Concept

Knightian Uncertainty

conceptinvestingphilosophyrisk-management

Knightian Uncertainty

The distinction between risk and uncertainty, introduced by economist Frank Knight in Risk, Uncertainty and Profit (1921). Risk has a known or estimable probability distribution — you can insure against it, price it, and make rational expected-value calculations. Uncertainty is irreducible: no amount of data resolves what probability to assign to an outcome, because the distribution itself is unknown.

Knight’s canonical examples: a fair coin is risk; the outcome of a genuinely novel business venture is uncertainty. Keynes developed a parallel formulation: in many domains of economic life, we simply do not know the relevant probabilities, and pretending otherwise is a form of false precision.

The practical difference

Under risk, the rational strategy is to diversify, hedge, or insure. Under uncertainty, these strategies do not cleanly apply — you cannot buy insurance against an event whose probability you cannot estimate. The rational response to uncertainty is a matter of philosophical debate; Knight’s own view was that entrepreneurs are defined by their willingness to bear non-diversifiable uncertainty.

In practice, most financial institutions are mandated to treat uncertainty as risk: they assign distributions to scenarios they cannot actually estimate, run Value at Risk (VaR) models, and report pseudo-precise confidence intervals to regulators and clients. This is manageable as a governance convention but misleading as epistemology.

In investing

Bill Miller invokes this distinction to explain his competitive advantage. Most institutional managers cannot hold deeply concentrated positions in Amazon (pre-2010), Bitcoin, or Chinese internet stocks because their mandates require risk to be bounded and explainable — and the probability distribution of these assets genuinely cannot be specified. Miller’s claim is that he is simply more comfortable than most with the discomfort of not knowing, which allows him to hold where others cannot.

The extension: many of Miller’s best investments looked like extraordinary risk when he made them but were, in his view, uncertainty — unknown distribution, but strong asymmetric upside if the core thesis proved correct. Amazon in 2002 (bonds at 20–30% yield, equity optionality not priced), Bitcoin in 2015 (supply fixed, demand trajectory unknown). The asymmetry is available precisely because the uncertainty is real — if the probability were estimable, the position would already be priced.

Connection to pragmatism

Miller draws Knightian uncertainty into his broader pragmatist philosophy (William James). James’s “On a Certain Blindness in Human Beings” is an essay on the opacity of other people’s inner lives; Miller reads it as a general epistemic caution: the most important truths about any situation are likely invisible to most observers. This is structurally the same as uncertainty — what looks like chaos or noise is, in fact, a signal that most market participants’ models cannot accommodate.

Where mainstream views differ

Efficient market variants: If markets are efficient, there is no Knightian uncertainty available to exploit — any genuine uncertainty is already priced via diversification and risk premia. Active managers who claim to be exploiting uncertainty are, in this view, simply taking undisclosed risk.

Behaviouralists: The discomfort people feel at true uncertainty (Ellsberg paradox — people prefer known-risk gambles to unknown-probability gambles even when the expected values are identical) is well documented. Miller’s comfort with uncertainty may be a genuine psychological differentiator — or a post-hoc rationalisation of concentrated bets that happened to pay off.

Risk management practice: Treating uncertainty as risk (assigning distributions you do not actually know) may be epistemically wrong but organisationally necessary. Institutions that refuse to model uncertainty cannot communicate to clients, regulators, or boards; the pragmatic solution is a convention everyone knows to be imprecise.