How to evaluate management

Why is capital allocation the CEO's most important job?

Because it is the decision that compounds. Buffett put the arithmetic in his 1987 letter: most chief executives rise through marketing, production, engineering or politics, and few arrive skilled at deploying capital; yet a CEO whose company retains earnings equal to 10% of net worth will, after ten years, have decided where more than 60% of all the capital in the business went. Strategy gets the press release. What was done with the cash is the record.

William Thorndike made the same point from the other direction in The Outsiders. A chief executive has five ways to deploy capital: reinvest in the business, acquire another, pay a dividend, pay down debt, or buy back stock; and three ways to raise it: cash flow, debt, or new shares. The eight CEOs he studied, Singleton at Teledyne, Murphy at Capital Cities, Graham at the Washington Post, Buffett among them, ran very different businesses and beat the market by multiples for the same reason: they treated those eight levers as the job, measured themselves per share rather than by size, bought their own stock only when it was cheap, and left operations to people who were better at them. Michael Mauboussin's 2015 paper on capital allocation went through the evidence behind each lever and found what owners suspect: acquisitions destroy value on average, buybacks create it only below intrinsic value, and the businesses that compound are the ones whose capital allocators can tell the difference.

Our own process asks for the same qualities in plainer words: management that is in it for the long run, reinvests to widen the moat rather than to please the next quarter, is honest, and is skilled in capital allocation. The question this essay answers is how that is checked, by the service and by you.

What does Invest Board read when it scores management?

An agent set up as a Buffett-style investor is given the numbers that describe the record: return on invested capital and return on equity, free cash flow against net debt, insider ownership, the CEO's pay against revenue and the CEO's name. It reads passages from the company's own filings and searches the proxy statement and the 8-K appointment notices for who runs the company and since when. Then it answers seventeen questions, candour and capital allocation first, and writes a hundred words with a score from 0 to 10.

The seventeen questions are Buffett's, and they are worth listing, because they are the checklist a reader can run without any software. Do they discuss mistakes openly? How well do they reinvest earnings or return them? Do they prefer long-term value to short-term results? Do decisions follow return on invested capital? Do they own the company, with meaningful insider ownership? Is the record one of execution? Are they honest? Do they stay inside what they understand? Is the balance sheet conservative? Is ROIC in the range of 10 to 15% or better, and not diminishing? Is ROE as good and as stable? Is the CEO's pay reasonable against the size of the business? Do they lead change or resist it? Does reinvestment strengthen the moat? The instruction that governs all of them is the one on our method page: actions over words. A management team is read by what it did with the cash, not by what the letter says about the cash.

Tenure gets its own search because it changes what the numbers mean. The agent is told to find the appointment date in the proxy statement or the 8-K that reports a departure or an appointment, and to treat the company's own record of its officers as authoritative. If the chief executive arrived two years ago, the decade of returns on the page is the previous management's report card, and the score is held down until the new one has written their own.

How is the management score built?

In three steps. A base score for capital allocation, ROIC, ROE and the track record. Then continuity: a CEO replaced in the last year costs three points, one replaced in the last two years costs two, and a new CEO cannot score above 8, because the historical returns belong to the predecessor; a founder with ten years in the chair earns half a point. Then bounded credit for proven turnaround, proactive change, culture, delegation and strategic focus. The bands read 0 to 3 poor, 4 to 6 average, 7 to 8 strong, 9 to 10 exceptional. The score carries 15% of the overall quality score and is passed on to the business model, optionality, valuation and bear case readings.

Two things about that construction are worth being clear on. The score is a judgement, written by an agent that reads and reasons, with the source beside each claim; it is not arithmetic on a spreadsheet, and the bands are the agent's instructions rather than a formula. And the continuity rules are deliberately harsh. A strong business with a CEO of eleven months reads as a strong business whose management is unproven, which is the truth, and the score says so instead of lending the newcomer the predecessor's decade.

Management is one of six pillars, with moat, business model, valuation, risk and optionality, and the quality score you see at the top of a company page is the weighted whole. But management is also passed forward: the business-model reading is told how management scored, the optionality reading asks whether this team has a history of finding and funding new lines, and the valuation reading is told that excellent allocators justify a higher multiple, and poor ones none. That is how a value investor thinks about it too. A great business in careless hands is a different investment from the same business in good ones.

Does the score reward vision and the will to lead change?

Yes, deliberately, and on top of the record rather than instead of it. After the base score for capital allocation, the agent asks whether this leadership leads change or resists it, whether the vision is clear and reasoned from first principles rather than borrowed from the industry, whether there is a proven turnaround, a culture that draws and keeps talent, real delegation with a credible successor, and strategic focus rather than acquisitions for their own sake. A founder, or a chief executive of ten years or more, earns credit for proven continuity. Each earns bounded credit, the whole of it capped, and a chief executive in the first two years cannot receive full credit for a vision that has not yet met a downturn. A clear vision with no capital-allocation record behind it is a story; the score is built so that the record comes first.

