How Do Companies Allocate Capital? Tom Murphy's Five Choices

Companies allocate capital by choosing, for every dollar the business produces, among five uses: reinvest it in the operation they already own, buy another business, pay down debt, repurchase their own shares, or pay it out to owners. Good allocation is ranking those five honestly, year after year, so that the cash flow belonging to each remaining owner grows.

Tom Murphy ranked them better than almost anyone. In 1954 he was sent to Albany, New York, to run a struggling television station housed in a former retirement home for nuns on a farm road in North Greenbush, near Troy.¹ The story the company kept from those years is that he had the building painted — but only the two sides that faced the road, where advertisers driving up would see them. The other two could wait. Four decades later the company that grew from that station, Capital Cities, had bought the American Broadcasting Companies, a business several times its size, and in 1995 agreed to sell the combined enterprise to Disney for about $19 billion.² The half-painted building and the ABC purchase are the same decision at two scales.

Capital Cities, ABC, CBS and Disney appear here as history, to illustrate a method. Nothing in this essay is a recommendation to buy or sell any security.

Five doors for a dollar, three ways to fund them, and the rank nobody keeps

Every company that produces cash faces the same short menu. A dollar of free cash flow — the cash left after the bills are paid and after the spending required to keep the operation running, which is not the same pile as reported profit — can go through one of five doors. Back into the business that produced it. Into another business. Against the debt. Into the company's own shares. Or out to the owners as a dividend. That is the entire universe of capital allocation decisions, whether the company has four employees or forty thousand.

The uses number five. The sources number three: cash the business already generates, money borrowed from lenders, or money raised by selling new shares to someone new. That third source is the menu run in reverse, and it is the one that slips past boards most easily, because no cash appears to leave the building when it is used.

Most management teams never rank any of this. They reinvest because growth is what executives are rewarded for showing. They acquire because a banker arrived with a book. They pay a dividend because the board has come to expect one. They issue shares because issuing shares feels free. The money moves every quarter, and nobody has asked the only question that matters: which of the five uses will leave the owners with more cash per share a decade from now.

Murphy's contribution was to turn the menu into a short ranking and then keep it for thirty years. Hold operating costs below every competitor's, so the business throws off more cash than rivals do from the same revenue. When a genuinely good property comes up at a sensible price, borrow to buy it. Use the cash that property produces to retire the debt quickly. When nothing worth buying is available and the company's own shares are cheap, buy those instead. And almost never issue new shares, because that hands a permanent slice of every future dollar to someone who did not earn it.

Notice what the first rule is doing. It is not an allocation decision at all — it decides how large the dollar is before anyone allocates it. The other four route that dollar. Debt repayment and buybacks are not personalities or goals; they are simply where a dollar goes when the best acquisition is not on offer. Dividends barely appear in the story: Capital Cities paid a token one that scarcely changed for years, because there was almost always something better to do with the cash. The ranking is the method. A preferred door, used out of habit, is how the method gets lost.

Paint only the sides the advertisers can see

The cost rule deserves the longest look, because it is the one owners who have run a company will recognise instantly, and because nothing downstream of it works without it.

Capital Cities owned television stations, radio stations, newspapers and, later, cable systems. A local television licence in the decades before cable and streaming was an unusual asset. A government permit limited the number of rivals in each city, advertisers had few alternatives, and once the transmitter and studio existed, the cost of broadcasting another hour was close to nothing. A well-run station could convert roughly half its revenue into pretax profit. The licence was the barrier; the cash was the proof the barrier was real.

But a great industry does not produce a great company on its own. Plenty of station owners spent the licence on executive dining rooms, thick headcount and the comfortable assumption that the permit would always cover the bill. Murphy and Daniel Burke — Murphy did the deals, Burke ran the operations — built Capital Cities on the opposite assumption. Headquarters was a few dozen people overseeing thousands of employees, with no public relations department and no layer of vice presidents whose job was to watch the people doing the work. Station managers were handed their budgets, sat down once a year with what they had promised against what they had delivered, and otherwise left alone. Decentralisation was not a mood or a management fashion; it was a refusal to spend cash on supervision that did not produce more cash. Warren Buffett, who watched the pair for decades, described them as the finest two-person management combination he had ever encountered.³

The result was a measurable gap on the same dollar of advertising revenue. Capital Cities properties simply kept more of what they took in than comparable properties in other hands, and that gap is the raw material of every later choice. A company with structurally higher margins can pay the same price for an acquisition and earn more from it, carry the same debt and retire it faster, and repurchase shares while a competitor is still covering overhead. Cost discipline is not a virtue layered on top of capital allocation. It is where the capital comes from. A company cannot allocate cash it has already spent.

A sceptic will object, reasonably, that this is the industry talking — that any careful owner of broadcast licences in those years was standing in a river. The useful comparison is the company standing in the same river and treating the cash differently. When Murphy became chief executive in 1966,⁴ CBS was a dominant media company many times Capital Cities' size. Over the following decades CBS routed network cash into a toy company, the New York Yankees, a publishing house and an expensive landmark headquarters in Manhattan,¹¹ supported by a corporate layer thick with presidents and vice presidents. Murphy's stated ambition was the opposite: not to build the longest train, but to arrive at the station first using the least fuel. In 1995, Capital Cities/ABC agreed to sell itself to Disney for about $19 billion.¹² In the same year, Westinghouse agreed to buy CBS for roughly $5.4 billion.⁵ Two companies collecting the same kind of advertising dollar, ending in very different places.

Owners who have run a business know exactly what the cost rule resists. After a good year, expenses drift toward revenue: headcount, travel, a second office, a layer of people whose work is to report on the work. Every one of those is an allocation decision — a dollar sent back through the first door that could have gone out one of the other four. Murphy refused to let success change the habit, and the two unpainted walls are the memory the company kept of where that was decided.

