Pre-tender estimates and cost control in public procurement

In public procurement the works estimate is not a single document produced once; it is a documented progression from the feasibility order of magnitude to the priced bill issued with the tender. Each level has an expected precision and a precise regulatory use.

The distinction between the developed design estimate and the priced bill issued with the tender is often misunderstood. The developed design estimate is the final estimate of the forecast cost of the works, prepared by the design team (R2431-11). The client's budget reference is the provisional financial envelope it sets before the outline and developed design studies and may refine before the detailed design studies (L2421-3): the estimate serves to check that the envelope is sufficient. If the authority has voted a budget authorisation for the project, this is also the moment to check its amount and, where needed, to ask the assembly to revise it (L1612-29 and R1612-51). The priced bill is the working document of the assessment, used to compare tenders. The two have to be coherent but they are not identical: the bill carries rate adjustments tied to the date of the tender and can differ from the developed design estimate where rates have moved between the two stages.

For local authorities, budget authorisations are optional: "Where the elected assembly so decides, the appropriations allocated to capital expenditure comprise budget authorisations and payment appropriations" (general code of local authorities, art. L1612-29, the regime common to all authorities since the 2026 financial year). Where the assembly votes them, the budget authorisation is the upper limit of the expenditure that may be committed for the project, and the payment appropriations that of the expenditure paid in the year. It is the ceiling for the project, and it governs both the placing of the contract and any variations.

• An inadequate authorisation: where the contract plus variations exceeds what was voted, the buyer has to go back to the assembly for a supplementary authorisation, which means a further resolution and therefore delay. Where a budget authorisation has been voted, committing beyond its amount requires it to be revised by a vote of the assembly, at the budget or in an amending budget decision (L1612-29 and R1612-51).

• An over-generous authorisation: an overstated authorisation ties up credit needlessly, but it is politically less risky than an inadequate one.

• The economist's role: the authorisation is calibrated on the provisional financial envelope of the project, and the developed design estimate serves to check that it is sufficient. An understated estimate leads to an inadequate authorisation and to a budgetary irregularity when the variations arrive. A sound estimate with a well-calibrated contingency protects the buyer from that.

ℹ A practical rule for calibrating the authorisation It has to cover the works contract, the design fees, the ancillary costs, surveyor, approved inspector, safety coordinator, insurance, and a provision for variations of at least 10 % of the works. For example: works at 850 000 €, design fees at 85 000 €, ancillary costs at 25 000 €, and a provision of 85 000 €, giving a recommended authorisation of 1 045 000 €. An authorisation of 900 000 € against a works contract of 850 000 € leaves no margin at all for variations.

The priced bill is what public buyers rely on to identify abnormally low tenders and to compare the tenders with one another. Its quality directly determines the quality of the assessment: a bill with wrong quantities or out-of-date rates produces a biased comparison.

• The buyer's own earlier contracts: rates from contracts placed on comparable projects are the best source, since they reflect the local market at a recent date. The economist should keep a database of the rates from the buyer's contracts.

• Published price bases and indices: national price references, brought forward with the published indices to reflect the date of the tender.

• Informal soundings: under the adapted procedure the buyer can consult contractors beforehand to calibrate its rates. That is not a competition and it can feed the bill without creating any commitment.

• Quantities taken from the drawings with no check on site: in refurbishment especially, the drawings do not always reflect the building.

• Rates not brought forward: using unit rates three years old without applying the indices can understate the contract by 15 to 25 %.

• Missing items: site establishment, cleaning, waste management, price adjustment. Items present in the contractor's signed pricing schedule but absent from the bill create a gap in the assessment that nobody can explain.

