Can the certainty equivalent exceed the expected amount?

Yes, under risk-seeking preferences represented by an appropriate convex increasing utility function over the outcomes.

Must everyone agree on the same certainty equivalent?

No. Preferences and circumstances can differ.

Does a certainty equivalent mean the risky outcome becomes certain?

No. It is a comparison value; the original prospect remains uncertain.

AI Glossary / Graphs and symbolic AI / Intermediate

Certainty equivalent

Definition

A certainty equivalent is a guaranteed outcome that a decision maker values equally to a specified uncertain prospect. In expected-utility analysis, it is the certain amount whose utility equals the prospect's expected utility. It translates a risky option into a comparable sure value while reflecting the decision maker's preferences and the stated outcome model.

What it means

Two people can assign different certainty equivalents to the same gamble because they value its consequences differently. A risk-averse person may value a risky gain below its expected monetary amount, while a risk-neutral person values it at that expected amount under a linear monetary utility model.
The number is not an objective market price or a guaranteed payout from the gamble. It describes indifference within a particular preference model or elicitation exercise. Timing, background wealth, obligations, and the meaning of the outcomes can affect the comparison, so those conditions should be stated rather than hidden behind a single figure.

How it works

List the possible outcomes and their probabilities. Under expected utility, transform each outcome through a utility function, multiply by its probability, and add the results. Then find the certain outcome with that same utility. With a strictly increasing utility function u, this is written CE = u inverse of E[u(X)].
This calculation differs from taking the utility of the expected outcome. Nonlinearity generally makes the two different. For a monetary gain variable and a concave utility function, the certainty equivalent is no greater than the expected amount. If a model uses final wealth instead of gains, include initial wealth consistently before calculating the equivalent.

Three practical examples

A decision analyst compares an uncertain operational benefit with a smaller guaranteed benefit using an explicitly chosen preference model.
A classroom exercise illustrates how identical expected amounts can have different values to a risk-averse decision maker.
A planning study expresses several uncertain outcomes as sure-value equivalents so stakeholders can inspect how assumed risk preferences affect the ranking.

Think of it this way

Think of asking, “What guaranteed result would make me equally satisfied as this uncertain option?” The answer is the certainty equivalent. The analogy is limited because casual answers may be inconsistent or depend on how the question is framed. A formal calculation requires a defined prospect and a preference model, not just an intuitive discount applied to every uncertain number.

Worked example

Consider a hypothetical nonnegative outcome X that is 16 units with probability one half and 64 units with probability one half. Assume utility u(x) = the square root of x. The two utilities are 4 and 8, so expected utility is (4 + 8) divided by 2, or 6.
The certain amount with utility 6 is 36, because the square root of 36 is 6. Thus the certainty equivalent is 36 units. The expected amount is (16 + 64) divided by 2, or 40. Under this stipulated model, the difference of 4 is the risk premium: the expected amount exceeds the sure value considered equally desirable. A guaranteed 37 units would be preferred to the gamble under this model, while 35 would not. The square-root utility is a teaching assumption over these nonnegative outcomes, not a measured preference for a person or a recommendation for valuing a real investment.

Illustrative case study

At fictional Ash Planning, a team compares a guaranteed resource benefit with an uncertain alternative. Analysts first show the raw outcomes and probabilities, then calculate equivalents under several clearly labeled preference assumptions. Stakeholders can see whether the recommendation changes when those assumptions change.
The team keeps the preference assumptions separate from estimates of what will happen. This fictional case illustrates transparent decision analysis. It does not claim that a certainty-equivalent calculation removes uncertainty or that one selected utility function represents every stakeholder.

Common misconceptions and mistakes

A certainty equivalent is not the expected value unless the relevant utility is linear or another special condition makes them equal. It is also not the certainty effect, which describes a behavioral pattern involving the special weight given to guaranteed outcomes.
Do not apply a utility function to gains in one option and final wealth in another. That changes the comparison. A model-derived equivalent also depends on probability estimates; incorrect probabilities can produce misleading values. Finally, certainty equivalence in control theory can refer to a different principle involving decisions based on estimated states or parameters. That technical use should not be silently substituted for this utility-based definition.

Business use

Businesses can use certainty equivalents to make risk-preference assumptions visible in comparisons of uncertain outcomes. They are most useful when readers can inspect the underlying distribution and the chosen utility model. Report sensitivity to those assumptions and preserve hard constraints separately: an attractive equivalent does not override a requirement that must hold in every permitted outcome. The calculation supports a decision discussion; it does not supply the organization's values automatically.

Check your understanding

In the square-root example, why is the certainty equivalent 36 rather than 40?

Answer: Utility is averaged before converting back to outcome units. Expected utility is 6, and the outcome with utility 6 is 36.

Related terms

Expected utility
Utility function
Certainty effect
risk premium
risk aversion
lottery
risk neutrality

Learn next

Repeat the worked example using linear utility and then an increasing convex utility over the same nonnegative range. Compare certainty equivalents while keeping probabilities unchanged. This isolates how preferences affect valuation without confusing them with beliefs about which outcome will occur.

Sources

Sources: Russell and Norvig, original AIMA utility and decision-making slides: https://aima.cs.berkeley.edu/slides-pdf/chapter16.pdf . MIT OpenCourseWare, original uncertainty teaching materials: https://ocw.mit.edu/courses/14-01sc-principles-of-microeconomics-fall-2011/pages/unit-6-topics-in-intermediate-microeconomics/uncertainty/ . Original research on certainty equivalents under broader preference models: https://arxiv.org/abs/1904.09456 .

Previous: Certainty effect Next: Certification