Borrowing against ABC, and letting ABC retire the loan

In March 1985, Capital Cities agreed to buy ABC for roughly $3.5 billion.⁶ ABC's revenue was about four times that of the company buying it,⁷ and it was the first time one of the three national networks had changed hands.⁶ The press called it the minnow swallowing the whale, and Murphy used the line himself.

How a smaller company pays for a much larger one is the whole lesson. Most of the money came from banks. A substantial further piece came from Buffett's Berkshire Hathaway, which paid $517.5 million for three million newly issued shares — a stake of roughly eighteen percent,⁸ and one of the very few occasions on which Murphy sold stock. He did it in a single block, to a buyer he had known for years, at a price he was comfortable with, in exchange for a permanent partner rather than a passing crowd. Federal rules then capped how many stations one company could own, so the combined enterprise had to divest properties it would not be allowed to keep, including its cable systems, which went to the Washington Post Company.⁹ Those proceeds went straight against the loan.

Then the ranking did its unglamorous work. ABC had been run in the expensive network style, and Burke's people arrived without a slogan — only the same insistence that every cost justify itself. Oversight staff at the owned-station group was cut to a fraction of its former size. The Manhattan headquarters building was sold. Margins at the network's stations climbed toward the levels Capital Cities took for granted, and the additional cash, together with the divestiture proceeds, was aimed at the debt. Within a few years, the loan that had made the deal look reckless was largely gone. What remained was a national network, a group of large-market stations, very little debt, and once again more cash arriving than the business could sensibly reinvest.

The sceptical question here is the right one: is this not simply leverage, the instrument that has sunk so many ambitious acquirers? It is leverage, and the distinction is precise. Murphy borrowed against businesses whose cash flows were already durable and observable — licensed stations with decades of operating history — and he borrowed with an explicit plan to repay from those cash flows rather than from a resale or a rising share price. Debt used to buy an asset that already produces cash, and retired by that cash, is a bridge. Debt used to buy growth that has not yet produced cash is a bet, which is why self-funding businesses tend to compound more reliably than debt-dependent ones. The ABC deal was a bridge. It looked like a bet only to observers who had not studied what the properties earned.

There is a second discipline inside the deal, and it is the one that kept Murphy from doing many more like it. His test for any acquisition was that the property should return double digits after tax over a ten-year holding period, calculated as though the purchase had been made entirely in cash — no borrowed money allowed to flatter the arithmetic. Most properties offered to him failed that test, and he passed. He entered auctions and lost them, because an auction is a machine for locating the buyer willing to overpay. Earlier purchases that did clear the bar — trade publications, Walter Annenberg's broadcast stations, local newspapers — were negotiated directly with owners who cared who inherited their work. ABC cleared it for the same reason. The willingness to wait years between purchases attracts the least admiration in Murphy's record and does the most work. A buyer who must do a deal this year has stopped ranking doors; the calendar is allocating the capital.

The door most companies use backwards

When no property passed the test — the normal state of affairs — the cash still arrived every quarter, and Murphy turned to the fourth door. Over his tenure, Capital Cities repurchased a large portion of its own shares, and bought hardest during the mid-1970s to early 1980s bear market¹⁰ when broadcasting stocks were marked down alongside everything else.

A repurchase is the simplest acquisition a company can make: it buys out the owners who want to leave, and the owners who stay hold a larger share of the same business. There is no integration risk and no due diligence problem, because management already knows the asset perfectly. But the arithmetic only works in one direction. Buying shares below what the business is worth transfers value to the owners who remain; buying them above that moves value the other way. Price is the entire question, which is why indiscriminate repurchasing — buying every quarter regardless of valuation, often to offset shares handed out as compensation — is a different activity wearing the same name. This publication has examined that distinction at length in the record of Henry Singleton at Teledyne; the point here is narrower. Murphy bought his own shares the way he bought stations: only when the price made the return obvious.

The mirror image is the sixth move, the one that does not appear on the menu. Issuing new shares feels costless because no cash leaves the treasury, and it is the most expensive currency a company owns. Every new share is a permanent claim on every future dollar of free cash flow the business will ever produce. Murphy's near-absolute refusal to dilute is the quiet reason the other four rules compounded for the people who stayed.

What the ranking actually compounds

The through-line in all of it is a single number: the free cash flow attributable to each remaining ownership slice, and whether that number grows. Cost discipline raises the cash the business produces. Disciplined acquisitions add cash bought at a sensible price. Debt repayment moves the claim on that cash from lenders back to owners. Well-priced buybacks shrink the number of slices the cash is divided among. Refusing to dilute stops new slices being created. Five different mechanisms, one destination. That is why the returns from a business like Capital Cities did not depend on the stock market agreeing to pay a higher multiple; the cash per owner grew, and the valuation eventually followed what was already true.

For owners now deciding where long-term capital goes, this converts into a short set of questions that can be asked of any business, public or private, without a finance background. Where did last year's cash actually go, and who decided? What did management pay for what they bought, and against what hurdle? Has the share count risen, fallen, or quietly crept up through compensation? Is debt being used to buy cash that already exists, or growth that has only been described? And does the company's cost base look like the habit of a lean operator or the drift of a comfortable one?

For those still running something, the test is older and simpler. Every expense approved after a good year is an allocation decision competing with four other doors, and the operators who survive the lean years are the ones who ran lean when they did not have to. Murphy's record is not a formula to copy — the regulated broadcasting economics that produced those margins no longer exist in that form. What survives is the habit underneath it: knowing there are only five doors, refusing to let any of them become a default, and judging each dollar by whether it leaves the remaining owners with more cash later than it would have anywhere else.