Expected Utility and Why a Risk-neutral Firm Is a Fiction

What this topic is

Expected utility is what you maximise once you admit that a dollar gained and a dollar lost are not the same size to you. You convert each possible outcome into a measure of how good it is for this specific balance sheet, average those, and convert back to dollars. The result is the certainty equivalent: the guaranteed amount you would swap the risky opportunity for. The gap between that number and the expected value is the risk premium, and it is a real cost that expected value cannot see.

Why it matters

Corporate finance teaches that a firm should be risk-neutral toward diversifiable risk, and for a large public company with no distress costs that is close to true. For almost everyone else it is false, and the falseness is expensive in both directions. Firms take bet-the-company risks because a spreadsheet said the expected value was positive. Other firms decline strings of small, genuinely favourable opportunities because the boss lost sleep once. Expected utility gives you one framework that says yes to the second set and no to the first, and it gives you the sizing rule that makes most large bets survivable.

Who should learn it

Owners deciding whether a project could end them, finance teams that keep approving positive-expected-value proposals, and anyone who has to explain to a board why a good bet is still too big.

What you will understand

• How to compute a certainty equivalent and a risk premium for a real project, with a formula that fits on one line
• Why the same opportunity is correctly a yes for a $10M balance sheet and a no for a $2M one
• Why repeated positive-expected-value bets can drive you toward zero almost surely, and the arithmetic that shows it
• How to size a bet rather than only accept or reject it, and why half the theoretically optimal size is usually the right answer

Prerequisites

• Bandit-based Price Testing Without Burning the Market
• Supervised Learning and the Bias-variance Tradeoff

Common misconception

"If the expected value is positive, take it. Over enough decisions it averages out." The averaging argument requires two things that are often absent: that the bets are many and independent, and that you are still in business to take the later ones. A bet that risks a third of your equity is not one draw from a large sample; it is a draw that changes the size of every subsequent draw. Once outcomes multiply your capital rather than add to it, the average across parallel universes stops describing what happens to you along your own path, and a sequence with a healthy positive expected value per round can still converge to ruin.