Rework: what it costs a construction contractor

What it is. 14.2% of original contract value, cost of design errors, direct and indirect, as estimated by survey respondents.

Sample. Questionnaire survey of 139 Australian projects. The 2012 paper gives a mean direct design error cost of 6.85% and a mean indirect cost of 7.36% of contract value; the 2014 paper gives a mean design error cost of 14.2%. 6.85 plus 7.36 is 14.21, so the two appear to describe the same sample, which the abstracts do not state outright.

What it is not. Respondents' estimates of design error cost, not measured cost records, and the abstracts do not say who carried the cost (contractor, designer or client), so it is filed as money at stake and not as contractor loss. Abstracts only: the full texts were not read. The 2012 abstract says cost did not vary significantly by project type, while the 2014 abstract reports significant differences by type, which the abstracts do not reconcile. The 2012 abstract does not name the country, so that source is filed as unknown; Australia rests on the two 2014 abstracts, which both say 139 Australian projects, and neither names the states or says how the projects were drawn, so how national the sample is cannot be told from them. It is larger than the 12.3% total rework cost the same group reports for 161 projects, so it is a cost of design errors and their effects, not a share of rework. That the 2012 and 2014 papers share one sample rests on the matching 139 and the sum 14.21, not on a statement.

Where it came from

• Love, Lopez, Kim & Kim (2014), Probabilistic assessment of design error costs
  abstract, as held in the Curtin University repository record (read through its mirror at cendekia.unisza.edu.my/neuaxis-e/Record/curtin-20.500.11937-10591/Details); the ASCE page was not opened and the full text was not read · source record
• Lopez & Love (2012), Design Error Costs in Construction Projects
  abstract, as held in the Curtin University repository record (read through its mirror at cendekia.unisza.edu.my/neuaxis-e/Record/curtin-20.500.11937-10508/Details); the ASCE page returned 403 and the full text was not read · source record

Considered and set aside

• Love, Lopez, Kim & Kim: Reports the same 14.2% for the same 139 Australian projects in a second paper by the same group, so it is the same sample read again, not independent corroboration.
• Love, Edwards & Smith: A survey of 161 Australian projects that found no significant relationship between contract documentation and rework cost. It gives no share of contract value for documentation, so it does not give a figure for this cause.
• Australian Constructors Association and Consult Australia: Says late design changes and unclear scope drive cost into major projects, with no figure, sample or data source.

rework/design_error_cost_au · AU, unknown · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 12.3% of total contract value, direct plus indirect.

Sample. National survey of 161 construction professionals about recently completed Australian projects. Mixed sectors.

What it is not. A perception survey. Design consultants' and project managers' estimates differed significantly. A 'total of 10%' appeared only in a search snippet, not the abstract, so it is not cited from this paper.

Our arithmetic. 6.4% direct + 5.9% indirect = 12.3% (both from the abstract; the sum is ours)

Where it came from

• Love & Edwards (2005), Calculating total rework costs in Australian construction projects
  abstract · source record

rework/rework_split_direct_indirect · AU · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 10% of contract value, including indirect costs.

Sample. Questionnaire survey covering 115 Australian civil infrastructure projects. The only peer-reviewed figure specific to civil work.

What it is not. Respondents' estimates, not measured cost records. Rework amount correlated significantly with cost and schedule growth (p<0.01).

Where it came from

• Love, Edwards, Watson & Davis (2010), Rework in Civil Infrastructure Projects: Determination of Cost Predictors
  abstract · source record

rework/rework_total_civil_survey · AU · F42 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. Per defect, total rectification cost to all parties (owner, builder, insurer and others), as estimated by dwelling owners; excludes owners' time and other costs.

Sample. Owners' estimates from a survey of 1,606 residential dwellings in all eight states and territories, Census-weighted, 2,574 defects reported. Of the 1,605 dwellings analysed, 1,231 are detached houses with 1,969 defects reported. The report's total cost per defect, including owners' time and other costs, is $4,366 for detached houses.

