On a large project, two families of undertakings circulate in parallel and resemble each other enough to be confused. On one side insurance policies, which indemnify damage. On the other bonds and contract guarantees: bid bond, advance payment bond, performance bond, retention bond. The latter are often issued by insurers, carry the word guarantee, are negotiated in the same meetings and filed in the same binder. They do not, however, follow the same logic, and the gap shows on the day one of them is called.
The difference comes down to one sentence. Insurance covers a fortuity: it indemnifies the insured for damage it suffers, and the premium is the statistical price of that risk. A bond guarantees an obligation: it assures the beneficiary that an undertaking given by a third party will be performed, and the commission is the price of a credit line granted to that third party. The consequence is that a guarantor who pays always turns against its principal: on a bond, the defaulting contractor reimburses in full, which has nothing to do with an insurer's subrogation against a liable third party. A bond is not a transfer of risk, it is a deferral of cash.
That nature explains the distinction that decides everything, the one between an on-demand guarantee and an accessory suretyship. An on-demand guarantee is paid on the beneficiary's simple call, conforming to the terms of the guarantee, without the guarantor being able to raise anything drawn from the underlying contract: neither the alleged non-performance, nor the dispute, nor even pending litigation. An accessory suretyship, by contrast, follows the fate of the guaranteed obligation, which allows the guarantor to raise the defenses the contractor could have raised itself. Two one-page documents, carrying the same title in the binder, and not producing the same effect.
The on-demand mechanism is brutal and that is its purpose. The beneficiary calls, the guarantor pays within days, the contractor is debited, and the argument about whether the call was well founded happens afterwards, between contractor and project owner, with the money already at the latter's. The reversal of bargaining power is total: the party that would have had to prove its right collects first and defends itself as a defendant, and the party that contested must now sue for restitution. The only limits are a manifestly abusive or fraudulent call, whose demonstration is demanding and made in haste, before a judge, within days.
An accounting and financial consequence grafts onto this, and it often escapes technical departments. Bonds consume credit lines at the contractor, whether issued by a bank or by a surety insurer. A heavily bonded contractor therefore sees its capacity to commit shrink project after project, independently of its real health, and a bond call degrades in a day a financing plan built over months. It is one of the mechanisms by which a site dispute turns into corporate distress, and the module on the stopped site describes what happens next.
What a bond does not do must also be seen, because it is regularly presented as general protection. It does not repair damage: if the works are damaged, insurance pays, and the performance bond does not stand in for it. It does not cover the contractor's liability toward third parties. It does not guarantee the quality of the works beyond the period and amount it names, generally a fraction of the contract. A project owner relying on a five percent performance bond to protect against a contractor's failure actually holds five percent, which is useful and bears no relation to the real cost of replacing a contractor mid-project.
The useful reading therefore consists of filing each undertaking in the right box before it is needed. For each document in the binder, three questions: does it guarantee damage or an obligation, which decides whether it is insurance or a bond; if a bond, is it on demand or accessory, which decides what can be raised against a call; and does the guarantor have full recourse against the contractor, which decides who really carries the burden in the end. Three questions per document, an hour for a whole project, and a confusion avoided at the moment it would cost most.
On a 96 million euro contract, a project owner holds a 5% on-demand performance guarantee, that is 4.8 million, and an 8 million advance payment guarantee, also on demand. A dispute opens in March over the quality of one package and the attribution of four months of delay, each side blaming the other. On April 14 the project owner calls both guarantees for their full amount. The guarantor pays on April 22. The contractor, whose annual profit is around 6 million, disputes everything and sues for restitution. Separately, a fire destroys works in progress in May to the tune of 3.2 million. Where do the parties stand?
Three things the file presents together must be separated. On the guarantees, the on-demand form produces exactly the effect it was designed for: the guarantor can raise neither the dispute, nor the contest, nor the pending action, and it pays 12.8 million on April 22. The contractor is debited the same amount by its guarantor, immediately and in full, because a bond is not a transfer of risk but a deferral of cash. It therefore takes, within a week, a loss of more than twice its annual profit, and it will have to sue for restitution, as claimant, on a merits question not yet decided. The reversal of bargaining power is total and it is the product of two words in a one-page document. The second point is the one the confusion of families causes people to miss: the May fire comes under insurance, not under the guarantees, and the performance guarantee called in April covers not one euro of it. The 12.8 million is not available for rebuilding, it guaranteed an obligation to perform, not damage. The third point is what follows, and it is predictable: a contractor losing 12.8 million of credit lines in a week sees its capacity to commit collapse across all its other projects. The March dispute, which was about the quality of one package, becomes corporate distress in April, and that is the starting point of the module on the stopped site.
- 01Insurance covers a fortuity and indemnifies damage; a bond guarantees an obligation and the guarantor has full recourse against its principal: it is not a transfer of risk, it is a deferral of cash.
- 02On demand or accessory: the first is paid on a simple call with nothing raisable against it, the second follows the fate of the guaranteed obligation. Same title in the binder, opposite effects.
- 03An on-demand call reverses bargaining power: the beneficiary collects first and defends later, the other must sue for restitution, and manifest abuse must be shown within days before a judge.
- 04Bonds consume credit lines: a call degrades a financing plan in a day, and it is one of the routes by which a site dispute becomes corporate distress.
- 05A performance bond repairs no damage and is worth the percentage it names, bearing no relation to the real cost of replacing a contractor mid-project.