Chapter 12: Global claims and causation

Every claim for money for delay or disruption has the same three elements, because the money is compensation. It puts the contractor back where it would have been had the employer's event not happened. That needs an event the employer answers for, an effect on the work, and a loss the effect produced. Akenhead J set out the three in Walter Lilly (TCC, 2012). The contractor must prove each of them on the balance of probabilities ([486(a)]). The balance of probabilities is the civil standard: more likely than not. The same three apply whether the claim is for loss and expense under the contract or for damages (Lord Macfadyen in John Doyle (OH) [35], quoted in Walter Lilly [480]). Either way the contract must give the right, or there must be a cause of action (SCL Protocol, Core Principle 18). Chapter 13 covers the two routes.

Time is easier to prove than money, because time has a model to reason with. The programme's logic shows which activities drive completion, so an expert can ask what an event did to the critical path (chapter 2). Cost has no such model. A cost ledger records what was spent, by trade and by month. It does not record why. Suppose output fell in the months the employer's changes arrived. The same months may also have brought bad weather, a new and slower gang, and the contractor's own rework. Each of them pushes up the same costs.

The test for money is usually the but-for test. Would the loss have happened without the employer's event? If it would, the event did not cause it, and the employer does not pay for it. That is why concurrent delay gives the contractor time but not money (chapter 8). In Thomas Barnes (TCC, 2022) counsel agreed that, depending on the contract's words, a contractor recovers loss and expense only where it satisfies the "but for" test ([118]). The judge proceeded on that agreed basis.

The practical way into a disruption claim starts with the loss and works back. I call this working top-down. The SCL Protocol, the Society of Construction Law's guidance on delay and disruption, starts with a review of productivity over time: when it fell, and on which work (18.7). Next, find what caused the fall, from the records and the people on site, before deciding how to measure it (AACE 25R-03, s.C.2). Then take out everything that is not the employer's. The productivity loss caused by all other events must be excluded from the claim (18.6). Disruption is not simply the gap between what was planned and what happened (18.6), and the burden of proving that it caused loss stays with the contractor (18.8).

The order follows from causation. The loss is the thing you can see: more hours spent than the work should have taken. Its causes have to be inferred. If you start from each event and add up its supposed effects, you invite double counting and charge the employer for losses it did not cause. If you start from the loss and strip out the other causes, what is left at the end is what the employer caused. It is the money-side twin of delay analysis that starts from the critical delay and looks back for its causes (chapter 7).

How to measure lost productivity, including the measured mile, is the subject of chapter 11.

Most disruption claims begin with a coincidence in time. Output fell in the weeks the employer's changes arrived. That correlation is evidence of cause. It is not proof, because other things were happening in the same weeks. My view is that close correlation, backed by records, can carry a claim to the balance of probabilities. Close means the dip is on the work the event touched, and it starts and ends when the event does. Backed means the records made at the time show the disruption and show that the other causes did not produce it.

The tribunal decides causation as a question of fact (Walter Lilly [377]), and it starts with the documents made when the work was done. In Van Oord v Allseas (TCC, 2015) the daily progress reports were signed by both parties ([50]). Coulson J treated them as a useful starting point for working out what was happening on site ([51]). They said very little about standing time or disruption, and the large claims came months or even years later. Standing time is a plain fact that a site records when it happens. The judge said it was precisely what the reports needed to include ([54]). So its absence, and claims made long afterwards, tell against it. Both facts undermined the credibility of the claims ([54]). He described them as essentially global ([47]).

Where the records do link the event to the extra resource, the costing is easy. In Walter Lilly the contractor showed which extra staff and resources each delay needed. Once that link was proved, Akenhead J said, the cost was found by counting the extra man weeks and their salary cost ([491]). The contractor's cost system recorded all the costs on the three units being built together ([488]), and he accepted its figures as reasonably accurate ([499]). He did not require a daily allocation made at the time. With staff shared across three units, that would have been immensely artificial on those facts ([488]).

Indirect resources, such as supervisors and standing plant, need their own proof. The Protocol asks the contractor to show the correlation between those costs and the loss of productivity in the direct resources (18.14). A foreman's extra weeks are recoverable only if the gangs he ran were in fact disrupted.

