The Price List Nobody Could Open

The most famous memo in American industrial history says a company counted the dead. In the fall of 1977, Mother Jones published “Pinto Madness,” and the country learned the shape of the story it would keep: a car with a dangerous fuel tank, a fix that cost eleven dollars a vehicle, and a memo that priced the burn deaths against the repair and found the deaths cheaper. The memo became the Pinto’s death warrant in the public mind, the smoking gun with a signature, the proof that a corporation had done the arithmetic of murder and filed it.

I went looking for the memo. What I found was stranger, and more useful.

The document is eight pages, and it carries no date and no addressee on its face. Its title is “Fatalities Associated with Crash-Induced Fuel Leakage and Fires,” and its title sheet names two engineers, E. S. Grush and C. S. Saunby. It went into the record as an attachment in two federal dockets, which is to say it was a public argument in a rulemaking, the kind of filing an industry makes when it wants to talk a regulator out of a rule. It counts 180 burn deaths, 180 serious burn injuries, and 2,100 burned vehicles. It values them at $200,000, $67,000, and $700. It weighs the total against an $11 fix across 11 million cars and 1.5 million light trucks, and it finds the benefits come to $49.5 million and the costs to $137 million. Nowhere in its eight pages does the word Pinto appear.

The legend needed the paper to be something it was not: a secret decision to let people burn, made in a closed room, because the fix cost more than the dead. The document is a comment on a proposed rule, in a public docket, about a hazard measured in rollovers across the whole American fleet, priced with figures the paper itself credits to the government. The famous trial kept it out of evidence. Gary Schwartz read the whole record for the Rutgers Law Review in 1991 and took the legend apart document by document, and the correction never traveled, because the myth fits in a headline and the correction needs a docket.

All of that is the correction, and the correction is not the interesting part. The interesting part is where the number came from.

The paper’s afterlife is the myth, and the myth’s appetite is worth measuring. An archival recording aired decades later by NPR estimated 500 Pinto burn deaths in six years, a figure far past the federal count of reported cases, which Schwartz gives in a footnote at page 1030: 38 reported rear-impact fire or leak cases, with 27 deaths and 24 burn injuries, from 1971 through 1977. And the second memo, the one the legend does not know it means, traveled its own road. Edward Ivey’s 1973 worksheet at General Motors valued each fatality at the same $200,000, spread it across about 41 million GM automobiles, and printed $2.40 per automobile in current operation. It recommended nothing. It surfaced in public in 1999, in a Los Angeles courtroom, in a case belonging to a mother, her four children, and a family friend, all burned when their 1979 Malibu was rear-ended on Christmas Eve, and all alive. The jury awarded $107 million in compensatory damages and $4.8 billion in punitive damages. The judge cut the punitive share to $1.09 billion. The Associated Press noted that the documents in the case never mentioned the 1979 Malibu itself. The distance between the memo in the courtroom and the car it never named is the same distance the Pinto paper kept, and a fair reader declines to build a second legend on it.

The paper’s own face answers. A reference on its last page names the highway safety agency’s preliminary societal-cost report of April 1972. The $200,000 was the government’s figure before it was Ford’s. The companies counted with the government’s numbers, in the open, in a filing anyone could read. That detail is what the legend could not carry: the arithmetic the myth calls murder was the government’s own arithmetic, handed to the companies for use in exactly this kind of filing.

So I went looking for the 1972 report, the government’s price list itself.

The paper’s other figures deserve their sentence first, because the ladder is the document’s own shape. The 1973 filing priced a serious burn injury at $67,000 and a burned vehicle at $700. A death, an injury, a truck: each with its price, in the government’s own currency of 1973, each doing work in the same column. A reader who wants to hate the arithmetic should have to look at the whole ladder, because the ladder was never a murder price. It was a benefits table, built for a rule, by the method the government had just taught.

It exists. The government’s own catalog lists it, under its Superintendent of Documents number. And the catalog’s listing carries a note that reads, in its own flat government language, that the report was not available for sale or distribution. The first price of an American death, in this machinery, was printed by the government and then kept from anyone who might want to read it. No reader can open it online today. Its itemization stays unquoted, and the figure the companies borrowed is known only from the borrowing. A price list that nobody could open: the ghost at the bottom of the legend.