The most delicate situation is the general overrun, with every tender above the estimate. In that case everything depends on the budget allocated to the contract, determined and established before the procedure is launched: only a tender that exceeds it is unacceptable (L2152-3), and a tender above the estimate that the buyer can finance is not. Where the budget allows, the buyer can therefore accept the best-ranked tender. Where, in a tender procedure, only irregular or unacceptable tenders have been submitted, a contracting authority can move to a competitive procedure with negotiation, or to a competitive dialogue, without substantially altering the initial terms of the contract, and without a new notice if it involves only the tenderers whose tenders met the time limits and formal requirements of the tender procedure (R2124-3, 6°, and R2124-5); it can also declare the procedure discontinued and inform the candidates of the reasons as soon as possible (R2185-1 and R2185-2): the courts require a reason of general interest, which may be budgetary, and the DAJ then asks for the existence and origin of the extra costs relied on to be demonstrated. Cutting the brief back or re-splitting the packages to the point of substantially altering the initial terms requires a new procedure. Declaring the procedure discontinued and going out again lengthens the timetable: the new tender leaves contractors the minimum periods of the code, in an open tender procedure thirty-five days from the dispatch of the notice, thirty if tenders can be submitted electronically (R2161-2 and R2161-3), then, in a formal procedure, at least eleven days between notification of the rejection of the other tenders and signature (R2182-1), not counting the time to rework the file and assess the tenders. Anticipating it in the programme is essential.

At tender assessment, the design team's commitment is also checked. Where its appointment includes assisting the client in placing the works contracts, the design team commits to keeping to the forecast cost of the works set at the latest before the launch of the procedure for placing the public works contracts (R2432-3). Compliance with that commitment is checked at the end of the tender: if the tolerance threshold is exceeded, the client can ask it to adapt its studies, with no additional fee. For design appointments whose tender was launched, or contract notice sent for publication, on or after 1 January 2023, the client can ask this only if the overrun does not result from circumstances the design team could not foresee; the earlier wording set no such condition (decree no. 2022-1683, art. 3 and 8). The tolerance threshold is the one set by the design appointment; failing any mention in its particular documents, article 13.2 of the 2021 CCAG-MOE, which applies only to appointments that refer to it, calculates it with the formulas forecast cost of the works excluding VAT x 1.05 on new build and x 1.1 on refurbishment. The 10 to 15 % tolerance in the table is an analytical benchmark, with no source: it does not replace that threshold.

ℹ The duty to inform the client The economist has a duty to inform the client as soon as a risk of budget overrun appears. It flows from the design appointment, and the administrative case law sanctions design teams that let a project drift without telling the buyer. In practice: any foreseeable overrun of the authorised budget must be the subject of a written note to the buyer, setting out the causes identified and the options available. That note has to go out before the overrun is incurred, not after.

Multi-year projects, refurbishing a school over three years, renewing an estate in phases, pose a specific difficulty: the estimate for the third phase is made three years before it is built, and prices will have moved, sometimes significantly.

The rule is that estimates for future phases have to be brought forward at each approval stage, not left frozen at the level of the initial study. The budget authorisation can be voted phase by phase, which allows the budget to be adjusted at each resolution. The economist should recommend that phased approach on any project whose total duration exceeds two years.

There are six levels of estimate, from an order of magnitude at ± 30 to 40 % through to the contractual pricing schedule. A local authority's budget authorisation has to cover the contracts, the fees, the ancillary costs and a provision of at least 10 %. The priced bill draws on three sources: the buyer's earlier contracts, published price bases brought forward with the indices, and informal soundings under the adapted procedure. The commonest errors are quantities never checked on site, rates not brought forward, missing items, inconsistency with the pricing schedule, and an undocumented gap between the developed design estimate and the tender. Where every tender overruns, the options are a reduced brief or going out again. The duty to inform the client is both contractual and established by case law, and any drift has to be reported in writing before it is incurred.

The nature of the figures quoted

Amounts and rates set by a text are quoted with their article and date. The other percentages and amounts on this page, dated reference points or a fictitious example, are orders of magnitude for framing, not measurements: they substitute neither for a tender exercise nor for an estimate prepared on drawings.

• Public procurement economics in France: the economist's role end to end
• Splitting works into packages and coordinating them in public contracts
• Assessing public tenders and abnormally low offers