Porter — a moat is a structure, not an adjective

The argument

Porter's claim is that profitability is mostly a property of industry structure, and only secondarily of how well a company is run.

Five forces set the ceiling: the threat of new entrants, the bargaining power of suppliers and of buyers, the threat of substitutes, and the intensity of rivalry among existing players. Where all five are strong, even excellent management earns ordinary returns. Where they are weak — high entry barriers, fragmented suppliers, captive customers, no substitute — even mediocre management earns well.

A moat, in this account, is not a quality of the company. It is the shape of the industry around it.

What structure looks like in the numbers

You cannot measure the five forces directly from a filing. What you can measure is their consequence, which is persistence.

Competition is a levelling force: high returns attract capital, capital competes the returns away. So a company earning returns on capital employed above its cost of capital, year after year, is telling you that something is stopping the levelling. That something is the moat, whatever it turns out to be.

The four things worth checking across a decade:

• Gross margin stability — pricing power that survives cycles
• Return on capital employed above the cost of capital — sustained, not once
• Margin trend — held rather than eroding
• Scale position — leadership within the industry rather than the market

Any one of these can be a good year. All four for ten years is a structure.

Where the reasoning inverts

The trap is running the argument backwards: seeing high returns and concluding there must be a moat.

High returns have other explanations — a cyclical peak, a temporary shortage, an accounting choice, a competitor's stumble, a patent about to expire. The persistence test is what separates a structural advantage from a good few years, and ten years of history is the minimum that makes it meaningful.

The limits

Porter's frame was built for industries with recognisable boundaries and is least comfortable where the boundaries move. A platform business can be protected by network effects that no cost curve or entry barrier describes, and a regulated business can have its moat granted or revoked by legislation rather than competition.

The other limit is temporal. A moat measured over the last decade is evidence about the last decade. Kodak's persistence statistics were superb right up to the point they stopped meaning anything.