Cost risks, contingencies and tolerances

A hidden contingency is built into the allowances by setting unit rates or quantities cautiously. It has three drawbacks: It is no longer identifiable in the result.

It cannot be explained to the client, because it is not visible. It dissolves under tender comparison, because real prices refute the cautious allowances. And it prevents any viability assessment, because a project estimated too cautiously can fail on apparent lack of return although it would be sound.

A stated contingency, by contrast, is a line of its own with three entries: amount, reason and release event. The third is almost always forgotten and is the most useful, because it makes the contingency testable.

Risk	Release event
Unknown ground conditions	receipt of the ground investigation
Unknown existing fabric	completion of opening-up works
Unclear consent conditions	grant of consent
Uncertain market	receipt of tenders
Open user requirements	completion of user consultation

The advantage shows as the project proceeds: the contingency falls stepwise and traceably, and its reduction is a robust progress indicator. A hidden contingency delivers none of that.

The contingency is to be distinguished from a risk allowance for named individual risks and from price movement up to procurement. The latter is not a contingency but a forecast and belongs as its own item with its calculation shown.

Fixed values circulate for the accuracy of the stages: plus or minus 30 per cent for the cost estimate, 20 for the cost calculation, 10 for the tender sum. They are routinely attributed to DIN 276.

The standard does not contain these values. It describes the stages by purpose, basis and degree of detail but prescribes no permissible deviation. The percentages are values drawn from professional literature.

They remain usable nonetheless, namely for communicating uncertainty to the client. They are not usable as justification for a deviation, since a different measure applies there.

Whether and for what a designer must answer on a cost overrun depends first on what was agreed. Only then does the question of tolerances arise, and not the other way round.

Without an agreed cost ceiling, the designer owes a cost determination free of defects. The determination is a partial result owed; omitting it is a defect. Where the result determined is exceeded, liability arises only on demonstrable, gross errors in the determination, or where actual costs diverge substantially. In this constellation the designer is granted a tolerance band; magnitudes of roughly 20 to 25 per cent are cited. For so-called Sowiesokosten, costs that would have arisen in any event, they are as a rule not liable.

With an agreed cost ceiling the position changes fundamentally. The ceiling is as a rule not a fault-independent guarantee, but it is a Beschaffenheitsvereinbarung, an agreement as to a characteristic of the work. Exceeding it therefore constitutes a defect in the design service.

Whether a tolerance band still applies in that second constellation is not answered uniformly. Part of the case law grants no tolerance band and treats the overrun as a breach of duty without more, unless the contract contains indications that the sum is not a strict limit. Other decisions apply a tolerance threshold even to such an agreement and allow liability only above that threshold.

For practice this yields no rule of calculation but a rule of caution: do not rely on a tolerance band where a ceiling has been agreed. A tolerance band does not replace an agreement.

The point with the greatest practical force is not the expressly agreed ceiling but the one that arises unnoticed. It comes into being without any contract clause and becomes visible only when someone relies on it.

A cost agreement requires two corresponding declarations of intent. The courts are, however, generous in finding such agreements: communication of the client's cost expectations can suffice, provided the designer does not expressly contradict them.

An observation in an early meeting or a figure in a set of minutes can therefore create a binding effect that was never intended. The countermeasure is simple and rarely taken: contradict unrealistic cost expectations expressly and on the record.

Two side effects compound the damage. First, on an overrun the fee may be based on the agreed ceiling rather than on actual chargeable costs. Second, professional indemnity cover is not assured in every case for agreements as to cost characteristics.

For contingencies to be justifiable, the risks must be named. A simple schedule works well, carrying each risk with four entries: description, cost groups affected, estimated range and release event.

The risks differ markedly by project type, and with them the appropriate size of the contingency. A flat percentage across all project types therefore rarely fits.

Project type	Principal risks
New build on an open site	ground conditions, groundwater, market at procurement
New build within an inner-city context	excavation support, neighbours, site logistics
Conversion and modernisation	condition of the structure, service routes, contamination
Change of use	consent conditions, fire safety, accessibility
Long construction period	price movement, interest movement, availability of trades

Two observations follow from this schedule. First, most risks in existing buildings lie not in the determination but in the unknown starting condition; they fall through investigation rather than through more careful arithmetic. Second, part of the risk lies entirely outside the designer's sphere of influence, and precisely those should be shown separately, because they are attributed differently.

Three points belong settled before a cost target becomes binding. Raised only after the commitment has been given, they leave the designer negotiating from a position they weakened themselves.

Test feasibility. The standard expressly provides for this test before the target is fixed. Where schedule, standard and target do not fit together, one side must be adjusted or the contract not concluded.

Name the scope of the target. A ceiling relates to particular cost groups, in practice frequently to cost groups 300 to 500 or 200 to 500. Which groups are covered and whether net or gross applies belongs expressly in the contract.

Reflect the dynamics. Instead of a rigid figure, the ceiling can be tied to the result of the cost calculation as approved by the client. The duty to inform should be regulated alongside, that is the obligation to notify without delay on a foreseeable overrun and to propose measures.

The third point is the one that best balances both sides' interests: the client gains cost certainty on the basis of a tested determination, and the designer does not answer for a figure named before any design existed. Each side gives something up, which is what makes the arrangement hold.