Knightian uncertainty is uncertainty about the future that cannot be measured: the possible outcomes, their probabilities, or both, are unknown and cannot be reliably estimated. It is named after the American economist Frank H. Knight, who in his 1921 book Risk, Uncertainty and Profit separated measurable "risk", where the odds can be worked out from experience or logic, from true "uncertainty", where they cannot. The idea is used in economics, finance and policymaking to describe situations, such as a new technology, a pandemic or a financial panic, where past data offer little guide.
The distinction matters because the tools of financial risk management, from insurance pricing to value at risk, assume that probabilities can be estimated. Under Knightian uncertainty they cannot. In everyday economics the broader word "uncertainty" is also used for any doubt about future conditions, such as unclear government policy; this entry covers both the strict Knightian sense and that wider use.
Knightian uncertainty in plain terms
Rolling a die is risky but not mysterious: you do not know which number will come up, but you know each has a one-in-six chance. Drawing a ball from an urn whose contents nobody has seen is different. You cannot even say what the chances are, because you do not know what is inside.
Knight called the first kind of not-knowing risk and the second uncertainty. Many big decisions in life and business are more like the mystery urn: starting a company with a product nobody has tried, or guessing how people will behave in a situation that has never happened before. No formula can give honest odds for them.
Frank Knight and the theory of profit
Knight, who later taught at the University of Chicago and is regarded as one of the founders of the Chicago school of economics, developed the idea from his 1916 Cornell doctoral thesis. He argued that measurable risks can be insured or pooled away, so they cannot explain lasting profit. Profit, in his view, is the reward the entrepreneur earns for making judgements and bearing the uncertainty that no market can insure. In a world of pure, calculable risk, competition would drive profit to zero.
Keynes and radical uncertainty
John Maynard Keynes reached a similar view in A Treatise on Probability, also published in 1921, and made it central to his economics. In a 1937 article defending his General Theory he wrote that for matters such as the price of copper or the rate of interest twenty years ahead there is no scientific basis for calculating probabilities. Because the future is unknowable, he argued, business investment depends on conventions and on animal spirits, and people hold money as a safe haven, an idea behind liquidity preference theory and the precautionary motive. Post-Keynesian economists, and later writers such as John Kay and former Bank of England governor Mervyn King in their 2020 book Radical Uncertainty, extended this line of thought; see also Keynesian economics.
Ambiguity in decision theory
Mainstream theory long sidestepped the problem. Following Leonard Savage's 1954 work on subjective probability, many economists assumed that a rational person can always assign personal odds to any event, so that uncertainty collapses back into risk; the theory of rational expectations took a similar view of how markets form forecasts. Daniel Ellsberg challenged this in 1961 with experiments showing that people prefer bets with known odds over bets with unknown odds, even when no probability assignment can justify the choice. This "ambiguity aversion" is now studied in behavioral economics, and formal models, such as Itzhak Gilboa and David Schmeidler's 1989 maxmin expected utility, describe decision-makers who plan for the worst plausible probability. Lars Peter Hansen and Thomas Sargent developed "robust control" methods for policymakers unsure which economic model is right.
Uncertainty in markets and policy
- Financial crises. Economists including Ricardo Caballero have argued that panics, such as the 2008 financial crisis, involve Knightian uncertainty: investors unable to judge who holds which losses withdraw from whole markets at once, far beyond what measured risk premiums would imply.
- Rare events. A black swan, in Nassim Nicholas Taleb's sense, is close to Knightian uncertainty: an event outside the range of past data. Models built on the normal distribution understate such fat tails.
- Monetary policy. Central bankers face uncertainty about the state of the economy and how it works. Alan Greenspan, then chair of the Federal Reserve, described US monetary policy in 2004 as a form of risk management under deep uncertainty.
- Policy uncertainty. Scott Baker, Nicholas Bloom and Steven Davis built an Economic Policy Uncertainty index from newspaper coverage, published in 2016; higher readings are associated with delayed hiring and investment. The US index spiked during the 2020 pandemic and reached new highs in April 2025, amid sweeping US tariff increases, above its 2020 peak.
Critics note that the line between risk and uncertainty is blurred in practice, since almost every real probability is an estimate with some doubt attached. Supporters reply that the distinction still warns against treating model outputs as certainties, and encourages buffers, flexibility and stress testing where the odds cannot be known.