Risk response and contingency management

Analysis tells the team which risks matter; responses decide what happens to them. Lesson 1 listed the standard responses: avoid, reduce, transfer, share and accept for threats, and exploit or enhance for opportunities. This lesson shows how to choose among them on the basis of cost and benefit, how to plan responses so they actually get done, and how to fund the risks that remain through contingency. Insurance and bonds, the main ways of transferring risk to third parties, come in lessons 7 to 10; contract clauses that allocate risk between owner and contractor come in lessons 5 and 6.

Choosing a response

A response is worth carrying out if it reduces expected loss by more than it costs, after allowing for the organization's attitude to large losses. A mitigation that costs $60,000 and reduces a risk's EMV from $180,000 to $40,000 saves $140,000 of expected loss for $60,000, a clear benefit. A mitigation that costs $100,000 to reduce an EMV of $90,000 to $30,000 does not pay on expected value alone, but may still be worth doing if the risk could cause a loss the organization cannot afford, such as a missed opening date with heavy liquidated damages and reputational damage.

Responses also create secondary risks. Dewatering can lower groundwater under a neighbour's building and cause settlement; early ordering of equipment commits money before the design is final; bonding a small subcontractor may push its price up or drive it out of the bid. A good response plan lists the secondary risks it creates and adds them to the register.

Decision trees

When a response involves a choice between options with uncertain outcomes, a decision tree helps. Squares mark decisions, circles mark chance events, and each branch carries a probability and a cost. The expected cost of each option is calculated by working back from the right-hand end.

The tree also shows what the probability would have to be for the decision to change. Here the options break even when 85,000 + 25,000p = 480,000p, so p = 85,000 ÷ 455,000, about 0.19. If the team believed heavy rain had less than about a 19 % chance, skipping dewatering would have the lower expected cost. Break-even analysis of this kind is useful when probabilities are hard to estimate: the team only needs to agree whether the probability is above or below the threshold.

Planning and tracking responses

A response that is only written in the register is not a response. Each one should be broken into actions with an owner, a due date and a cost, and the actions should appear in the schedule and budget where they belong. Dewatering needs a subcontract, a permit for discharge, a design and a start date before excavation. Early procurement of dehumidifiers needs approved shop drawings and a purchase order. The risk owner checks that actions are completed and updates the residual score once they are.

Contingency plans, sometimes called fallback plans, are different from mitigation. They are prepared in advance but carried out only if the risk occurs: a standby pump arrangement, a second shoring subcontractor prequalified in case the first fails, or an agreed procedure for handling contaminated soil if it is found. Their value lies in speed; a team that has already planned what to do loses days rather than weeks when the event happens.

Contingency: money for retained risk

After responses, some risk always remains. Contingency is the money set aside to pay for it. It is not a slush fund, a hidden profit or a reserve for scope the owner has not yet decided on. CON-300 lesson 9 shows how contractors build contingency into a bid, and CON-310 how it is tracked in the cost report. This lesson focuses on how much is needed, who holds it, and how it is released.

The distinction between contingency and management reserve is important on large projects. Contingency covers identified risks and is controlled by the project manager within agreed limits. Management reserve covers risks nobody has identified, and releasing it requires approval above the project, from a sponsor, board or executive committee. Keeping them separate stops the project from quietly spending the reserve on known problems and leaving nothing for genuine surprises.

Sizing contingency

Three methods are common, in rising order of rigour. A flat percentage, such as 5 % of construction cost, is quick but ignores the specific risks of the project. Summing the EMVs of identified risks ties the figure to the register but, as lesson 3 showed, is often exceeded. A probabilistic method reads contingency from the Monte Carlo S-curve at a chosen confidence level.

The confidence level is a policy decision. Public owners that must return to a council or legislature for more money often budget at P80 or higher, because an overrun is politically expensive. Contractors bidding competitively cannot carry a P90 contingency and still win; they typically carry less and rely on their margin and company resources for the tail. This difference in appetite is one reason owners and contractors view the same risk differently. CON-300 lesson 11 shows how competitive pressure pushes contractor contingency down.

# Contingency at different confidence levels, and the cost of each extra point (illustrative)
base = 46_200_000
levels = {"P50": 47_063_630, "P80": 47_717_848, "P90": 48_105_209}   # from the lesson 3 simulation
prev = None
for name, value in levels.items():
    cont = value - base
    line = f"{name}: contingency {cont:>10,.0f}  ({cont / base:.1%} of base)"
    if prev:
        line += f"   extra over {prev[0]}: {cont - prev[1]:,.0f}"
    print(line)
    prev = (name, cont)
bid_margin = 0.05 * base
print(f"For comparison, a 5 % fee is {bid_margin:,.0f}")

P50: contingency    863,630  (1.9% of base)
P80: contingency  1,517,848  (3.3% of base)   extra over P50: 654,218
P90: contingency  1,905,209  (4.1% of base)   extra over P80: 387,361
For comparison, a 5 % fee is 2,310,000

Moving from P50 to P80 costs about $650,000 on this project; moving from P80 to P90 costs another $390,000 for just ten more points of confidence. The curve flattens at the top because the tail is long. Deciding where to stop is a matter of risk appetite, not arithmetic.

Drawing down and releasing contingency

Contingency is spent as risks occur and released as risks expire. A dewatering allowance can be released once the parking level is backfilled; a contaminated soil allowance once excavation is finished. A well-run project plans how contingency should decline over time, based on when each risk's exposure ends, and compares actual drawdown with the plan each month. Drawdown that runs ahead of plan is an early warning that the project is consuming its cushion faster than the risks are expiring.

Release rules should be written down before the project starts. Under a guaranteed maximum price contract (CON-340 lesson 4) the contract often says how unspent contractor contingency is treated at the end: returned to the owner, shared, or partly kept by the contractor. Under a stipulated price contract unspent contingency is the contractor's, which is fair because overspent contingency is the contractor's too. On the owner side, releasing contingency to fund scope additions is a common temptation that leaves the project exposed later.

Risk-adjusted pricing

Contractors price risk in three places: in specific line items where a risk is concrete enough to estimate, in contingency for identified risks, and in the markup, which must cover general business risk as well as profit. CON-300 lesson 11 explains how competitive pressure and bid strategy affect markup. From a risk management point of view, the key discipline is to price known risks explicitly rather than burying them in markup, so that if the job is won the team knows exactly what money is available for which risk.