What it is not. Owners' own estimates, not invoices, and the cost is the owner side (the buyer's cost of putting a defect right), not the contractor's loss, which depends on warranty and who pays. Houses, not civil works. The average covers every defect owners reported, including some the CIE weights as caused by maintenance or alteration rather than the initial build (1,024 of 1,969 detached-house defects are weighted as initial-build). The text calls the $3,440 a cost per dwelling in one sentence, but Table 2.17 and the total-cost line give it per defect. The report does not state the price year of the estimates; 2021 is the report's date. Workmanship is not separated from design and approval failures in this figure.

Where it came from

• Centre for International Economics (for the Australian Building Codes Board) (2021), Building Confidence Report, A case for intervention
  Section 2.5 (Table 2.5), Tables 2.8 to 2.10, 2.16 and 2.17 (printed pages 24 to 36), executive summary Table 1 · source record

Considered and set aside

• Love, Matthews & Mahamivanan: Gives a mean non-conformance cost of 0.41% of contract value on infrastructure, but the abstract does not name the country, so it cannot fill an Australian cell.
• Love, Ika, Ahiaga-Dagbui, Locatelli & Sing: Recorded rework of 0.39% of contract value at one Australian contractor, but it is not split by cause and its ledgers undercount quality failures.

rework/defect_rectification_cost_detached_au · AU · F41 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. Incidence and profit impact, not a share of contract value.

Sample. 98 projects with complete cost data (of 346 projects and 19,605 rework events), one contractor, six years of ledger data, public and private clients, building and infrastructure.

What it is not. Recorded rework was only 0.39% of contract value - the ledgers undercount. Margins were 6-9%, which is why 0.39% wiped out roughly a quarter of yearly profit. The 2018 paper on the same data reports 28%; cite one and say which.

Where it came from

• Love, Ika, Ahiaga-Dagbui, Locatelli & Sing (2019), Make-or-break during production: shedding light on change-orders, rework and contractors margin in construction
  abstract; results p.11 of the accepted manuscript · source record

rework/rework_contractor_bears_it · AU · F41, F42 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 23.4% of original contract value, cost of design errors on the civil engineering projects in the sample, as estimated by survey respondents.

Sample. The civil engineering subgroup of the 139 Australian projects in design_error_cost_au. The abstract does not give how many of the 139 were civil. Fit-out projects averaged 22.50% and projects above A$101 million 26.18%.

What it is not. Graded low because the subgroup size is not in the abstract, and the 2012 abstract on what appears to be the same sample says cost did not vary significantly by project type. Respondents' estimates, not cost records, and the carrier of the cost is not stated. Abstract only; full text not read. Whether the sampled civil projects were road, rail or utilities is not stated.

Where it came from

• Love, Lopez, Kim & Kim (2014), Probabilistic assessment of design error costs
  abstract, as held in the Curtin University repository record (read through its mirror at cendekia.unisza.edu.my/neuaxis-e/Record/curtin-20.500.11937-10591/Details); the ASCE page was not opened and the full text was not read · source record

rework/design_error_cost_civil_au · AU · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. Per defect, rectification cost in an apartment building; owners' own costs scaled to total cost to all parties by the CIE; excludes owners' time and other costs.

Sample. Apartment owners' personal rectification cost averaged $8,494; the CIE multiplied it by the detached-house ratio of total cost to personal cost, about 1.10, to give $9,349 (Table 2.17 and the text above it). The report's total cost per apartment defect, including owners' time and other costs, is $12,238. Waterproofing in apartments averaged $19,648 a defect and flammable cladding $34,375, the latter set from one VCAT judgment on the Lacrosse Tower.

