In 2000, Tyco International’s chief executive, L. Dennis Kozlowski, told the company’s head of human resources to forgive the relocation loans of 51 employees, and to pay them enough extra to cover the taxes. The board, he told her, had approved the program.
She asked for documentation. On September 11, 2000, he sent her a memo. It said that “a decision has been made to forgive the relocation loans” of the people on an attached list. It did not say who had made the decision. The employees who received the money were asked to sign confidentiality agreements.
The program paid out $95,962,000 in gross wages. Kozlowski’s share was $32,976,000.
Two years later, after he had resigned, Tyco filed its own account with the Securities and Exchange Commission. The benefits, the company wrote, “were not approved by, or disclosed to, the Compensation Committee or the Board of Directors.” The board that was supposed to watch the owners’ money had not been told.
Less than a cent on the dollar
The money belonged to the shareholders, and there were a great many. At the end of November 2001, Tyco had nearly two billion shares outstanding. Its proxy statement that winter named only two holders known to own more than 5 percent. Kozlowski, counting more than 10 million shares he could buy through stock options, held less than 1 percent.
In widely held companies, owners rarely run the firm, and managers rarely own much of it.
In 1976 Michael Jensen and William Meckling worked out what that does to spending. A manager who owns 95 percent of a firm, they wrote, will spend its money on perks until a company dollar is worth no more to him than 95 cents of his own. Split a dinner bill a hundred ways and the lobster costs you a hundredth of its price.
At under 1 percent, each Tyco dollar spent on Kozlowski cost him, as an owner, less than a cent. In 2001, by the company’s account, Tyco bought an apartment at 950 Fifth Avenue for his use for $16.8 million, then spent $3 million on improvements and $11 million on furnishings. Among them was a shower curtain for $6,000. His share of it was under $60.
Two billion shares, nobody steering
Adolf Berle and Gardiner Means named the pattern in 1932 in The Modern Corporation and Private Property. In 1929, they found, the principal shareholder of the Pennsylvania Railroad, of AT&T and of U.S. Steel each owned less than one percent. Ownership had split into two parts, they wrote, “control and beneficial ownership,” leaving “a large body of security holders” who “exercise virtually no control over the wealth which they or their predecessors in interest have contributed to the enterprise.”
Owners elect a board to do the watching for them. For any one small shareholder, digging further can cost more than his slice of what he might find.
Kozlowski testified that the committee’s late chair had been “fine with” the bonuses. A jury convicted him and his finance chief of grand larceny over four bonuses, the loan forgiveness among them, and New York’s highest court upheld the verdicts in 2008.
Not every split is a crime
Berle and Means did not call the split a mistake. “It is precisely this separation of control from ownership,” they wrote, “which makes possible tremendous aggregations of property.”
Owners can also lose a fortune with no crime at all. In 1995 Disney hired Michael Ovitz as president. Fourteen months later he was terminated without cause, with a severance valued at about $130 million. The board never voted on the firing. Delaware’s courts found that the compensation committee’s process in approving his contract fell far short of “best practices,” and that no director had breached a fiduciary duty.
And the pattern is not universal. In 1999 a study of large companies in 27 wealthy economies found that, except where shareholder protection was very good, few were widely held. They were typically controlled by families or the state, with power well beyond their claim on the profits. There, the split runs between the controlling owner and everyone else.
The roof bid nobody read
Imagine you own one of 120 flats in an apartment building. The owners’ board brings in a single bid to replace the roof: $900,000, in a 60-page packet. Your share is $7,500.
If the price is 10 percent high, you lose $750. Reading it and finding two more quotes would take a weekend, and any saving is split 120 ways. Nine owners come to the vote. The bid passes. Nobody has asked why there was only one.
Control without a stake, on purpose
Berle and Means warned that giving controllers free rein risked “an era of corporate plundering.” OpenAI built the split on purpose, as a safeguard. Its 2023 structure page said the nonprofit board fully controlled the for-profit company, that its independent directors “do not hold equity in OpenAI,” and that Microsoft, despite “a multibillion dollar investment,” had “no board seat and no control.”
On November 17, 2023, the board announced that its chief executive, Sam Altman, would depart, saying he “was not consistently candid in his communications with the board.” Twelve days later OpenAI announced he was back, with a new initial board. Of the four directors named in the first announcement, one stayed on. In a message to staff, Altman thanked Microsoft’s leaders, who “were ready to welcome all of us if we couldn’t achieve our primary goal.”
If your company runs on another firm’s model, you sit nearer the shareholders’ seat: you carry the results, someone else controls the changes. If your provider offers dated model versions, call one by its exact name, keep twenty of your own test cases, and rerun them before you switch.
In October 2025, OpenAI said its recapitalization was complete. The nonprofit, it said, “remains in control” of the for-profit. It also held equity then valued at about $130 billion.