Investing Classics: Quality InvestingBuy Good Companies, Don't Overpay, Do Nothing

These three sentences part ways directly with Graham's cigar-butt tradition. The cigar butt buys "a mediocre company below liquidation value" and earns a one-off price gap; quality investing buys a good business that can roll its profits forward year after year at high ROCE (return on capital employed), and earns the steady growth of intrinsic value. Smith uses ROCE as his first filter—anything below the market average he won't even look at. "Do nothing" is not laziness but a deliberate defense against transaction costs and behavioral biases.

From its November 2010 launch to end-2021, Fundsmith Equity returned ~18% annualized, beating MSCI World over the long run—yet it was highly concentrated (20–30 names), with single-digit-percent annual turnover and an average holding ROCE near 30%. But in 2022 it fell ~14% and lagged the market—when rates rise fast, highly valued quality stocks are hit first. That is exactly the second sentence: "don't overpay" is no ornament but a constraint as important as "buy good companies."

Quality investing is most comfortable in low-rate, steady-growth environments; in years of sharp rate moves or deep-value rallies (e.g. 2022) it lags badly. The deeper risk is "quality decay"—yesterday's good company disrupted by technology or business model (Nokia and Kodak were both once high-ROIC good companies). Quality is dynamic, not a lifetime label.

• Has ROCE been stable above 15%–20% for 10 years, not just a single peak?
• Does the high return have an identifiable moat, or is it a cyclical/accounting illusion?
• Has the premium I pay for quality already pre-spent years of future growth?
• Is "do nothing" based on continually verified conviction, or just neglect to recheck?

ROIC = after-tax operating profit ÷ invested capital, measuring "how much each dollar invested earns back." Damodaran's core thesis: growth in itself is neutral. When ROIC > cost of capital (WACC), growth creates value; when ROIC < WACC, growth destroys it—the faster it expands, the more shareholders lose. So "high growth" is no automatic good news; the real moat is the positive, durable spread between ROIC and the cost of capital—translating "moat" into a calculable number.

The airline industry is the counter-example. Buffett's 1989 US Air preferred stake nearly went to zero; he joked investors should call an "airline addiction hotline." The industry's ROIC has long sat below its cost of capital—huge capital outlays, price wars, unions and oil prices mean growth just burns cash faster. The contrast is high-ROIC franchises: consumer brands, exchanges, rating agencies, which expand on minimal incremental capital. Same "growth," but one devours cash and one creates it—all in the relative position of ROIC versus cost.

ROIC can be distorted by accounting: goodwill, operating leases and expensed R&D all skew the figure. High ROIC can also be a cyclical-peak illusion (a boom year in commodities) or the last glow of something about to be disrupted. Look at "adjusted, sustainable" ROIC, not an isolated single-year high.

• Is the company's ROIC durably and stably above its cost of capital?
• Have I adjusted for goodwill, leases and R&D to see true invested capital?
• Is its "growth" creating value, or expanding at returns below cost?
• Is today's high ROIC a structural advantage, or cyclical/one-off?

The table at the top is the arithmetic of that quote. But there's an often-ignored premise: high ROIC isn't enough—you also need a "reinvestment runway," room to keep plowing retained profits back at high returns. See's Candies had an extremely high ROIC but couldn't absorb large incremental capital, so it could only return cash to Berkshire—a superb business, but not an "internal compounder." A true compounding machine combines high ROIC + a long reinvestment runway. Hence Munger: the first rule of compounding is never to interrupt it unnecessarily.

From 1972 to 2007, See's Candies contributed over $1.35 billion in pre-tax profit to Berkshire on less than $40 million of added capital—astonishing returns, but with almost nowhere to deploy incremental capital. It "spat out" cash for Buffett to allocate elsewhere (that is Berkshire's way of compounding). By contrast, a company with both high ROIC and the ability to keep reinvesting profits into its own high-return projects compounds internally. Both are good businesses, but only the latter is a textbook compounding machine.

Reinvestment runways run dry—once a market saturates, forced "growth" tends to pour cash into low-return expansion, diluting ROIC. Management's empire-building urge is the number-one killer: M&A for the sake of size, turning cash that should have been returned into mediocre assets. Here the "high ROIC" label deceives you, because the return on marginal capital has already collapsed.

• What share of profits can it reinvest at its existing high ROIC?
• How long is the runway left—penetration, geography, categories?
• When the runway peaks, does management return cash or expand at low returns?
• Am I looking at overall average ROIC, or the return on marginal new capital?

The 1972 "Nifty Fifty" is the classic specimen. They were called "one-decision stocks"—the market deemed them so good you could buy at any price and hold forever. They were indeed mostly fine growth companies (Coca-Cola, IBM, Polaroid, Xerox, Avon), with top-tier ROIC and brands. The problem was price: in 1972 their P/Es ran a sweeping 40–90x, pricing in over a decade of flawless growth in advance.

In the 1973–74 bear market, Polaroid fell ~91%, Avon ~86%, Xerox ~71%—the fundamentals didn't collapse in lockstep; mere valuation reversion wiped out years of returns. Jeremy Siegel's 1998 study gives an intriguing sequel: held to 1996, the Nifty Fifty roughly matched the S&P—quality eventually "grew back." But the price was enduring more than 20 years, and only on average; Polaroid and Kodak later went to zero. "Overpaying" can only be salvaged by very long holding plus survival—costly, and not everyone can bear it.

But don't swing to the other extreme. Fearing valuation too much on a genuine compounding machine and waiting forever for "cheap" means missing it entirely (the very tension in Card 3—Munger says a slightly dear price is fine). The distinction: does the premium you pay truly have a 20-year high-ROIC reinvestment runway to digest it? A 40x P/E on 8% growth is a trap; on verifiable 20 years of rapid compounding it may be reasonable. The question isn't "is it expensive?" but "is the expense justified?"

• How perfect must the growth/return assumptions embedded in the price be to pan out?
• If future growth comes in 30% below expectation, do I still have a margin of safety?
• Does the premium I pay have a long enough runway to be amortized?
• Am I paying for business quality, or for a "certainty narrative"?