Silence in the records usually hurts a claimant, as it did in Van Oord. Silence of another kind can occasionally help. In Cleveland Bridge the main contractor made only limited complaint about the erectors, and that supported a minimum award. What records to keep, and how tribunals read them, is chapter 14.

English courts use Akenhead J's description. A global claim identifies many causes of delay and disruption, a total cost on the job and a net payment from the employer. It then claims the balance between costs and payment, attributed by inference to the causes relied on (Walter Lilly [484]). "Global" and "total cost" are labels, not terms of art ([484]). The Court of Appeal said the same of "global claim" and "all-or-nothing claim" in Alame v Shell ([95]). A label settles nothing. The question is always what the claimant has proved.

The idea has a sound logic. If every event is one the employer answers for, there is no need to prove which part of the loss each caused, because the employer pays for all of it anyway. Lord Macfadyen said so in the Scottish case John Doyle in 2002 ([35], quoted in Walter Lilly [480]). Causation is then proved by inference. The gap between cost and payment is the evidence, and the list of the employer's events is the explanation. The Inner House, hearing the same case on appeal, called that in essence the nature of a global claim (John Doyle [10]).

The weakness is what else sits in the gap. The difference between cost and payment holds the effect of every cause on the job. That includes the employer's events, but also bad weather, the contractor's own inefficient working and any shortfall in its price. The Inner House gave bad weather and inefficient working as its own examples ([10]). In the invented example, a global claim asks the tribunal to accept that none of the rain, the rework or the price is in the £600,000.

Claimants still bring global claims, because on a troubled job the events tangle. Fifty late instructions arrive over a year. Each disrupts a little, and together they disrupt a great deal. Separating the effect of each may be impossible, or not worth the cost. That is the case the concept was made for. It lets a claimant recover where the loss comes from many events, all the defendant's responsibility, and it is impossible or impracticable to separate the loss each one caused (Alame [94]). The idea is old. An early case is J Crosby & Sons Ltd v Portland UDC (1967) 5 BLR 121. The judge saw no reason why an arbitrator should not award the separable items one by one and the rest as a composite whole (as quoted in Walter Lilly [474]). He attached two conditions: no duplication and, under that contract, no profit element.

A global claim need not cover the whole claim. The Inner House described the American modified total cost claim. The contractor proves some items in the ordinary way and uses the global method only for the rest ([11]). Delay costs such as site establishment are often easy to tie to a period of delay; disruption costs are harder ([11]). And the method is the claimant's choice. It is an additional weapon, not one a claimant must use (Alame [94]). A court cannot make a claimant run its case that way against its will ([96]).

The leading English statement is Akenhead J's in Walter Lilly at [486]. There is nothing in principle wrong with a total or global cost claim ([486(d)]). Subject to any restriction in the contract, the contractor simply has to prove its case on the balance of probabilities ([486(a)]). But in many cases, though not all, it faces added evidential difficulties ([486(d)]). It must generally show that the loss, meaning the difference between cost and payment, would not have been incurred anyway. That means proving its accepted tender was priced well enough to make some net return. And it must show, in effect, that no other matters caused the loss ([486(d)]). What happens if others did comes next.

Each requirement follows from the logic of the inference. The gap between cost and payment points to the employer only if it would not have been there anyway. If the tender was too low, the contractor would have lost money on a job where nothing went wrong, and part of the gap is its own pricing. If other things went wrong, part of the gap is theirs. So the contractor proves the two things that make the inference safe: a price that worked and no rival cause. Nothing here is special to global claims. It is the but-for test, applied to the whole gap at once.

The burden of proof stays with the contractor. Counsel in Walter Lilly argued that it moved to the employer, and Akenhead J held that wrong ([486(d)]). That is the ordinary rule: the party that says an event caused a loss must prove it. The employer may still lead its own evidence. It may show that the tender was so low the loss would have happened regardless, or that other events caused all or part of it ([486(d)]).

Two things are not required. The contractor need not first prove that ordinary cause-and-effect proof was impossible ([486(a)]). Nor is it barred because its own conduct made the effects impossible to untangle; Akenhead J held that argument wrong ([486(g)]). Both follow from the civil standard. A claimant may prove its case by any evidence that makes it more likely than not, and there is no set way to prove the three elements ([486(c)]). A tribunal may be more sceptical where direct linkage was readily available and not used ([486(f)]). That is no reason to reject the claim out of hand ([486(f)]). Check the contract as well. It may restrict global claims, and a notice condition precedent bars a claim to the extent it was not met ([486(a), (b)]; chapter 4).