What it is not. Graded low because the CIE did not use apartment owners' reported total cost: respondents confused the cost to the whole building with their own unit's, so it scaled personal cost by the detached-house ratio. Owner side, not the contractor's loss, and residential apartments rather than civil works. Owners' estimates, covering every defect reported, including some weighted as caused by maintenance or alteration. The report does not state the price year; 2021 is the report's date.

Where it came from

• Centre for International Economics (for the Australian Building Codes Board) (2021), Building Confidence Report, A case for intervention
  Section 2.5 (Table 2.5), Tables 2.8 to 2.10, 2.16 and 2.17 (printed pages 24 to 36), executive summary Table 1 · source record

rework/defect_rectification_cost_apartment_au · AU · F41 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None - Australian sources measure material waste in tonnes and name over-ordering as a practice, but none gives a cost of wrong quantity, wrong type, or wrong sequence.

Sample. Searches: "Australian construction material waste over-ordering percent of material purchased cost study", "Australian residential construction site material wastage percent over-ordering study", "National Waste Report 2022 construction and demolition waste", and the Love-group rework searches listed under the scope_change record, read for procurement and sequencing content. Frequencies that exist but are not a cost: wasted material averaged 141 kg per square metre, 22% of material weight, on the GBCA sample; the GBCA methodology note says industry practice is a 10% over-order, with no source; the national construction and demolition diversion rate is about 64%.

What it is not. Recorded as none on purpose. GBCA prices waste at up to about $384 per square metre, but by assuming wasted material costs the same as used material and counting all waste, not only over-ordering, wrong type or rework. That is a valuation of waste, not a cost of this cause. The 10% over-order is an unsourced statement of practice. A vendor blog seen only in search results claims 10 to 15% waste on purchased material; it was not opened and cites no study. No source read measures temporary-works failures or sequencing error cost.

Where it came from

• Green Building Council of Australia (2025), Australia's Waste[d] Opportunity 2025
  pages 4 and 5 (cost of wasted materials per square metre and per apartment), and the methodology note on over-ordering · source record

Considered and set aside

• Ratnasabapathy, Perera, Alashwal & Lord: 10 Australian residential projects. Waste quantities and diversion rates, no cost and no cause.
• Ratnasabapathy, Alashwal & Perera: National diversion rate of 64% and 12 residential projects. Quantities only.
• Love, Matthews & Mahamivanan: Names processes and methods not followed and defective materials as root causes of non-conformances on infrastructure, with a mean rework cost of 0.30% of contract value, but the abstract does not name the country and does not split cost by cause.
• Before You Dig Australia (prepared by Clear Ridge Consulting): Mentions resumption costs when work stops at an unexpected asset, in words only, with no figure for them.
• Love, Ika, Ahiaga-Dagbui, Locatelli & Sing: Recorded rework of 0.39% of contract value at one Australian contractor, reported as a total per project with no breakdown by cause in the manuscript.
• Love & Edwards: Direct 6.4% and indirect 5.9% rework from 161 Australian projects; no split by cause.
• Love & Li: Two Australian projects, rework 3.15% and 2.40% of contract value; names client change and contract documentation as the main causes, not procurement or sequencing.
• Love: 161 projects. Finds rework did not vary with procurement method, which is the contract route, not wrong or late buying.
• Centre for International Economics (for the Australian Building Codes Board): Defect cost per defect in housing; attributes defects to initial build or to maintenance, with no split for procurement.

rework/planning_procurement_rework_none_au · AU · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None - Australian studies name client and end-user change as a main cause of rework and measure change-order value, but none gives the cost of the work lost.

Sample. Searches: "Love Edwards rework Australian construction design change causes percentage of contract value", "rework client-initiated changes cost percentage contract value Australia questionnaire change orders", "Make-or-break during production change-orders rework contractors margin", "Infrastructure Australia OR Australian Constructors Association cost of rework design changes major projects", "Australia latent conditions variation cost percent of contract value", "Love 2002 influence of project type and procurement method on rework costs", "Off the rails cost performance of infrastructure rail projects scope changes". Infrastructure Australia and Productivity Commission material turned up in searches only and gave no cost of lost work. Frequencies and values that exist but are not a cost of lost work: change-orders averaged 10.6% of original contract value on 67 Australian public infrastructure projects of one contractor (11% of work value 2011 to 2014); in the 98-project sample, where change-order data existed for 67 projects, they were 3.6% of awarded value. Those are the value of additional or altered work the client pays for, not rework.

