The Lesson from Irving, Texas

Lee Raymond and the Discipline of ROCE

In Private Empire, Steve Coll’s definitive account of ExxonMobil under Lee Raymond, one detail stands above the rest: Raymond’s singular, relentless obsession with Return on Capital Employed.

Raymond, known internally as “Iron Ass,” understood something that most public market investors, and virtually all private equity sponsors, either ignore or actively obscure: in a capital-intensive business operating at extraordinary scale, the only honest measure of management quality is the return generated on every dollar of capital the business has absorbed.

His logic was straightforward. ExxonMobil’s global operations spanned dozens of countries, hundreds of thousands of employees, and investment horizons stretching forty years or more. At that scale, the natural drift of any large organization (the slow accumulation of marginal projects, undisciplined spending, empire-building by division heads) compounds into catastrophic value destruction. Small inefficiencies, multiplied across a $200+ billion asset base, become large numbers quickly.

Raymond’s framework was built for oil fields and refineries, but it has never been more relevant than it is today. The hyperscalers (Alphabet, Meta, Microsoft, Amazon) are no longer software companies that happen to own servers. With combined capital expenditure exceeding $300 billion in 2025 and projected to surpass $400 billion in 2026, they have become physical infrastructure companies operating at a scale that rivals the integrated oil majors.

What ROCE Measures

And Why Other Metrics Fall Short

ExxonMobil’s own definition, applied consistently for decades, uses net income excluding after-tax financing costs divided by average capital employed. The key insight is that Capital Employed captures the full investment in the business: the equity shareholders have contributed, the earnings the company has retained over its entire life, and the long-term debt it has taken on. Nothing is excluded. Nothing is adjusted away.

This is the whole point of business. Earn high returns on the capital base. Retain earnings. Reinvest at similarly high returns. Compound. ROCE is the metric that captures this entire cycle in a single number.

The Portfolio: 10 Years of ROCE

High Returns on Growing Capital Bases

Each of these businesses has sustained or improved ROCE over a full decade, through commodity cycles, a pandemic, rate hikes, and massive incremental capital deployment. The pattern is not coincidence. It is the signature of durable competitive advantage.

The chart also reveals a recurring pattern in Sheephill’s entry timing:

In each case, the entry point coincided with a major deviation from the company’s historical ROCE trend. When an otherwise high-quality business shows a sharp, temporary decline in capital productivity, the market tends to reprice the equity as though the decline is permanent. That gap between temporary impairment and permanent repricing is where the opportunity lives.

Portfolio Performance

The Result of Focusing on Business Quality

A portfolio built around ROCE discipline, balance sheet quality, and entry timing during temporary dislocations has produced the following annualized returns through March 31, 2026.

The ROCE framework does not predict macro events. It does not require a view on oil prices, interest rates, or election outcomes. It identifies businesses with durable capital productivity and waits for the market to misprice them. The returns above are the cumulative result of that discipline applied consistently over a decade.

The Compounding Mechanic

Growing Capital, Sustained Returns

Each of these businesses has substantially grown its asset base over the past decade while maintaining or improving returns on that capital. This is the signature of a competitive moat.

The critical distinction: BHP shrank its capital base and improved returns through divestiture discipline. Alphabet and Meta massively expanded their capital bases and improved returns through superior reinvestment. ExxonMobil and Shell grew moderately and maintained strong returns through the commodity cycle. Each is a different expression of the same principle: deploy capital only where it earns above-hurdle returns.

When a business earns 25%+ ROCE and retains a significant portion of earnings, the compounding math is extraordinarily powerful. Meta’s shareholders’ equity grew from ~$44B in 2015 to over $153B in 2024. Internally generated, retained, and reinvested at rates that justified retention. That is the whole point of owning equity in a business.

Public market investors fixated on P/E and P/FCF miss this entirely. They evaluate a company on a single year’s earnings without asking whether the retained portion is being deployed productively. A company trading at 30x earnings with a 33% ROCE is a fundamentally different proposition than one trading at 10x with a 6% ROCE, even though the latter “looks cheaper” on a headline multiple.

From ROCE to Total Shareholder Return

Two Paths to Compounding

ROCE measures how productively a business uses its capital. But what matters to shareholders is what happens after that return is generated: how much is returned via dividends and buybacks, and how much is retained and reinvested?

The portfolio contains two distinct models. The energy and materials companies (XOM, SHEL, BHP) earn moderate returns on large, mature capital bases and return the majority of earnings to shareholders. The technology companies (GOOG, META) earn high returns on rapidly growing capital bases and retain the majority of earnings for reinvestment. Both models create value, though through different mechanics.

Dividend and buyback yields approximate, based on 2025 distributions relative to average market capitalization. Earnings retained = 1 − (total distributions / net income). Source: Company filings, Sheephill analysis.

The reconciliation matters because it explains why a portfolio can hold both 9% ROCE energy companies and 32% ROCE technology companies without contradiction. The energy names compensate for lower reinvestment returns with higher direct capital return. The technology names compensate for lower direct yield with superior reinvestment economics. Both are disciplined allocators of capital; they simply allocate in opposite directions.

One important clarification: ROCE is most powerful as a comparative and cyclical tool within an asset class, not as an absolute threshold across asset classes. A 9% ROCE in integrated energy consistently ranks first or second among IOC peers deploying similar capital into similar assets. In a capital-intensive commodity business with forty-year investment horizons, that peer-leading consistency signals the same operational discipline that a 30% ROCE signals in technology. The relevant question is not whether 9% clears some universal hurdle, but whether it represents best-in-class capital productivity for the opportunity set. When that peer-leading ROCE is then combined with a total shareholder yield of 8 to 11%, the effective return to shareholders on a well-timed cost basis can far exceed what the headline ROCE alone would suggest.

ROCE Is the Corrective

The public market’s standard toolkit (P/E, EV/EBITDA, P/FCF) evaluates businesses as static snapshots. ROCE evaluates them as compounding machines.

The distinction matters because the long-term value of an equity is not determined by what the market is willing to pay for today’s earnings. It is determined by the rate at which retained earnings can be reinvested and the returns those reinvestments generate.

Every company in this portfolio has demonstrated the ability to grow its capital base while sustaining or improving returns on that capital. That is the definition of a competitive moat, and it is the foundation of long-term equity compounding.