Contingency reserve vs management reserve, and how to size them

The difference in one table

Every estimate carries uncertainty. Reserves are how you budget for it openly instead of padding each line item. The two reserves answer different questions: contingency asks "what will the risks we know about probably cost?", while management reserve asks "what if something happens that nobody listed?"

Aspect	Contingency reserve	Management reserve
Covers	Identified risks (known unknowns)	Unidentified risks (unknown unknowns)
Sized from	The risk register: EMV, or a quantitative analysis	Organisational policy, project size and novelty
Part of the cost baseline?	Yes	No, it sits on top of the baseline
Included in BAC for earned value?	Yes	No
Who releases it	The project manager, within agreed rules	Sponsor or management, through change control
Effect of using it	Baseline unchanged; money moves from reserve to work	Baseline increases by the amount released

The structure stacks like this: base estimate + contingency reserve = cost baseline, and cost baseline + management reserve = project budget. This is the layering described in the PMBOK Guide, and most cost control frameworks use the same idea under different names.

How to size contingency reserve

Contingency should come from your risks, not from a habit of adding a round percentage. There are three common methods, from simplest to most rigorous:

1. Expected monetary value (EMV). For each risk, multiply probability by cost impact: EMV = P × I. Add the EMVs of threats and subtract the EMVs of opportunities. This works well when you have a reasonable risk register and no simulation tool.
2. Quantitative analysis. Run a Monte Carlo simulation of total cost, then set contingency as the gap between the base estimate and a chosen confidence level such as P80. Use this on large or unusual projects.
3. Percentage of base estimate. A fixed percentage by project type. It is quick, but it ignores the actual risk profile, so treat it as a cross-check rather than the answer.

Schedule contingency works the same way, in days instead of money. You can add expected delays from the register, or read a P80 finish date from a schedule risk analysis. Our guide to Monte Carlo and P80 dates explains the time side.

Worked example: building the budget

A fit-out project has a base estimate of $400,000. The risk register holds three threats and one opportunity:

Risk	Probability	Cost impact	EMV
R1 Ceiling services clash needs redesign	30%	$40,000	$12,000
R2 Imported glazing delayed, extra site costs	20%	$25,000	$5,000
R3 Asbestos found in existing walls	50%	$8,000	$4,000
O1 Reuse existing raised floor (saving)	25%	−$10,000	−$2,500
Contingency reserve			$18,500

Check: 12,000 + 5,000 + 4,000 − 2,500 = 18,500. The cost baseline is $400,000 + $18,500 = $418,500. This is the BAC you use for earned value.

The organisation's policy (an example, not a norm) sets management reserve at 5% of the base estimate for this kind of project: 0.05 × $400,000 = $20,000. The project budget is $418,500 + $20,000 = $438,500.

Notice that contingency is less than any single big risk. That is normal. EMV assumes not every risk will happen, so the pool covers the expected total, not the worst case. If R1 alone would sink the project, add a specific fallback plan or a higher confidence level instead of hoping the pool is enough.

Drawing down the reserves during the project

Continuing the example, three things happen:

• R3 occurs and costs $8,000. The project manager moves $8,000 from contingency to the work package. Contingency falls to $18,500 − $8,000 = $10,500. The cost baseline stays at $418,500.
• R2 closes because the glazing arrives on time. Re-assess what the open risks still need: R1 ($12,000) and O1 (−$2,500) give $9,500. The remaining contingency of $10,500 exceeds that by $1,000, which can be released back to the sponsor if your rules allow.
• A supplier goes out of business, a risk nobody listed, and replacing it costs $15,000. This is a management reserve case. The project manager raises a change request, management approves it, the reserve falls from $20,000 to $5,000 and the cost baseline rises to $418,500 + $15,000 = $433,500.

The rule of thumb: identified risk, use contingency; unidentified event, ask for management reserve through change control; scope change requested by the client, neither reserve, because that is a new piece of work with its own budget.

Common mistakes

• Hiding contingency in line items. Padded estimates cannot be tracked or released, and everyone spends them. Keep the reserve as a visible line.
• Using contingency for scope changes. This drains the pool silently and leaves real risks unfunded.
• Including management reserve in BAC. It distorts CPI and every EAC that depends on BAC. Earned value uses the cost baseline only; see the earned value guide.
• Never re-assessing. Contingency should track the open risks. Recalculate it whenever risks close, occur or change.
• Forgetting opportunities. They reduce the reserve you need and deserve owners too.
• Treating one percentage as universal. A routine repeat project and a first-of-a-kind project do not carry the same uncertainty.

How to do this in Critova

In Critova each risk in the register carries a probability, cost and schedule exposure, and its EMV, so the contingency figure is the sum you can see in the risk register and on the risk dashboard. Threats and opportunities sit in the same list, so opportunities reduce the total. When an unidentified event needs management reserve, raise a change request: it records the budget and date impact and goes through approve or reject. For the schedule side, the Monte Carlo analysis gives P50, P80 and P90 finish dates, so you can size a time contingency from the gap between the deterministic date and P80. Critova simulates the schedule, not total cost, so a cost P80 still needs a separate cost model. See risk management in Critova.