This is where our method parts company with a purely Graham reading of management. The businesses we want to own for a decade are the ones that will not be the same business in a decade, and the people running them have to be able to see that and act before the numbers force them. The agent is told to look for the pattern that Satya Nadella showed at Microsoft, Jensen Huang at Nvidia and Reed Hastings at Netflix, a leader who moved the company while the old business was still profitable, and for its opposite, the Kodak and BlackBerry pattern, where every quarter was defended until there was nothing left to defend. It asks what the chief executive said the company would be five years ago and whether that is what it became. It asks whether the strongest executives stay, because a culture that keeps talent is the only durable evidence of a leader worth following.

A founder in the chair helps, and the score says so. Founders own the company in a way hired managers rarely do, so the alignment question answers itself; they think in decades because they expect to be there; they will invest through a downturn that a caretaker would manage for the quarter; and when the business has to change, they have the standing to change it, as Bezos, Huang and Zuckerberg did against their own boards and analysts. Bain's study of the S&P 500 found that companies where the founder was still deeply involved returned about three times the rest of the index between 1990 and 2014. The agent gives a founder or a chief executive of ten years or more credit for proven continuity, and reads a founder-led push into a new line more generously than a hired one. The caveat is the other side of the same coin: a founder can also be the key-person risk, the succession gap and the board that never says no, which is why the questions about delegation and a credible number two sit beside the credit.

The order matters. Vision is scored after the capital-allocation record, it can add to a strong record but not rescue a poor one, and a founder's ten years in the chair earns more than a newcomer's keynote. The growth readings on the company page, Lynch's categories and the Gardner frame, read the same management score and call it visionary leadership once it reaches 8; the value reading reads it as whether the capital will be allocated well. Same people, same evidence, two questions.

What are the red flags in a management team?

A recent change at the top or three executives gone in two years. A chief executive from outside the industry, or an interim one. Pay out of scale with the size of the business. Insiders who own almost nothing. ROIC and ROE that grind lower year after year while the letters stay cheerful. Debt that free cash flow cannot clear in a few years, a share count that grows, stock compensation that eats the revenue. And the two the value-trap check watches: buying back stock while intrinsic value falls, and insiders selling a stock that looks cheap.

The buyback flag deserves a sentence of its own, because repurchases are the capital-allocation decision most often praised and least often examined. A buyback below intrinsic value transfers value from the sellers to the owners who stay; a buyback above it does the opposite, and a buyback funded with debt while returns on capital fall is a management team spending the owners' money to flatter a per-share number. The value-trap check reads the pattern directly: repurchases rising while ROIC declines, and insiders selling into a price that screens as cheap. Both are what a team in denial does, and denial about a weakening business is the most expensive management quality there is.

Where does the agent stop, and what should I check myself?

The agent reads the numbers as they stand today and the passages it retrieves; it does not hold the ten-year series or the repurchase history in its hands. Those are on the page beside it: the Capital track record under the Management pillar plots ROE and ROIC over the decade, the Context card carries three months of insider buying and selling, and the stress signals carry dilution and stock pay. The judgement that remains yours is the one-dollar test: did each dollar the company kept become at least a dollar of value? Open the proxy, read how the pay is earned, and compare the years of heavy buybacks with where the price stood against fair value.

The one-dollar test is Buffett's from the 1983 and 1984 letters, and it is the shortest honest test of a capital allocator: over time, every dollar of earnings retained should have produced at least a dollar of market value, or the owners would have been better off with the dividend. You can run it on any company with ten years of history in an afternoon. Add up the earnings kept, compare them with the change in the value of the business, and read the Capital track record for the trend underneath: ROE and ROIC that hold or rise across a cycle pass; a steady grind lower fails in slow motion, usually a year or two before the market says so.

The proxy statement is the other document the agent searches and you should read. It states how the executives are paid and against what. Pay tied to per-share value, returns on capital and multi-year targets points one way; pay tied to revenue growth, adjusted earnings or the share price over a single year points the other. An owner looks for the structure that would make the CEO poorer if the decisions were bad.

How do I check management on Invest Board?

Open a company and go to Understand, then the Management pillar: the score, the hundred-word read, the source beside each claim, and the Capital track record beneath. On Decide, the kill-criteria shelf holds four management-bound rules you can adopt in a click: insider net selling over six months, net debt against free cash flow, dilution year on year, and stock compensation as a share of revenue. The briefing reports when one trips. And Ask takes the question in plain words: what did management do with the cash over the last five years?

One last thing about scores. A management score of 8 is where the reading begins, not where it ends. The hundred words beneath it name the evidence and the open questions; the citations take you to the filing; the track record shows the direction. The agent has read more than you have, and it has understood less. Its job is to put the right questions and the right documents in front of you so that your judgement of the people running your money is made on the record rather than on the interview.