What it is not. Recorded as none on purpose. A change-order is usually paid work and mostly revenue to the contractor, so its share of contract value is not a loss, and no Australian source read gives the cost of work done and then lost to a change. Every peer-reviewed Australian rework source found comes from one research group and mostly from perception surveys, or from one contractor's records.

Where it came from

• Love, Irani, Smith, Regan & Liu (2017), Cost Performance of Public Infrastructure Projects: The Nemesis and Nirvana of Change-Orders · free copy
  Table 2; accepted manuscript, Change-Orders section · source record

Considered and set aside

• Love, Ika, Ahiaga-Dagbui, Locatelli & Sing: Accepted manuscript opened and searched for change and cause content. 98 projects of one contractor. A footnote records that data on rework arising from change-orders and design omissions was not made available, so it cannot say what change cost in lost work.
• Love & Li: Two Australian projects with rework at 3.15% and 2.40% of contract value. Names client and end-user changes as a main cause but gives no cost for it.
• Love & Sohal: The same two projects, same result: cause named, no cost per cause.
• Love & Sohal: 161 Australian projects. Client-initiated change is a significant regression predictor of total rework cost, with no money figure for it.
• Love & Edwards: Same 161-project survey, same predictors, no share of rework or of contract value attributed to change.
• Love: 161 projects. Reports rework as 52% of cost growth and 26% of its variance, shares of cost growth rather than of contract value, and not split by cause.
• Love, Edwards & Smith: Looks at contract documentation, not change, and finds no significant link with rework.
• Love & Edwards: 161 projects, direct 6.4% and indirect 5.9% of contract value in total; not split by cause.
• Australian Constructors Association and Consult Australia: Says late design changes drive cost and time pressure, with no figure or data source.

rework/scope_change_rework_none_au · AU · F42 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. Per reported utility strike, direct cost of repairing the damaged asset only; indirect and social costs excluded.

Sample. A direct cost per strike that the industry body advised as conservative and the report used to price about 15,276 strikes reported a year by the 26 largest network operators ($137.5 million a year). The 16 case studies it examined (four each in gas, energy, water and telecommunications) averaged $26,732 in direct cost, which the report says is likely higher than the average strike. Including indirect and social costs, the report puts the total at $4.6 billion a year, about 32.4 times the direct cost.

What it is not. An advised planning figure used as a model input, not an observed mean. The cost is the asset owner's repair cost: the utility bears it unless it recovers it from whoever caused the damage, so it is the contractor's exposure only where that recovery is made, and it is not the contractor's loss. It excludes the contractor's own standing-time and reinstatement costs, delay, and the societal cost of the outage. Strikes by pipe, cable or sector differ widely (telecommunications repairs in the case studies were the largest). The report gives no price year for the $9,000 (it escalates its case-study costs to 2024), so 2024 is the report's date. The strike count covers reported strikes only, which the report thinks is two to three times fewer than the strikes that happen. The report's own basis for the $9,000 is "BYDA data and experience" (Appendix C), with no records shown.

Where it came from

• Before You Dig Australia (prepared by Clear Ridge Consulting) (2024), Economic Assessment of Utility Strikes in Australia
  Section 3.3 (Table 1, 16 case studies) and Section 3.4 (Table 3 and the extrapolation), report pages 5 to 8, and Appendix C for the basis of the $9,000 · source record

rework/utility_strike_direct_repair_cost_au · AU · ALL ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None: New Zealand sources give how often documents have gaps and how often design is blamed for defects, not what it costs.

