Risk versus Uncertainty
The tools of project risk analysis rest on a distinction that goes back to Frank Knight. Risk describes situations where the outcomes are unknown but their probabilities are known, like the roll of a fair die. Uncertainty, sometimes called Knightian uncertainty, describes situations where even the odds are unknown, so no reliable distribution can be written down. The practical consequence is sharp. Monte Carlo simulation needs probability distributions, so it lives in the world of measurable risk. Where genuine uncertainty dominates, those distributions are guesses, and the analyst leans more on scenario thinking, judgement and flexibility than on a precise computed probability.
Why it matters
If you bet on a die you can compute exact odds, so the situation is risky but quantifiable. If you open a new cafe and ask how many customers arrive next month, you cannot even pin down the probabilities, so the situation is uncertain in Knight’s deeper sense. Most investment decisions live somewhere between the two. Recognising which one you face keeps you honest, because dressing up deep uncertainty as a tidy probability distribution gives false confidence in a number that the world never promised to obey.
Formulas
Worked examples
A firm models next year’s commodity price with a distribution estimated from decades of history, and separately weighs the impact of a never-before-seen regulatory regime on its new product. Which is risk and which is uncertainty?
The commodity price is closer to measurable risk, because a long, stable history supports an estimated probability distribution that simulation can use. The novel regulatory regime is closer to Knightian uncertainty, because there is no precedent from which to derive reliable odds, so any probability attached to it is a judgement rather than a measured frequency. The firm should simulate the first and lean on scenarios, flexibility and conservatism for the second.
Common mistakes
- ✗Risk and uncertainty are just two words for the same thing. Risk has known probabilities, uncertainty does not, and the difference dictates which analytical tools are honest to use.
- ✗Any uncertainty can be turned into a probability with enough effort. Genuine Knightian uncertainty lacks a reliable basis for odds, so a manufactured probability can mislead rather than inform.
- ✗Monte Carlo simulation resolves uncertainty. Simulation handles measurable risk by sampling known distributions. It cannot manufacture trustworthy distributions where none genuinely exist.
- ✗Deep uncertainty means analysis is useless. Even without reliable odds, scenario analysis, decision trees and built-in flexibility help manage outcomes that cannot be precisely quantified.
Revision bullets
- •Risk means outcomes unknown but probabilities known, like a fair die
- •Uncertainty in Knight’s sense means even the probabilities are unknown
- •Monte Carlo needs distributions, so it belongs to measurable risk
- •Under deep uncertainty, manufactured probabilities give false confidence
- •Scenario analysis, decision trees and flexibility help where odds are unknown
- •Most real investment decisions sit between pure risk and pure uncertainty
Quick check
In the Knightian sense, the difference between risk and uncertainty is that under risk
Monte Carlo simulation is best suited to problems of
Connected topics
Sources
- Titman & Martin, Ch. 3Titman, S., & Martin, J. D. Valuation: The Art and Science of Corporate Investment Decisions. Pearson.Frames investment analysis as decision-making under uncertain future outcomes that risk tools attempt to quantify.
- Knight (1921)Knight, F. H. Risk, Uncertainty and Profit. Houghton Mifflin, 1921.Original distinction between measurable risk and unmeasurable (Knightian) uncertainty.