Expected Monetary Value (EMV)

A common mistake among project managers and business students is to treat expected monetary value (EMV) as a prediction. EMV of $50,000 does not mean you will get $50,000. It means that if you could repeat the exact same risky situation thousands of times, the average outcome would be $50,000. For a one-off project decision, you will get one result, not an average. The math is still useful because it lets you compare options under uncertainty. The core formula is EMV = probability × impact, summed across all identified risks. The PMI PMBOK Guide (7th ed.) defines it that way, and the Practice Standard for Project Risk Management calls EMV a statistical concept that calculates the average outcome when the future includes scenarios that may or may not happen. OpenIntro Statistics (4th ed.) section 3.4 gives the same definition for any random variable: E(X) = Σ x_i * P(X = x_i).

To compute EMV for a single risk, multiply the probability of the risk occurring (as a decimal) by the monetary impact if it does occur. For a threat with a 30% chance of causing a $100,000 loss, the EMV is 0.30 × −$100,000 = −$30,000. That negative sign matters: threats reduce value, opportunities add it. The PMBOK Guide sums EMV across all risks in a risk register to get a total contingency reserve. The Practice Standard for Project Risk Management walks through this as a standard calculation step.

When you have multiple risks, sum their individual EMVs. The sum gives you the overall expected monetary value of the project's risk exposure. But this sum only makes sense if the risks are independent. Blitzstein & Hwang (2nd ed.) ch. 4 explains linearity of expectation: E(X + Y) = E(X) + E(Y) for any random variables. That theorem is what lets you add EMVs without worrying about whether the risks are dependent. The catch is that variance does not add as cleanly, so the total EMV says nothing about the range of possible outcomes.

Weigh Threats and Opportunities Separately

Break your risk register into threats (negative impact) and opportunities (positive impact). A threat with a 20% chance of costing $50,000 has an EMV of −$10,000. An opportunity with a 15% chance of saving $80,000 has an EMV of +$12,000. The net EMV of these two risks is +$2,000. That positive number suggests the project should take both risks, but it hides the fact that one is a cost and the other is a saving. The net EMV is a single figure you can plug into a decision tree, but it does not tell you whether the project will lose money on the threat before the opportunity materializes.

The Practice Standard for Project Risk Management advises using EMV in a risk register to prioritize risks by their probability-impact score. The PMBOK Guide treats EMV as one input to the quantitative risk analysis process. Do not confuse EMV with the most likely outcome (the mode) or with the probability of winning. A low-probability, high-impact risk (like a 1% chance of a $1M loss) has an EMV of only −$10,000, but it can ruin the project if it occurs. The EMV alone does not capture that tail risk.

A decision tree is the standard tool for applying EMV to choices. You draw branches for each decision option, then branches for each uncertain outcome, with probabilities and monetary outcomes at the end. To solve it, you start at the leaves and work backward, a process called rolling back. At each chance node, compute the EMV of the outcomes branching from it: sum of (probability × outcome). Choose the path with the highest EMV at each decision node.

The PMI Practice Standard for Project Risk Management includes decision trees as a technique for comparing options under uncertainty. Blitzstein & Hwang (2nd ed.) ch. 4 covers the linearity that makes this addition valid. The tree does not tell you which option is safer, only which has the higher long-run average. For a one-off decision, the EMV-maximizing branch might still be the one that bankrupts you if the bad outcome hits. That is why the tree should accompany a discussion of variance and risk attitude, not replace it.

Building the Tree

Label each branch with the action. Below it, draw a chance node with branches for each possible outcome, labeled with its probability (all must sum to 1.0) and its monetary impact. At the end of each branch, write the net outcome. Multiply probability by outcome at each chance node, sum across that node's branches, and write the EMV above the node. At a decision node, choose the branch with the higher EMV. The tree typically appears before you commit to a strategy, not after.

Your project needs a software module. You can build it in-house for a fixed cost of $120,000, or buy a commercial off-the-shelf package for $80,000. The risk: the build option has a 60% chance of working perfectly (no extra cost) and a 40% chance of requiring $50,000 in rework. The buy option has a 90% chance of working with $10,000 in customization and a 10% chance of failing to integrate, costing $200,000 to fix.

Calculate the EMV for each.

Build: Outcome A (60%): cost $120,000, no rework. Outcome B (40%): cost $120,000 + $50,000 = $170,000. EMV = (0.60 × −$120,000) + (0.40 × −$170,000) = −$72,000 + −$68,000 = −$140,000.

Buy: Outcome A (90%): $80,000 + $10,000 = $90,000. Outcome B (10%): $80,000 + $200,000 = $280,000. EMV = (0.90 × −$90,000) + (0.10 × −$280,000) = −$81,000 + −$28,000 = −$109,000.