Sample. Search terms: New Zealand construction design errors documentation cost percent of contract value; BRANZ consent documentation quality; design defects rework cost New Zealand; New Zealand design change cost estimators; cost of defects causes design workmanship. A frequency exists but is not a cost: BRANZ inspectors found that in 43% of 52 sets of new-house consent drawings not all details were provided and in 31% uplift fixing details were missing (SR355, section 5); and ER49 Table 2 (from BRANZ inspections) attributes 12% of defects found to poor or incomplete design against 52% to poor workmanship.

What it is not. Recorded as none on purpose: the New Zealand sources read count documents with gaps and defects by cause, and the one New Zealand cost model prices all defects together. No figure is inferred from the cause split.

Where it came from

• BRANZ (Ian Page) (2016), Consent Documentation Quality for New Housing (BRANZ Study Report SR355)
  Abstract; section 5 Discussion · source record

Considered and set aside

• NZ Institute of Economic Research (NZIER), for BRANZ: Prices all quality defects in new houses together (5.4% of output, modelled). Table 2 attributes 12% of defects found in inspections to poor or incomplete design, which is a share of defects, not of cost.
• BRANZ: Summary of the same model; lists poor design as one direct cause of defects and gives no cost for it.
• PricewaterhouseCoopers, for the Department of Building and Housing: Prices the leaky-building bill (NZ$11.3 billion, 2008 dollars) as a whole and does not split design from workmanship or materials; design fees appear only as a claimant transaction cost ($15,000 for a full reclad).
• Atapattu, Domingo & Sutrisna: Estimators name inadequate tender documentation and frequent design changes among the three most critical cost factors; a perception ranking with no cost magnitude in the abstract, and the full text was not available.
• Massey University (Asadi, Wilkinson, Rotimi): Rates how well the standard contract addresses rework causes, with errors in design, drawings and specifications and incomplete design among five scored adequate (Table 3, means 3.50 and 3.39 on a 5-point scale); contract coverage, not cost. The 25% of contract value rework figure it quotes comes from earlier international literature.
• New Zealand Infrastructure Commission / Te Waihanga: Notes that policy-driven added scope and design requirements raise infrastructure cost; compares unit costs and attributes no share to design or documentation.
• New Zealand Infrastructure Commission / Te Waihanga: Names an under-estimated initial cost and governance as contributors to the motorway overrun; no design-error or documentation figure.
• BRANZ (Roman Jaques): Contractors link non-standard design and inaccurate detailing (15%) to material waste; measures waste causes by respondent share, no cost.
• The Conversation (Mark Kirby, Auckland University of Technology): Opinion piece; its nearly-10% loss figure misreads the NZIER GDP effect as a defect cost and gives no design-specific number.

rework/design_documentation_none_nz · NZ · F41 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None: New Zealand sources measure waste by mass or volume and rank planning as a cost factor, and none prices wrong or surplus orders as a share of project cost.

Sample. Search terms: BRANZ construction waste New Zealand over-ordering materials percent; wrong materials ordered rework New Zealand; temporary works failure cost New Zealand; construction procurement supply chain delay cost percent of project New Zealand; Construction Sector Accord procurement. A frequency exists but is not a cost: AUT's scoping study puts construction and demolition waste at about 40% of New Zealand landfill, with estimates of 30% to 50% by source, and names over-ordering of wet concrete as a source of concrete waste (ER82). A BRANZ and EBOSS supplier survey of freight cost and delays turned up in search but was not opened.

What it is not. Recorded as none on purpose: waste by mass, landfill share and contractors' own waste percentages by material (1998 to 1999) are not a cost as a share of project value, and are not filed as one.