Whether a global claim is allowed is a different question from whether it is pleaded well enough. The Privy Council's decision in Wharf Properties v Eric Cumine (1991), often cited against global claims, upheld the striking out of a pleading as hopelessly embarrassing. It was concerned with a pleading issue (Walter Lilly [476]-[477]). An arbitrator may insist on as much particularity as possible without rejecting global claims in principle (AMEC v Cadmus (1996) 51 Con LR 105, 115, 118).

Walter Lilly itself shows how a claim that looks global may turn out not to be. The job was one of three houses built together for an employer with high expectations, and design was nowhere near complete when the contractor started ([1]). Akenhead J found remarkably little design, hundreds of variations and hopelessly late information ([487]).

The employer called the contractor's preliminaries claim global. It was not. The contractor had linked its extra resources to the events and costed them from its records, so it was not on analysis a global claim at all ([491]). In the alternative, the judge found its original prices realistic, and at a level at which no net loss would have arisen had the events not happened ([492]). A national firm of quantity surveyors had reviewed them at tender stage ([492]). One risk stayed with the contractor. It had assumed a number of procurement packages and took the risk that there might be more, so the extra cost of the additional packages was not recoverable ([493]). The net sum due to the contractor was £2,330,666.26 ([658]).

The hardest question is what happens when something other than the employer's events is in the gap. Pure logic gives a harsh answer. The inference from gap to cause holds only if all the events that contribute to the loss are ones the defendant is liable for (John Doyle (OH) [36]). If a material cause is not, the gap no longer shows what the employer caused. Lord Macfadyen concluded that a global claim, as such, must then fail, and called advancing one a risky enterprise ([36]-[37], quoted in Walter Lilly [480]).

In Walter Lilly, Akenhead J softened that answer by asking what the evidence still proves. If the other cause's effect can be valued, take it out. What remains is still the employer's. Akenhead J held that an unpleaded or contractor-risk cause does not necessarily mean the contractor recovers nothing. It depends on that cause's impact ([486(e)]). His example was a £1m global loss where the accepted tender had overlooked, and not priced, a £50,000 item. The claim does not fail. The global loss is reduced by £50,000 ([486(e)]). His second example came from the case: management time spent on the contractor's own lift problems, deducted if it could be quantified precisely or at least by way of assessment ([486(e)]).

A claim that fails as a global claim may also leave something standing. The evidence may still show links between individual losses and individual events, or allow a rational apportionment of part of the loss (OH [38]). Ramsey J said much the same in London Underground v Citylink (TCC, 2007). An undermined claim will fail as a global claim ([144]). But pleaded events that caused delay may remain within it, and a claim may be established on them if the evidence gives enough basis for causation ([144]-[145]).

Deduction is not guaranteed. It assumes the other cause can be valued. If it cannot, the tribunal cannot tell how much of the gap is the employer's, so the inference from the gap fails. The SCL Protocol warns that a global claim may fail entirely if a material part of the loss had another cause and the tribunal cannot value that part on the evidence (17.5). That is the real risk for a claimant. Its own problems need not sink the claim if it has measured them. They may sink it if it has not.

Scotland reached its own middle position. The Inner House in John Doyle (2004) ruled on the pleadings before any evidence was heard, so what it said is guidance rather than a holding. It said a claim presented globally fails only where the matter the employer is not responsible for played a significant part in causing the loss ([10]). Even then, the court may treat one cause of an item of loss as the dominant cause ([15]), or apportion the loss between the causes ([16]). For delay it suggested an equal split, at least where the other cause is not the contractor's responsibility. Where it is, it may be right to deny recovery for the period of the contractor's default ([16]). But it also said the pursuer must show it was impossible or highly impracticable to trace the costs to each event ([30]). English law does not require that (Walter Lilly [486(a)]). So Scots law is softer than the pure logic in one way and stricter than English law in another. Apportioning loss here is a different thing from apportioning time for concurrent delay, which English courts refuse to do (chapter 8).

