooms, and messenger routes. Estimates of the value of securities stolen from Wall Street firms during the crisis ran as high as $400 million by 1970, much of it attributed to organised crime, which discovered that a bearer-transferable certificate extracted from a chaotic back office could be pledged as loan collateral long before anyone noticed it was missing.

Firms could not reliably distinguish a stolen certificate from one that was merely lost inside their own operation, and in many cases they could not establish which customer's position a given certificate was supposed to represent. Customer account statements had become assertions rather than records. That was the condition of the industry when the market turned.

When the Market Fell

Prices peaked in December 1968, with the Dow Jones Industrial Average near 985, and then declined for eighteen months to a low around 631 in May 1970 — a fall of roughly a third. For brokerage firms the price decline mattered less than what came with it. Volume dropped, and commission income dropped with it, at precisely the moment when firms were carrying the bloated clerical payrolls and half-built computer systems they had acquired to fight the backlog. Costs had been structured for a boom that had ended.

Most Wall Street houses were partnerships with thin capital, and partnership capital was itself often invested in the market that was falling. A firm whose records were unreliable could not compute its own net capital position with confidence, which meant neither the exchange nor the firm's partners knew how close it was to insolvency until it arrived. Operational failure had converted itself into a solvency crisis, and the solvency crisis arrived across the industry at once.

Roughly 160 New York Stock Exchange member organisations disappeared between 1969 and 1971, about half through merger and the remainder through liquidation or resignation from the exchange. More than a hundred brokerage firms failed or were forced into combinations they would not otherwise have accepted (Seligman, 1982). The membership of the exchange had been reduced by attrition rather than by policy.

Goodbody, Hayden Stone, and the Trust Fund

Handling the failures fell to the exchange itself, through a crisis committee chaired by Bernard "Bunny" Lasker, the NYSE chairman, with Felix Rohatyn of Lazard Frères doing much of the negotiation. Their instrument was the exchange's Special Trust Fund, a members' pool intended to make whole the customers of failing member firms. It had never been designed for systemic use. The membership voted an additional $20 million into it in August 1970, by a margin of nearly six to one, and the fund was drawn on repeatedly as firms went down.

Hayden, Stone & Co. — an old retail house that had grossed about $113 million in 1968 and whose business had nearly tripled over the decade — was among the first large casualties. Its back office had failed, the exchange had ordered it to restrict its business, and by early 1970 it was looking for a buyer. It was acquired by Cogan, Berlind, Weill & Levitt, a much smaller firm whose partners included Sanford Weill, who made control of the combined back office his personal project and built from that transaction the acquisition sequence that eventually produced Shearson and, decades later, Citigroup.

Goodbody & Co. was the harder case. It was the fifth-largest brokerage in the country, with about 225,000 customers served through 95 offices, and its collapse would have stranded more retail accounts than the trust fund could plausibly cover. In October 1970 Merrill Lynch agreed to absorb it, completing the merger that December in what was then the largest combination of American brokerage firms ever undertaken. Merrill extracted a hard bargain: the exchange guaranteed it against losses discovered later in Goodbody's records, which nobody could yet read reliably. That indemnity cost the trust fund about $21 million.

One rescue went badly wrong, and it belonged to Ross Perot. At the request of the Nixon administration, which feared what a failure of that size would do to public confidence, the founder of Electronic Data Systems put money into duPont Glore Forgan, one of the largest houses on the street and a significant EDS customer. He later added Walston & Co. Perot committed roughly $97 million over three years, intending to apply EDS data-processing discipline to a brokerage back office. Investigators found that duPont's customers believed they owned some $15 million more in securities than the vaults contained. Perot dissolved the business in 1974, having lost an estimated $60 million to $70 million.

What Replaced the Certificate

Three structural changes came out of the wreckage, and together they rebuilt the plumbing of American securities markets.

The first was immobilisation of the certificate. The exchange had established its Central Certificate Service in 1968 to hold member firms' shares in a single depository and settle transfers between them by book entry, eliminating the physical movement that had caused the jam. Adoption was initially slow and the service itself struggled under the same backlog it was meant to relieve, but by late 1969 it held 464 million shares on deposit. In 1973 the function was reconstituted as the Depository Trust Company, an independent entity with bank status, which became the institution through which American shares are now held and transferred. The certificate did not disappear; it stopped moving.

The second was customer protection. Congress concluded that the exchange's members could not be expected to indemnify each other's customers indefinitely out of a private fund, and passed the Securities Investor Protection Act, which Richard Nixon signed on 30 December 1970. It created the Securities Investor Protection Corporation, an industry-funded body that protects customers against the loss of cash and securities held at a failed member brokerage, initially up to $50,000 per account. The federal government had taken over a function the exchange had performed as a self-regulator, and the self-regulatory model lost an argument it had held since 1792.

The third was supervisory. The SEC delivered to Congress in 1971 its "Study of Unsafe and Unsound Practices of Brokers and Dealers," which examined how firms had been permitted to expand their trading without maintaining records adequate to establish their own solvency. Net capital rules were tightened and made to depend on the demonstrable accuracy of a firm's books. Regulators had learned that a broker-dealer's operational competence is a prudential matter, not an administrative one.

A fourth change followed indirectly. In November 1970 the exchange's president, Robert Haack, used a speech to call for an end to fixed commission rates — an argument aimed at the economics that had let firms compete on branch offices and research instead of on the cost of processing a trade. The position was awkward for an exchange whose chairman ran a specialist firm, and Haack did not survive long in the job. Fixed commissions ended on 1 May 1975 (Welles, 1975).

The Settlement Layer as Infrastructure

Read from the present, the paper crunch is the episode in which American finance discovered that its settlement layer was infrastructure rather than administration — a utility whose failure could close the market and bankrupt its members, and which therefore could not be left to each firm's clerical discretion. Every subsequent construction of that layer rests on the conclusion reached in those cages: that the movement of ownership must be centralised, electronic, and legally distinct from the firms that trade.

A parallel gap surfaced in foreign exchange four years later, when the failure of a German bank in the middle of a settlement day produced the concept documented in the Herstatt settlement risk episode of 1974. The same decade's other great institutional failure, the Penn Central bankruptcy of 1970, struck while the exchange's crisis committee was still triaging member firms, and the two episodes together set the terms on which the Federal Reserve and the SEC understood contagion for the next decade. The durability of what was built is visible in the contrast with later failures: when a market-maker's order system malfunctioned in the Knight Capital meltdown of 2012, four million erroneous orders cleared and settled without operational incident, and the firm died alone.

What the cages could not do in 1968, the depository does now without anyone watching. The certificates themselves mostly stayed where the crisis left them — in a vault, immobilised, never again carried down Broad Street by a messenger running a five-day deadline.