Where it came from

• Auckland University of Technology, for BRANZ (2023), Understanding waste generation in the New Zealand construction sector: Scoping study (BRANZ External Research Report ER82)
  Executive Summary (key issue 1); concrete waste section; supply-chain material · source record

Considered and set aside

• BRANZ (Roman Jaques): Contractors estimate material waste by material (for example plasterboard 5% to 15%) in 1998; 38 Wellington-area respondents, an estimate of quantity wasted, not cost, and dated.
• BRANZ (Roman Jaques): 84 Auckland respondents; non-standard sizing (53%) is the main cause of waste named by contractors; quantities and perceptions, no cost.
• Massey University (Asadi, Wilkinson, Rotimi): Materials and equipment and process causes of rework are among those the standard contract covers; clause coverage, no cost.
• Atapattu, Domingo & Sutrisna: Poor planning and scheduling is among the three most critical cost factors for New Zealand transport projects; a perception ranking with no magnitude (abstract only).
• NZ Institute of Economic Research (NZIER), for BRANZ: Table 2 prints BRANZ inspection results of 12% for materials and 12% for construction methods (builders' opinion 7% and 14%); shares of defects as printed, and the model prices defects as a whole.
• New Zealand Infrastructure Commission / Te Waihanga: Discusses procurement and delivery capability as drivers of unit cost; benchmarking of unit costs, no share attributable to wrong quantities, sequencing or temporary works.
• New Zealand Infrastructure Commission / Te Waihanga: The builder not anticipating the consenting system is named as a contributor; no figure.
• SCIRT Learning Legacy: Describes the Christchurch rebuild procurement model and its $2.2 billion total; no planning, over-order or sequencing cost.

rework/planning_procurement_none_nz · NZ · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 5.4% of the output of the residential construction industry ($995 million of $18.43 billion, central case): the modelled cost of rectifying quality defects found during construction, from a 90% defect prevalence (informed by New Zealand surveys of new houses) and an assumed 6% average cost per defective job (set from overseas studies).

Range 2.6% to 8.6% (scenario)

Sample. A computable general equilibrium study (NZIER, ORANI-NZ) for BRANZ. Inputs: 90% prevalence (85% to 95%), 6% average cost of a defect (3% to 9%), residential output $18.43 billion for 2018. The 90% rests on BRANZ new-house quality surveys (92% of 225 houses in 2014 had a quality or appearance defect; 96% had a compliance defect at the first inspection); the 6% is set from the literature review, which cites Australian and UK studies of rework or defect cost averaging 2.4% to 6.4% of contract value (UK 5.8%; Australia 2.4% and 3.15%, 4.1% with a 3.0% to 11.2% range, and 6.4% direct and 5.6% indirect). Cost $995 million central, $420 million to $1,744 million across the range.

What it is not. A model, not observed rework. Only the prevalence is informed by New Zealand data; the average cost of a defect is an assumption taken from overseas studies, and the report assumes those estimates apply to New Zealand. The 5.4% is prevalence times average cost (0.90 x 6%), applied to all residential construction output, including the residential share (45%) of construction services; the report scopes the study to defects found during construction of new houses but assumes non-new work has the same prevalence. Out of scope by the report's own list: civil and non-residential work, alterations and maintenance, undiscovered defects, leaky buildings and earthquake repair. It is an industry-wide resource cost that the report says falls on builders but is priced into construction costs, so it is not the loss of any one contractor. The denominator is industry output, not a contract value, so it is filed as ratio_of_other and is not multiplied by a job value. The range is the report's low and high scenarios (ER49 Table 7: 2.6%, 5.4%, 8.6%; Appendix Table 13 agrees); ER49's own Table 8 and the BRANZ summary's Table 2 both print the 2.6% and 5.4% in the wrong columns.

