Expected monetary value: contingency you can defend
Build a probability-weighted risk allowance without mistaking expected loss for a guaranteed funding reserve.
By Dr. Hassan Khames EliwaLast updated

Decide what the allowance is supposed to mean
Expected monetary value multiplies an event probability by its monetary consequence. Summing those amounts estimates mean exposure under a stated model. It is useful for comparing responses and explaining a risk allowance. It does not establish how much funding provides a chosen confidence of staying within budget.
A low probability of a large loss can produce a modest average while leaving a substantial possible invoice. Calling that average a safe reserve hides the shape of the exposure. Start by deciding whether the question concerns a response decision, an expected forecast or a funding confidence level. Those are different questions, and a decimal-heavy answer cannot make them interchangeable.
The expected monetary value calculator separates positive threat exposure, negative opportunity savings and net expected cost. Keep the underlying register beside the output. The GAO Cost Estimating and Assessment Guide offers a wider framework for credible estimates and uncertainty analysis; the figures below are explicit illustrative assumptions.
Define an event that somebody can recognize
A usable row describes a trigger, a consequence and a project horizon. “Supplier risk” is too broad. “A late control assembly requires temporary operation until delivery” identifies a consequence that can be estimated and monitored. Ask what costs arise if the event occurs and which costs are already in the baseline.
This article uses positive impacts for additional costs and negative impacts for savings. Other decision models use negative values for losses because they measure profit. Both conventions work. Mixing them does not. State the convention on the register before colleagues copy rows from a different template.
| Input | Meaning | Check before approval |
|---|---|---|
| Probability | Chance within the stated horizon | Is the window defined? |
| Impact | Extra cost if the event occurs | Is it already in the base? |
| Response fee | Certain intervention cost | Is it paid regardless of outcome? |
| Residual risk | Exposure after response | Is elimination assumed too casually? |
| Owner | Responsible decision-maker | Can this person act on the trigger? |
An impact is conditional, not already probability-weighted. Entering an expected loss in the impact column and multiplying it again understates exposure. Use one currency and price basis. Where timing is material, apply a consistent discounted-cash-flow framework rather than adding distant nominal amounts as if they occurred together.
Worked register: reduce expected cost with expediting
A contractor has a $500,000 base for a short installation. None of the following incremental exposure is already included. Amounts use current dollars and timing effects are ignored over the short horizon.
| Threat | Probability | Impact | EMV |
|---|---|---|---|
| Late equipment | 20% | $30,000 | $6,000 |
| Unsuitable ground | 10% | $50,000 | $5,000 |
| Controls correction | 25% | $12,000 | $3,000 |
| Total | — | — | $14,000 |
The weighted loss is $14,000. Base plus mean allowance is $514,000, assuming the register captures the relevant uncertainty and impacts are additive. This is not a statement that $514,000 is sufficient at any particular confidence level.
For a certain $3,000 expediting fee, the team assumes delivery probability falls from 20% to 5%, with the same conditional $30,000 consequence. Residual delivery EMV is $1,500. Ground and controls remain $5,000 and $3,000. Residual threat exposure becomes $9,500.
Enter a revised base of $503,000, because the committed response is now planned spending, and the residual probabilities in the calculator. Base plus allowance becomes $512,500. Compared with $514,000, the response improves expected total cost by $1,500. The gross exposure reduction is $4,500; subtracting the $3,000 fee explains the net difference.
For an additional distribution check, explicitly assume independent occurrences. The chance of no event before mitigation is 0.80 × 0.90 × 0.75 = 54%. After mitigation it becomes 0.95 × 0.90 × 0.75 = 64.125%. Independence is needed for these joint probabilities, not for summing expected costs.
A $50,000 ground event alone exceeds the $9,500 residual average. That is not a calculation failure. It demonstrates why a mean allowance and funding sufficiency are different. A sponsor seeking a budget quantile needs a cost distribution, including dependencies and other uncertainty, rather than a relabeled EMV total.
Worked supplier decision: a premium can offset risk
Supplier A charges $100,000 and has an assumed 30% chance of lateness. Lateness adds $40,000 of standby and temporary-operation costs. Supplier B charges $108,000 and has an assumed 10% chance of the same consequence. Purchase prices are certain; quality differences, financing and other penalties are excluded.
