A shared cause can make several project costs rise together

A project may contain separate estimates for contractors, equipment hire, supervision and temporary premises, yet one delay can increase all four. Review the events behind the estimates before combining them. Separate line items are not necessarily separate risks.

GAO's cost-estimating guide explains that dependencies and correlation affect the distribution of total project costs. Although written for government programs, its treatment of shared drivers is useful when reviewing a business estimate. Ask the estimator whether a common schedule or productivity assumption reaches every cost it affects.

Begin with an operating conversation. Which costs move if commissioning takes another month? Which depend on the same supplier, exchange rate or technical redesign? Write down the mechanism and the affected lines. A correlation coefficient without an explanation of the relationship is difficult for management to challenge.

Consider a deliberately simplified hypothetical project with four cost items. Each can overrun by $10,000 with an assumed 50% probability. If those events are independent, the probability that all four overrun together is 6.25%: one-half multiplied by itself four times. The expected total overrun is $20,000.

Now assume the same four costs all rise only when one shared event occurs, again with a hypothetical 50% probability. The total overrun is either zero or $40,000. Its expected value remains $20,000, but the probability of the full $40,000 outcome is now 50%. An unchanged average conceals a very different exposure to the high-cost outcome.

These are contrasting mathematical assumptions, not measured probabilities for a real project. Actual relationships may be partial and more complicated. The example shows why reviewing only the expected total cannot establish that a proposed allowance covers the relevant downside.

When a delay directly changes the number of months of equipment hire and supervision, link those costs to the same schedule driver. Adding a second generic correlation adjustment to a relationship already captured by formulas can duplicate the effect. Ask the analyst to distinguish explicit dependencies from additional statistical assumptions.

Keep the cost boundary consistent. A delay allowance may already include temporary facilities or lost operating contribution. If those appear elsewhere in the business case, reconcile them before combining totals. Risk analysis should describe the complete consequences without charging for the same event twice.

For uncertain relationships that cannot be estimated reliably, show transparent joint scenarios. Management may understand an extra-month case better than a simulation with unsupported coefficients. Explain what is assumed to happen together and which consequences remain outside the model.

Request the total-cost range, the major shared causes and the spending dates in an adverse case. A delay can increase total cost while also postponing the revenue expected to fund it. Keep that timing effect in the connected cash plan rather than treating the allowance as the only consequence.

Ask which actions could reduce the common exposure. An earlier technical decision or confirmed installation window may improve several cost estimates at once. Record the cost and practical limits of the mitigation instead of assuming it succeeds merely because it appears in the risk register.

When the schedule or project scope changes, revisit the relationships. Retain the original estimate and explain the revised range. Management should know whether it is approving more spending, accepting a larger downside, or funding an action intended to reduce the risk.