Where it came from

• NZ Institute of Economic Research (NZIER), for BRANZ (2020), The Economic Cost of Quality Defects (BRANZ External Research Report ER49)
  Table 7 (p.22), Table 8 and Table 9 (p.24) · source record
• BRANZ (2021), BRANZ Research Now Quality #2: The economic cost of defects in new residential buildings in New Zealand
  Table 2 (p.2): confirms the $995 million central estimate · source record

Considered and set aside

• PricewaterhouseCoopers, for the Department of Building and Housing: Prices the leaky-building repair bill (NZ$11.3 billion in 2008 dollars, 42,000 failing dwellings built 1992 to 2008), which ER49 puts out of scope. The report does not split failures between design, workmanship and materials and 69% of the cost falls on owners, so it is not rework by a builder and is not filed under any cause.
• The Conversation (Mark Kirby, Auckland University of Technology): Presents the $2.5 billion as nearly 10% of sector value lost to quality failure. In the NZIER study $2.5 billion is a GDP effect in the economy-wide model, not a cost of defects, so the 10% is not used.

rework/quality_defects_residential_cost_nz · NZ · F41, F43 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None: New Zealand sources rank change as a cause of rework or cost growth but give no cost, and the one national benchmarking dataset has no New Zealand overrun data.

Sample. Search terms: New Zealand construction variations change orders cost percent of contract value; NZS 3910 rework variation survey; design change cost New Zealand infrastructure; Auditor-General Treasury variations contract value New Zealand; Transmission Gully cost scope change. A frequency or ranking exists but is not a cost: New Zealand estimators name frequent design changes among the three most critical cost factors (Atapattu 2023, abstract only), and Asadi 2023 (80 completed responses) scores changes first among the rework causes the standard contract addresses (Table 3, mean 3.83 of 5), a rating of contract coverage. The Asadi paper quotes rework of up to 25% of contract value from earlier international work.

What it is not. Recorded as none on purpose: no New Zealand source read measures the cost of client or design change, abortive work or change orders that caused rework. Variation values for Australia exist in the corpus but do not apply to New Zealand.

Where it came from

• Massey University (Asadi, Wilkinson, Rotimi) (2023), Mixed-Method Approach to Evaluate Rework Provisions within New Zealand Construction Contracts · free copy
  Abstract; Introduction (rework cost paragraph); Method (81 respondents); Table 3 (causes of rework addressed by the contract) · source record

Considered and set aside

• Atapattu, Domingo & Sutrisna: Estimators name frequent design changes among the three most critical cost factors on New Zealand transport projects; a perception ranking with no cost magnitude in the abstract, full text unavailable.
• New Zealand Infrastructure Commission / Te Waihanga: Says added scope and design requirements raise infrastructure unit costs and that overrun analysis is outside its scope; no figure.
• Oxford Global Projects, for Te Waihanga: Overrun reference classes are international; the report states New Zealand estimated-versus-actual data were lacking, and it does not attribute overruns to scope change.
• New Zealand Infrastructure Commission / Te Waihanga: Summary names estimate accuracy, governance, consenting and external events; it does not quantify or separate scope or design change.
• SCIRT Learning Legacy: Describes the Christchurch rebuild as having an evolving scope; reports the $2.2 billion programme total, no change or rework cost.
• BRANZ (Roman Jaques): Architects and quantity surveyors rate avoiding variations on site as a way to cut waste; 38 respondents, opinions only, no cost.
• BRANZ (Roman Jaques): Repeats that buildable design avoiding variations reduces waste; 84 respondents, opinions only, no cost.
• NZ Institute of Economic Research (NZIER), for BRANZ: Prices defects found during construction as a whole; client or end-user change is not separated and is outside the defect definition.

rework/scope_change_none_nz · NZ · F ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. None: New Zealand sources count utility strikes or describe ground and weather costs qualitatively; none gives a cost per strike or a share of contract value.

