Retention is ending.

The questions it raised aren’t.

Will Sarhangian

Managing Director

First Demolition Services

For years the industry has argued about whether retention protects clients or punishes contractors. The government has now moved on it: the Commercial Payments Bill, currently before Parliament, is set to ban withholding retention under construction contracts.

Plenty of firms are celebrating getting their 3 to 5% back. That is both a narrow reading and a shallow first-order effect. I want to make a different point, from both our vantage point as a strip-out contractor and as students of the industry as a whole: the cost of retention was never the strongest argument against it. The mismatch between what it costs and what it secures is, and how this was balanced across parties. The changes proposed will reverberate through these linkages, in these second-order but deeper effects.

Retention was always an awkward fit for our scope. The mechanism exists to secure a client against latent defects during a rectification period. But our work is removal; when a floor is stripped back to slab and the arisings are gone, there’s very little to hold 5% against. Snagging notwithstanding, it is very hard to make the case for anything being deserving of a year-long, interest-free security, or longer still when release is tied to Main Contract PC. It is harder still to see how a defect in that work could survive unnoticed all the way through to Main Contract PC.

Cash flow and solvency exposure to the holders of the retention will be the focus of most commentary, and they are worthy considerations. But cash flow is the cost of the mechanism, not the case against it. The case against it is that the security often bears little relation to the risk it claims to cover. One of the wider issues retention can sometimes mask is accountability in sign-off. When we complete an enabling package and hand over to a main contractor appointed after us, responsibility for confirming our works can drift to parties who had no part in our appointment and may not even be privy to our Contract or its precise scope. In other circumstances, our demobilisation is sanctioned by operations staff who claim no commercial responsibility.

The arithmetic nobody says out loud

There is an arithmetic problem beneath all this that rarely gets stated. Main contracting runs on thin margins, with net profit typically in the low single digits. Retention runs at 5% of the works. On most projects, the retention held across a main contractor’s supply chain will comfortably exceed the profit it stands to make on the job itself. A buffer larger than the margin it sits above cannot really be said to be funded by the party holding it. At that scale, retention becomes difficult to describe as security proportionate to risk, and starts to look more like working capital, provided by the supply chain.

Which is why “the main contractor should simply carry the risk itself” was never realistic. You cannot self-insure a 5% exposure out of a 2% margin; the arithmetic doesn’t close. So once the float is gone, the system can only re-settle three ways: prices rise to fund the risk explicitly, margins compress further (they can’t, for long), or main contractors and professional teams get appointment, scope and sign-off right so there is less to secure in the first place. Only the third is an improvement, and it is a difficult one to enact. The other two just move the cost around.

We don’t yet know how this shakes out

And it’s worth being honest that it could go either way.

At one pole, the ban does what its backers hope. Diligent, well-delivered specialist contractors stop being treated as a source of security their work never required. Their retention was only ever held because they were easy to hold it from, not because their work was ever in doubt. Main contractors can no longer treat downstream retention as a hedge or a cashflow line of their own, so the burden of getting it right (who they appoint, how clearly scope is defined, how sign-off actually works) lands where it should.