Warren Buffett talks about a lot of things. Capital allocation. Moats. The wonders of See’s Candies. But one topic he returns to again and again and one that I think is criminally underappreciated is executive compensation.
This clip is a good one.
His core point is almost offensively simple: pay people for what they can control, not for what they can’t.
He uses copper mining as the example. If copper is at $3.50 a pound, the village idiot running a copper company looks like a genius. If copper is at 80 cents, even Warren himself would be struggling. So why on earth would you write a fat bonus cheque tied to copper prices? That’s not measuring management, that’s measuring the commodity cycle and handing the manager credit for it anyway.
The same logic applies to oil. If crude is at $70, every oil CEO is strutting around like they personally drilled Texas. The moment they get called before Congress, though, suddenly none of them had anything to do with prices. Funny how that works. Buffett’s answer: pay them for finding costs. Did your team discover and develop oil cheaply over a 5 to 10-year period? That’s skill. That’s worth paying for. Riding the commodity wave isn’t.
At GEICO, he measures two things: unit growth, and the profitability of seasoned business. New customers cost money to acquire and you don’t punish the team for that. You want new business. But you measure what’s actually under their control.
This framework travels well. Capital-intensive business? Measure cost efficiency. Cyclical business? Strip out the cycle. Asset-light compounder? Focus on returns on incremental capital. The principle doesn’t change, only the variables do.
Then Charlie Munger, ever the warm and fuzzy one, drops the real diagnosis: it’s not greed driving bloated comp packages. It’s envy.
Think about that for a second. You can hand someone a $2 million bonus and they’ll be perfectly happy until they find out the person at the next desk got $2.1 million. Then they’re miserable for the next twelve months.
Charlie’s observation was that of the seven deadly sins, envy is the silliest. At least gluttony has an upside, Buffett notes he’s personally had some of his best times being gluttonous. We won’t discuss lust. But envy? All it does is make you feel worse. You gain nothing. You just sit there, stewing, comparing yourself to someone else’s number.
And yet envy is the engine behind the entire compensation arms race. Every year, the SEC mandates more pay transparency which sounds great in theory. In practice, it becomes a shopping list. One CEO sees that a peer gets their haircuts expensed, and suddenly they too require company-funded haircuts. And they become big tippers while they’re at it.
The disclosure meant to shame excess ends up spreading it.
One more gem from this clip: Berkshire has never in forty years lost a manager over a compensation disagreement. They’ve never hired a compensation consultant. No long meetings, no elaborate frameworks, no 80-page proxy exhibits explaining why the CEO deserves a “retention award” on top of his existing retention award.
As Buffett puts it: it’s not rocket science. It’s made to look complicated, because complexity serves the people who benefit from complexity.
(A small note: if you’re familiar with the full canon, you’ll know Charlie has a classic line about he would rather throwing a viper down his shirt than hiring a compensation consultant Alas, that gem didn’t make it into this particular clip. We mourn its absence.)
Then there’s Constellation Software. And honestly, it might be the best compensation system I’ve ever seen.
While Buffett laid down the philosophy, Mark Leonard and his team have been quietly living it for decades and at this year’s AGM, the new CEO, Mark Miller made clear that the tradition continues. He announced he will follow Mark Leonard’s lead and take zero compensation. No salary. No bonus. No options. Nothing.
Let that sink in for a moment. The CEO of one of the greatest compounding machines in public market history has decided he doesn’t need to be paid. Because he already owns the stock. His incentive is perfectly, uncomplicatedly aligned with yours.
But it goes beyond the top. Employees receive bonuses and then are required to use a meaningful portion of those bonuses to buy Constellation shares at market price. No discounts. No options with a low strike. Just: here’s your money, now put it where your mouth is, right alongside every other shareholder. You want to build wealth? Build it the same way the company does, by compounding capital at high rates of return over long periods.
And the metric that drives those bonuses? ROIC. Return on invested capital. Not revenue growth. Not EBITDA margins in isolation. Not total shareholder return, which as Buffett would point out, can be flattered by a rising tide just as easily as copper at $3.50 a pound. ROIC strips all of that out. It asks a simple question: for every dollar of capital entrusted to you, how much did you earn? That’s a metric management can actually influence. That’s something worth paying for.
Now here’s where it gets genuinely interesting. Constellation has long been a serial acquirer, buying vertical market software businesses, wringing out inefficiencies, redeploying capital. Organic growth was never really the story. But at this AGM, Miller flagged something worth paying attention to: they are beginning to explore rewarding organic growth as well, precisely because AI might actually make that achievable in ways it never was before.
Think about what that means. If AI enables Constellation’s portfolio companies to expand their addressable markets, build new features faster, and grow within their existing customer bases, then organic growth stops being the consolation prize and starts being a legitimate value creation lever. And Miller is getting ahead of it, structuring incentives now so that when and if it happens, the people driving it are rewarded for it.
That’s not reactive compensation design. That’s thinking several moves ahead.
The takeaway for us as investors is actually quite practical. When you’re evaluating management quality, look at what the incentive structure rewards. If a CEO is getting paid for commodity prices, currency moves, or a bull market in their sector, things entirely outside their control, that’s a yellow flag. Not because the CEO is necessarily dishonest, but because misaligned incentives produce misaligned behaviour over time.
Pay people for what they can control. Ignore what they can’t. And whatever you do, don’t tell them what the person next to them is making.
Buffett told us the principle. Constellation is showing us what it looks like in practice. The gap between those two and the rest of corporate America is, frankly, embarrassing.