Sometimes the tribunal is sure the employer caused some loss but cannot see how much. Two principles pull against each other. A claimant that has proved a breach and some loss should not leave with nothing because the amount is hard to measure. A defendant should not pay for loss nobody has proved. In Cleveland Bridge, below, Akenhead J's answer was to award the least the evidence shows was probably caused. The difference between what was spent and what should have been spent is a ceiling. The award is a floor.

Cleveland Bridge v Severfield-Rowen (TCC, 2012) shows the method in numbers. Cleveland Bridge supplied fabricated steel for the Shard in London to Severfield-Rowen (SRS), which erected it. Deliveries were late and in some respects out of sequence ([155]). SRS counterclaimed for the disruption to its erection work, but said it could not say which individual events caused which individual periods of disruption ([152]). Akenhead J did not call it a global claim. It had the same weakness. Chapter 10 tells the acceleration side of the same case.

SRS's disruption claim was amended belatedly, in November 2012, to £225,055.43 ([152]). Its own quantum expert put the figure at £122,547.33 ([153]). He had been able to use neither the measured mile nor earned value. He had no timesheets detailed enough to identify standing time. And he accepted it was logically possible that the extra shifts came from poor planning, inefficient use of cranes, or an erection subcontractor that was not very good at its job ([154]).

The judge worked from what was reliable. The steel up to level 9 should have taken 630 man shifts to erect, and 808.5 were used up to 18 April 2010 ([155]). He was satisfied there must have been disruption in the sense of reduced productivity ([155]). The disruption could not exceed the difference, 178.5 shifts. But without detailed evidence tying lost productivity to the breaches, it would be wrong to award the whole difference ([156]). Where some more than trivial disruption must have occurred, the court makes a reasoned assessment, based on the minimum probably caused by the breaches ([156]). He found that at least 40 man shifts must have been wasted, about half a week's shifts ([156]). At £217.84 a shift, with payroll costs added, SRS was entitled to £10,256 ([157]-[158]).

One piece of evidence helped SRS, and it was an absence. Mace, the main contractor, was supervising the work closely and had made only limited complaint about the erectors. The judge thought there would have been more if SRS's erectors had been "dragging their feet" unduly ([156]). That supported a minimum award, not the whole gap. The £10,256 was under a twentieth of the sum claimed and about a twelfth of SRS's own expert's figure.

The same judgment shows the other side of the line. SRS also claimed back what it had paid its decking subcontractor for disruption. The judge took that to be at most £60,000. That part got nothing ([159]-[160]). The court had a nagging feeling that part of it could have been proved, and that was not enough to make an award ([160]). What separated the two results was evidence that set a floor.

The broad axe, a tribunal's licence to estimate a loss that is hard to measure, has the same limit. It helps to quantify a proved loss but does not prove one (Lumley Baxter v Aviva [124(v)]). In that 2026 County Court case a financial adviser proved that an investment platform's breaches disrupted its staff. The claim was not rejected for being advanced globally. It failed because nothing linked the disruption to any lost revenue or extra cost ([294]), and the adviser recovered £1 in nominal damages ([296]). It is not a construction case and binds no one, but its logic is general. Disruption is not the same as loss ([116]).

The SCL Protocol is warier of global claims than the courts are. It discourages composite or global claims made without any attempt to prove cause and effect, while noting the courts' more lenient trend (Core Principle 17). Accurate records should usually make a global claim unnecessary, and a failure to keep them is unlikely to justify one (17.1). In rare cases it accepts two stages: price what can be linked, then claim the rest as a composite whole (17.2). It advises a statement of the steps taken to link each event and of why that proved impossible (17.3). A tribunal need not dismiss a global claim out of hand, but it is not the tribunal's job to do the analysis the contractor has not done (17.4).

Parts of that ask more than Walter Lilly does. English law does not require proof that linkage was impossible ([486(a)]). Nor does it bar a contractor whose own conduct caused the tangle ([486(g)]). The Protocol is guidance, not a statement of the law (Introduction, para B). Its second edition in 2017 came after Walter Lilly. It recognised the courts' more lenient approach, adding that the risks of proceeding on a global basis remain (para K(e)). Follow its advice as good practice. Do not treat it as a rule of law.

How lost productivity is measured is in chapter 11, and how the recovered sum is priced is in chapter 13. The records that make any of this provable are in chapter 14.