The buy option has a higher EMV (−$109,000 vs. −$140,000), so a risk-neutral decision-maker chooses buy. But the buy option has a 10% chance of a $280,000 loss, a tail risk that the build option does not have. If your project cannot absorb a $280,000 hit, the EMV-maximizing choice might be the wrong one. This is why the PMBOK Guide recommends coupling EMV with a sensitivity analysis or a Monte Carlo simulation.

Option	Outcome	Probability	Cost	EMV Contribution
Build	Works perfectly	0.60	$120,000	−$72,000
Build	Requires rework	0.40	$170,000	−$68,000
Build	Total EMV	1.00	—	−$140,000
Buy	Works with customization	0.90	$90,000	−$81,000
Buy	Fails to integrate	0.10	$280,000	−$28,000
Buy	Total EMV	1.00	—	−$109,000

EMV assumes risk neutrality, valuing each dollar equally. Most people are not risk-neutral. Blitzstein & Hwang (2nd ed.) ch. 4 covers utility functions, which can be nonlinear in money. A risk-averse person would reject the buy option in the example above because the 10% chance of a $280,000 loss outweighs the $31,000 EMV improvement. Expected utility, not expected monetary value, determines rational choice under risk.

The second limit is that EMV is meaningless for a single trial. The Law of Large Numbers says the average outcome converges to the EV over many trials, but a project is usually one trial. A positive EMV does not guarantee a profit, and a negative EMV does not guarantee a loss. This is the failure case: a project manager who chooses an option based solely on EMV and then blames the math when the bad outcome hits. Always pair EMV with a range of outcomes, variance, standard deviation, and worst-case scenario, before making a final decision.

Decision tree expected value is the EMV computed at each node of the tree. You use it to compare discrete alternatives. In the build-versus-buy example, the decision tree shows the buy option at −$109,000 and the build option at −$140,000. The tree itself is the diagram with branches, probabilities, and outcomes. Rolling back the tree to compute the EMV at each chance node is the same as applying the EMV formula to each set of branches.

The Practice Standard for Project Risk Management and the PMBOK Guide both include decision trees as a tool for quantitative risk analysis. OpenIntro Statistics (4th ed.) section 3.4 covers the underlying random variable concept. When you present a decision tree to stakeholders, show the EMV at each node and highlight the chosen path. But add a note that the EMV is an average, not a guarantee. The single most practical thing to do next is to compute the variance of each option's outcomes: the standard deviation tells you how much the actual result can swing from the EMV.

For the PMP exam, you need to know the EMV formula and how to apply it to a decision tree. You will be given probabilities and monetary impacts and asked to compute the EMV of a single risk or a simple tree. The PMBOK Guide (7th ed.) is the reference. The exam also tests your understanding that EMV is a long-run average, not a prediction. A common exam question asks which option to choose based on a decision tree, always choose the branch with the higher EMV unless the question says the firm is risk-averse.

The Practice Standard for Project Risk Management adds detail on how to use EMV in a risk register. You sum the EMVs of all risks to estimate the contingency reserve. But the exam will not ask you to compute a reserve from a full register; it will give you a handful of risks to sum. The key is to treat threats as negative and opportunities as positive, then add them. A mistake is to forget the sign.

Expected value decision making means choosing the option with the highest EV. It is a rational benchmark under risk neutrality. But in practice, you should ask three questions before relying on EV: (1) Is this a repeated decision, or a one-off? For repeated decisions, EV converges to the average. For one-off decisions, variance matters more. (2) Can my organization absorb a worst-case loss? If not, the EV-maximizing option may be too risky. (3) Are the probabilities and impacts accurate? EMV is only as good as its inputs. Subjective probabilities can be off by 50% or more, as noted in the research.

Blitzstein & Hwang (2nd ed.) ch. 4 covers linearity of expectation, which makes EV addition valid even for dependent risks. OpenIntro Statistics (4th ed.) section 3.4 gives the formal definition. The PMBOK Guide's EMV is the same concept applied to project risks. The single most practical thing to do next is to run a sensitivity analysis: change each probability and impact by ±20% and see how the EMV changes. If the decision flips, the inputs are not precise enough to trust the EV.

Open a spreadsheet. List every identified risk with its probability (as a decimal) and its monetary impact. Multiply each pair to get each risk's EMV. Sum the positive EMVs (opportunities) and negative EMVs (threats) separately. Then compute the net EMV. This is your baseline. Next, compute the standard deviation of the total outcome using the variance formula from Blitzstein & Hwang (2nd ed.) ch. 5: Var(X) = E(X²) − (E(X))². The standard deviation tells you the range of outcomes you should expect. If the standard deviation is larger than the net EMV, your risk exposure is dominated by variance, and the EMV alone is misleading. That is the failure case: a project manager who reports only the EMV and gets blindsided by a bad outcome.