DEMITONResearch / 2026 report
What the five diseases cost an Australian civil job, 2026
disease-economics by Demiton. Published 10 October 2026 from corpus release v5-17-g86fab48; each figure’s own page carries the current record.
On 115 Australian civil infrastructure projects, total rework averaged 10% of contract value (Love, Edwards, Watson and Davis 2010). Planned gross margin on Australian civil and transport work sits at about 9.9% (Love et al. 2017, with two later studies from the same group at 6% to 9%). If both numbers held on the same job, rework alone would take the whole margin. They come from different samples, so read that as the scale of the risk, not a measured result.
This record covers the five ways a contractor loses margin: rework, missed claims, disputes, cost drift and lapsed compliance. For each figure it gives the source, the sample, a confidence grade and what the figure does not measure. Every figure below is a current record; its page on research.demiton.io carries the calculation and the full caveat.
Rework
| Figure | Source and sample | Grade | What it does not measure |
|---|---|---|---|
| 10% of contract value, total rework including indirect cost | Love, Edwards, Watson & Davis 2010; questionnaire covering 115 Australian civil infrastructure projects | Moderate | Respondents’ estimates, not cost records |
| 12.3% of contract value, 6.4% direct plus 5.9% indirect | Love & Edwards 2005; national survey of 161 construction professionals, mixed sectors | Moderate | A perception survey; designers’ and project managers’ estimates differed significantly |
| Contractor bore the cost on 93% of projects; 23% of yearly profit lost | Love, Ika, Ahiaga-Dagbui, Locatelli & Sing 2019; 98 projects with full cost data, one contractor, six years of ledgers | Moderate | Not a share of contract value. Recorded rework was only 0.39% of contract value, because ledgers undercount it |
Missed claims
| Figure | Source and sample | Grade | What it does not measure |
|---|---|---|---|
| 36.9% of the amount claimed sat in Queensland adjudications dismissed for no jurisdiction | Adjudicate Today 2023; 246 published Queensland decisions, March 2022 to August 2023 | Moderate | A nominating authority’s own tally. Failure on procedure (timing, form), not merits; other routes to recovery may remain |
| 10.6% of original contract value in client-approved change orders | Love, Irani, Smith, Regan & Liu 2017; 67 public infrastructure projects, one contractor, 2011 to 2014 | Moderate | The size of the pot, not a loss. No study measures how much of it contractors fail to claim |
| 9% of project cost in contract variations | Queensland Audit Office 2022 (Report 16: 2021-22); the Queensland Government projects examined, mostly schools and public buildings | Moderate | A handful of audited building projects, not civil work. Money the client agreed to pay |
| AU$870 to AU$2,010 per working day of time-related site overhead on a $5M job (AU$17,400 to AU$40,300 over 20 wet days) | O’Leary & Vaz-Serra 2025; 30 senior Australian estimators | Low | A worked example, not a measurement. One input is a UK 2002 figure read second-hand; the survey was mostly building work |
Disputes
| Figure | Source and sample | Grade | What it does not measure |
|---|---|---|---|
| 12% (NSW), 18% (Qld), 35% (Vic) of the amount claimed was determined in adjudication, 2018 to 2023 | Liu (DGA Australia) 2023; state regulators’ statistics, 9,784 applications, 5,657 determinations | Moderate | The denominator includes the roughly 40% of applications that never reached a determination, so it understates recovery on decided disputes |
| 50.9% of the amount claimed awarded once adjudicated, NSW 2011-12 | NSW Department of Finance & Services 2012 (statistics by UNSW); 779 released determinations | High | One year, fourteen years old, decided disputes only. Not comparable with the 2018 to 2023 ratios |
| 14.9% of the amount claimed awarded, Queensland 2020-21 to 2024-25 pooled; 8.6% to 45.6% by year | QBCC annual reports; 983 released decisions | Moderate | Pooled dollars, so one A$634m claim in 2024-25 dominates |
| AU$288,000 lost per unpaid subcontractor, Queensland | Special Joint Taskforce (Byrne) 2019; 146 submissions, mostly small contractors | Moderate | Self-reported and self-selected. A per-case figure, never multiplied by a contract value |
Cost drift
| Figure | Source and sample | Grade | What it does not measure |
|---|---|---|---|
| 7% loss at completion, a 16-point swing from a 9% tendered margin | Ryan & Duffield 2017; 28 completed Australian megaprojects over A$500m, Tier 1 open-book data | Moderate | Megaprojects only, selected for risk. Not the loss on a typical civil job |
| 23% mean overrun from contract award (median 12.3%) | Love, Zhou, Edwards, Irani & Sing 2017; 16 rail projects, one contractor, mostly WA | Moderate | Mostly scope variations the owner paid, not contractor losses |
| 24% over the first promised cost | Terrill & Danks (Grattan Institute) 2016; 836 Australian transport projects of $20m or more, 2001 to 2016 | Moderate | The owner’s overrun, mostly before a contractor is signed. Median overrun is 0% |
Lapsed compliance
| Figure | Source and sample | Grade | What it does not measure |
|---|---|---|---|
| AU$1,200 to AU$97,408 per accident borne by the employer, short absence to fatality | Allison, Hon & Xia 2019; Safe Work Australia 2012-13 data plus three Tier 2 case studies | Moderate | Not independent of the Safe Work data; no average per accident. Penalties are separate |
| 19% of the total cost of work-related injury falls on employers, premiums included | Safe Work Australia 2015; national estimate, AU$61.8bn total, construction AU$5.84bn | High | Societal cost, not the contractor’s bill. Latest edition covers 2012-13 |
What the record cannot tell you yet
- How much of the variation pot a typical contractor fails to claim. The 10.6% is the pot; the loss inside it is unmeasured.
- How many unclaimed wet days a real civil job loses. The wet-day figures are worked examples.
- What rework costs in the books. Surveys put it near 10%; one contractor’s ledgers recorded 0.39%. The gap is the finding: most rework is never written down as rework.
Method
Every figure is a current record in the disease-economics corpus (release v5-17-g86fab48); withdrawn records are excluded. Each record was read from its source on a recorded date and holds the value, the sample, the denominator, a confidence grade (high, moderate or low) and a caveat naming what it does not measure. Figures are estimates with stated confidence, not measurements of any one job. Only a share of contract value may be multiplied by a contract value; per-incident and per-day amounts stay in their own currency and price year. Any record can be disputed through the public repository, github.com/demitonapp/disease-economics.
How to cite
Data licensed CC BY 4.0. Credit “disease-economics by Demiton” with the figure’s URL, for example: disease-economics by Demiton, “Mean total rework cost, 115 Australian civil infrastructure projects”, https://research.demiton.io/rework/rework_total_civil_survey/ (release v5-17-g86fab48). Quoted passages remain their authors’.
Demiton watches a contractor’s own registers for these five, so the figures above become that job’s figures.