Sample. Search terms: Before U Dig New Zealand utility strikes annual cost per strike; NZ Utilities Advisory Group strikes; Waka Kotahi utility strike research; SCIRT service strikes cost; New Zealand weather delays construction cost; Transmission Gully ground conditions cost; setting-out survey error cost New Zealand. A frequency exists but is not a cost: in a 2020 survey, 34 respondents reported 60 utility damage incidents in 2019 tied to missing or inaccurate as-built drawings and 50 for other reasons (Maree 2021, section 4.3); the same paper fits a line to those counts and says about 8 to 9 strikes per 100 excavations, a frequency from a 34-respondent sample, not a cost. Vendor blogs quote a New Zealand strike count (12,861 in 2020) and per-strike costs built on a UK multiplier; they were not opened at source and are not used. The Auditor-General's 2016 SCIRT report and MBIE weathertightness statistics were blocked to automated access and not read.

What it is not. Recorded as none on purpose: the New Zealand strike statistics found are counts of incidents or enquiries. Australian damage costs per strike in the corpus do not apply to New Zealand.

Where it came from

• Massey University (Maree, Rotimi, Rotimi) (2021), The Primacy of As-Built Drawings in the Management of Underground Utility Operations: A New Zealand Study
  Section 3 (method), section 4.3 (utilities damage records), Table 1 · source record

Considered and set aside

• Before U Dig (BUD): Reports 186,614 enquiries lodged in 2022 and 186 asset owners; publishes no national strike counts or costs.
• NZ Transport Agency, via the SCIRT Learning Legacy: Says tenders will ask for service-strike performance and that damage must be monitored; no strike counts or costs.
• SCIRT Learning Legacy: Christchurch rebuild summary with the $2.2 billion programme total; no strike, ground or weather cost.
• New Zealand Infrastructure Commission / Te Waihanga: States tunnelling in difficult ground can cost up to three times as much as in favourable ground; an international tunnelling comparison, not a New Zealand share of contract value and not about ground differing from the record.
• New Zealand Infrastructure Commission / Te Waihanga: Lists external events (earthquake, weather, COVID-19) among contributors to the overrun; no figure and no separation of ground conditions.
• Oxford Global Projects, for Te Waihanga: Says New Zealand estimated-versus-actual cost data were lacking, so no New Zealand overrun benchmark or cause split exists in it.
• PricewaterhouseCoopers, for the Department of Building and Housing: Prices the leaky-building repair bill (NZ$11.3 billion, 2008 dollars); those are building defects, not weather or ground conditions, and the report does not split them by cause.
• NZ Institute of Economic Research (NZIER), for BRANZ: Prices construction defects in new houses; setting-out error and weather damage are not separated.

rework/site_conditions_none_nz · NZ · F42 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 5% of total construction cost, direct rework only.

Sample. The paper analyses 359 projects in the CII database, mostly US industrial and commercial.

What it is not. The 5% is the abstract's opening framing, not the paper's own measured result, and who bears the cost is not measured.

Where it came from

• Hwang, Thomas, Haas & Caldas (2009), Measuring the Impact of Rework on Construction Cost Performance
  abstract, read on TRID: https://trid.trb.org/view/884832 · source record

rework/rework_field_direct_framing · US · F41, F42 ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page

What it is. 0.38% of contract value, direct cost only, as recorded pre-completion.

Sample. One contractor's project data, collected by its quality managers. Adding post-completion corrections raises the average to 0.76%.

What it is not. SECONDARY - the paper itself returned 403 and was read through ASCE's news write-up. The article gives no sample size, country or sector. The same study found actual costs underreported by around 300%, which is the number that matters here: most rework is never written down as rework.

Where it came from

• Love (2026), Quantifying the Costs of Field Rework in Construction second-hand
  abstract, as reported · source record
  Read via: ASCE Civil Engineering Source, 22 Jan 2026: https://www.asce.org/publications-and-news/civil-engineering-source/article/2026/01/22/how-much-does-field-rework-in-construction-actually-cost (the paper itself returned 403)

rework/rework_recorded_in_ledgers · unknown · unknown ·the record ·link to this figure ·dispute this figure(or email support@demiton.io)

Open this figure's own page