For A, on-time cost is $100,000 and late cost $140,000. Expected cost is 0.70 × $100,000 + 0.30 × $140,000 = $112,000. For B, the corresponding outcomes are $108,000 and $148,000. Its expected cost is also $112,000. B’s premium exactly offsets its lower expected delay loss.
| Option | Planned cost | Delay probability | Expected delay loss | Expected total |
|---|---|---|---|---|
| A | $100,000 | 30% | $12,000 | $112,000 |
| B | $108,000 | 10% | $4,000 | $112,000 |
| A with expediting | $103,000 | 20% | $8,000 | $111,000 |
A offers a $3,000 expediting package reducing assumed lateness to 20%. It is paid in either outcome. The revised expected total is $100,000 + $3,000 + 0.20 × $40,000 = $111,000, which is $1,000 below either unmodified option.
The break-even response fee is the expected loss reduction: (0.30 − 0.20) × $40,000 = $4,000. Below that fee, expected cost improves; at that fee, it ties. This is a useful negotiation boundary under the assumptions, not a valuation of every nonfinancial consequence of late delivery.
B still has the lowest modeled chance of disruption. A sponsor with an immovable opening date may choose it despite its higher mean compared with expedited A. Explain both the expected-cost ranking and the downside being accepted. An average saving cannot settle a question involving safety, compliance or a critical operating obligation.
Dependencies and opportunities need explicit treatment
Expected values add even when events are correlated, provided the probabilities and additive consequences describe actual exposure consistently. Correlation changes the distribution and the chance of a large aggregate loss. If one event changes another event’s probability or duplicates its cost, repair the model before summing rows.
Do not list mutually exclusive alternatives as simultaneous unconditional risks. A decision tree may be more appropriate when one branch replaces another. Likewise, opportunity savings may improve net expectation without providing cash when a threat arrives. Show gross threats and opportunities separately before deciding whether one may offset the funding of the other.
The PERT calculator can describe a duration estimate from optimistic, most likely and pessimistic inputs. Its weighted duration is not automatically the probability of missing a deadline. Define the deadline event first, estimate that probability appropriately, and then convert the consequence into money using a defensible cost rule.
Keep the register usable after the estimate is approved
- Record the evidence trail. Retain the estimator, date, horizon and information supporting each judgment.
- Separate committed spending. Approved inspections and expediting are planned costs; residual exposure remains uncertain.
- Check common causes. A delayed design release can trigger multiple consequences that need coordinated treatment.
- Show sensitivity. A preferred response that changes under modest probability revisions needs more evidence.
- Update at milestones. Retire resolved events and reconcile actual response costs instead of carrying obsolete exposure indefinitely.
The recommendation should state the mean result, the main downside and the reason for the funding policy. If probabilities are weak judgments, label them. Extra decimal places do not substitute for evidence. A register is defensible when another person can trace its assumptions and challenge them without reverse-engineering an unexplained contingency percentage.
Calculators used in this guide
Frequently asked questions
Must risks be independent to sum their EMVs?
No. Expected values add for additive costs even when events are dependent. Independence matters when calculating joint probabilities and the full distribution.
Can opportunities offset threats?
They can reduce net expectation, but uncertain savings may not fund a loss when it occurs. Show them separately before approving an offset policy.
Is EMV a P50 budget?
No. EMV is a mean; P50 refers to a median or corresponding funding quantile. Skewed and discrete distributions can make them quite different.
What if probabilities are weak estimates?
Label them as judgments, identify their owners and test a plausible range. Collect better evidence when the preferred response changes within that range.
Where does mitigation spending belong?
Once approved, a certain response fee belongs in planned costs. The risk register should then show remaining probability and impact, not count the response again as uncertain exposure.
Sources and standards
About the author
Dr. Hassan Khames Eliwa holds a PhD in Construction & Built Environment and has more than 20 years of international project controls experience. He is Director of Eliwa Project Controls Limited in Auckland and founder of PMMilestone.org and PMMilestone.com.
Buy me a coffee