← All publications

From the literature

Insure, diversify, consolidate: how organisations eat up uncertainty

Knight does not present uncertainty as an absolute inevitability. A large part of economic organisation serves precisely to reduce it, spread it or turn it into measurable risk.

Chapter VIII is the most practical in the book. Having defined uncertainty, Knight methodically surveys the means the economy has devised to digest it. There are five of them, and most contemporary risk management arrangements are variations on them.

1. Consolidation, or the law of large numbers

Insurance is the most obvious example. If a sufficiently homogeneous group of cases can be assembled, individual fluctuations offset one another and become predictable at portfolio level. What was a nasty surprise for one person becomes a statistical loss for the whole. Uncertainty is converted, at least partly, into measurable risk, and therefore into a budget line.

Knight extends this reasoning to the organisation of firms, and this is where the argument becomes interesting. An activity conducted on a large scale brings together more decisions, customers, projects and events. Good and bad results partly offset one another. The large firm is therefore not large merely because it owns more impressive factories; it also benefits from a consolidation effect on uncertainty that is an economic advantage in its own right.

2. Diversification

Diversification follows a similar but distinct logic. Consolidation multiplies similar cases; diversification combines different ones. A single decision can be extremely hard to predict. A portfolio of independent or imperfectly correlated decisions produces a more stable outcome, without any individual decision becoming predictable as a result.

This principle became familiar in market finance with the work of Markowitz, thirty years later. Knight places it within a more general theory: reducing uncertainty is an economic function in itself, which deserves to be rewarded as such.

3. Information and forecasting

Scientific research, statistics, market studies, forecasts and expertise make it possible to turn certain unknowns into better-established facts. Here again, the process does not remove all uncertainty. It shifts the boundary between what has to be guessed and what can reasonably be calculated. A geotechnical survey turns an unknown about the subsoil into data; it says nothing about the price of steel in three years' time.

4. Control of the future

Some uncertainties cannot be foreseen but can be constrained. A long-term contract, vertical integration, an indexation clause or a framework agreement do not predict the future: they make it partly binding. This is a form of uncertainty reduction through action rather than through knowledge, and it is often more effective than the previous one.

5. Specialisation in bearing uncertainty

The fifth mechanism is the most Knightian. Uncertainty can be transferred to those best placed to bear it, either because they understand it better, because they can consolidate it, or because they have the financial strength to absorb it. This is the function of futures markets, venture capital, reinsurance and, more generally, of any co-financing structure.

The limits, and a word of caution

An important limit remains, which Knight stresses insistently: situations that depend on human judgement are difficult to consolidate from the outside. The performance of a decision depends on factors internal to the decision-maker, on their understanding of the context and on their behaviour, which change precisely because they are insured. Here Knight identifies, in different vocabulary, what economic theory would later call moral hazard. It is one of the reasons why entrepreneurial risks cannot be insured like a fire or a car accident.

The practical lesson is twofold. First, a well-designed organisation can genuinely reduce exposure to certain unknowns, and this has to be designed deliberately: it is not a by-product of growth. Second, it is dangerous to label as "risk management" every attempt to put a process around a problem. Sometimes the process measures. Sometimes it organises. Sometimes it simply shifts responsibility to someone else. And sometimes, to be fair, it mainly adds a committee.

Frank H. Knight, Risk, Uncertainty, and Profit (1921), chapter VIII. Editorial adaptation, not a quotation.