Four years later the agency’s own revision of the figure, written by Barbara Moyer Faigin and dated December 1976, explained what the new method removed. The older estimate had allowed something for pain and suffering, for what a death feels like. The revision declined to. Read the two facts side by side, slowly: the government priced a death, would not sell the price list, and then corrected the method by taking the suffering out.

What the number became has its own history, and it is short enough to tell. Thomas Schelling’s 1968 essay made the move from lives to risks, and its title carries the whole argument in one clause: “The Life You Save May Be Your Own.” The statistical life is an identified life whose name you do not know. The economist’s question was the government’s own: what should the public spend to save a life it cannot name. In H. Spencer Banzhaf’s history of the concept, the idea was born at the RAND Corporation, where the Air Force objected to analyses that gave pilots’ lives no weight. The concept was born when someone priced a life at zero, and everything since is the correction.

The modern figure is the government’s own, and the government concedes the label confuses. The Department of Transportation’s guidance defines it as the value of safety improvements that, summed across a population, prevent one expected death: a measure of what people will pay for small reductions in risk, gathered up until the reductions add to one fewer death. Nobody is buying a life. Thousands of people are each paying a little against a small chance. The figure prices risks, not persons, and the label says otherwise, and the Department’s own document concedes that the conventional label has confused the public and decision-makers alike.

W. Kip Viscusi’s method, as his publisher summarizes his 2018 book Pricing Lives, asked how much more workers demand for dangerous jobs. The wage premiums became the modern figure’s foundation: divide the premium by the risk, and a worker’s price of a danger becomes a population’s value of a death prevented. The premium a worker demands for a dangerous job is a real price, observed in wages, paid by somebody. The figure built on it is an inference, and the inference is the government’s. By the publisher’s account, the older cost-of-death method had undervalued life, the move to the new method raised the government’s benefit figures by an order of magnitude, and corporations were urged to abandon the older arithmetic. The second meaning is now the government’s own, and its label, the Department concedes, confuses.

The series runs like this, each number dated and labeled: $9.1 million for a 2012 base year, $13.2 million in 2023 dollars by a 2024 memo, $13.7 million for 2024, and $14.2 million for analyses using a 2025 base year, under guidance effective March 20, 2026. The guidance does not offer its figure as a point; it tells its analysts to test a range. The candor is the document’s own: the government does not claim to know the worth of a life within five million dollars in either direction, and it prints the range where anyone can read it.

The Bureau of Labor Statistics’ own table puts the consumer price index’s annual average at 44.4 for 1973. The 2025 average stands at 321.943, with October 2025 missing from the series because of the lapse in appropriations. By that arithmetic, $200,000 in 1973 comes to about $1.45 million in 2025 dollars. The Department’s $14.2 million is about 9.8 times that, and the 9.8 divides two different instruments. The 1973 figure measured an economic loss. The 2025 figure measures what the living will pay to shave a risk. No reader should take the gap to mean a person is worth nearly ten times more than they were; the gap is a change of method as much as of prices and incomes, and the two instruments should never have shared a label.

The first is the economic loss of a death. The second is what people will pay for a small reduction in risk. The third is a company’s expected liability: the chance of a claim times the claim’s cost, summed across the fleet, the only one of the four a seller computes for itself, and the only one shaped like a part’s price. The fourth is compensation after the harm, paid to identified people, audited by the acceptance of the paid. The legend of the memo treats the first meaning’s figure as the third meaning’s proof. The government’s own label confuses the second with the first. The funds pay the fourth. And the recall notice, the one instrument in the machinery that must be read by a person with a name, prices nothing.

The government has stood on the seam once in public, and it is worth watching what it did there. In December 1994, with the government’s own finding on the pickups’ tanks vacated for procedural reasons, the Secretary of Transportation traded the recall for a fund: about $51 million in safety programs over five years, and the government’s investigation, its attempt to compel a recall, and the company’s lawsuit all ended together. The Secretary defended the trade in public, and his defense was arithmetic: the programs would start saving lives before a court fight could end. The critics said the trucks would stay on the road, and the trucks stayed. The error was in the denominator: the fund’s lives were statistical strangers, and the trucks’ owners were identified people the government had spent two years learning to count. A federal appeals court, reviewing the deal the next year, wrote that the defect finding had come over the objections of some of the agency’s own engineers, and that the company admitted no liability.

Where is the wrong, then, if the arithmetic was the government’s own? Put the five questions of the docket plainly: what it counts, what it leaves out, who reads it, what it decides, who can answer. The 1973 filing counted the deaths at the government’s own figure and the fix at the company’s, and it counted the rule’s benefits at $49.5 million against costs of $137 million. The people whose deaths were in the table were not in the docket. That is the wrong worth naming. The arithmetic was honest work; the audience was missing. A price list no one can open fails the same test whatever its figure was, and the 1972 report’s catalog note is the smallest and most damning document in the machinery: the government’s concession, printed in advance, that the count would not bear looking at.

The fourth meaning deserves its own paragraph, because it is the one place in this machinery where the arithmetic is paid to identified people and the totals are audited by the acceptance of the paid. General Motors’ compensation fund for the ignition switch, at its final count in December 2015, found 124 death claims eligible out of 473 filed, and paid $594.5 million in all, at a floor of one million dollars for each death, plus $300,000 for each surviving spouse and dependent. Every death offer was accepted. The company’s own public number for the dead had long stood at 13, because its counting rule admitted only head-on crashes in which the air bags failed. Two counts, two different rules: the company’s rule excluded the nurse on her birthday, and the fund’s rule counted her. The counting rule is the instrument in its cruelest form, a definition of which deaths the number will include, and the difference between the two counts is the difference between a defense and a fund.

Takata’s fund for the injured paid by points, a severity grid with a dollar figure attached, the injury priced by its severity at $64.34 a point. The first payments, nearly $9.8 million to 102 people, went out in August 2019, and the individual payments ran from $643.40 to $608,013. The administrator of GM’s fund was Kenneth Feinberg; he had been a trustee of the Dalkon Shield claimants’ trust before it, and he was named to administer Takata’s restitution fund after. The compensation of the identified dead is a profession, and the profession has a roster. A compensation fund is the seller’s arithmetic turned to the other side of the seam: it applies rules, counts, and tariffs, the grammar of the statistical dead, to identified people with names and claims.

That seam is the whole map. Statistical deaths on one side, identified ones on the other. The memos of 1973 priced the statistical dead, in the open, at the government’s own figure. The funds of 2015 and 2019 paid the identified ones, by their own rules. The government’s 1994 settlement sits exactly on the seam: it took a finding about identified trucks and converted it into a fund aimed at statistical lives, and the appeals court reviewing the deal wrote that the programs were unrelated to the pickups. Each instrument lives on one side of the seam, and the recall notice is the only instrument that must cross it, because its subject is identified, the owner, and its reason is statistical, the count. A memo can count a statistical death. A notice has to find an identified person.

That difference is where all of this lives.

The legend of the memo will outlive me, because a legend wants a villain with a signature, and a villain with a signature can be punished, and the punishment ends the story. The document is less satisfying and more useful. Companies do count. The counting is on paper, the paper has a home address in a federal docket, and the figures on it were borrowed from the government’s own arithmetic, which was printed and never sold. What the myth got right is the thing worth saying. What it got wrong is where the guilt lives. The truth needs a file, because the truth of the count is a system that prices death in the open and calls it comment on a proposed rule, and a file has no ending.

I have spent the better part of two years inside this machinery, reading the dockets and the wires and the owner letters, and the book that came out of it is published this month: No Injuries Have Been Reported: The Recall Notice, from the Wire to the Update, by David Boles, from David Boles Books Writing and Publishing. It reads the recall as a drafted document across eighty-nine years of American paper, from the 1937 telegram to the over-the-air update, and it keeps its figures dated and its sources named in the sentence that relies on them. If this essay’s ground interests you, the book walks it end to end. It is at BolesBooks.com, and on Amazon for Kindle and in paperback.