Statistical Confidence & Hourly Rate — Valuation Explainer
Why does the statistical confidence gauge remain steady while the likely range shifts when you customize the hourly rate? Here is the exact mathematical and accounting rationale.
The core question: Why doesn't the percentage change?
When viewing the Valuation tab of an IFC report, users often customize the blended hourly rate (defaulting to €85/hr, or converted into local currencies such as USD, GBP, CAD, or AUD) to match their corporate charge-out rates or regional engineering frameworks.
A natural question arises: "If I enter a custom hourly rate, shouldn't the statistical confidence of the valuation change?"
The short answer is no: the confidence percentage measures the physical modeling effort (hours), not the commercial currency rate applied to those hours. However, the Likely range displayed inside the confidence box (e.g. €21,000 – €44,000) does scale dynamically with your rate.
Two halves of the valuation formula
Under international accounting standards (IAS 38, IFRS 13, and IPSAS 31/46), replacement cost is calculated by multiplying quantitative labor effort by a supportable unit rate:
🔒 What stays constant: Confidence %
The dial gauge and percentage (e.g., 70% · High) reflect how well the IFC model fits our empirical effort benchmark. Entering €120/hr instead of €85/hr doesn't make the building model take fewer or more hours to draw.
⚡ What updates live: Likely Range
The monetary spread inside the box (Likely range €... – €...) scales directly: [hours_low × rate, hours_high × rate]. Your financial range shifts immediately with your rate.
How the 70% statistical confidence is derived
ForensicBIM's authoring effort heuristic is not a rule-of-thumb guess; it is calibrated against an empirical benchmark:
- 121 blind estimates from 7 independent BIM experts across architectural, structural, and MEP disciplines.
- 43 diverse calibration models: ranging from single-trade precast elements to complex hospitals, residential towers, and bridge infrastructure.
- Leave-one-out cross-validation: Candidate regression formulations were evaluated by predicting each model with a fit calibrated strictly on the other 42 models it had never seen before.
Out of sample, 70% (30 out of 43) of expert consensus totals fell inside the reported effort range (defined as $total / 1.45$ to $total \times 1.45$). This is why the baseline statistical confidence is set at 70%.
When does the confidence percentage decrease?
The confidence gauge drops only when the IFC file possesses technical characteristics outside the calibrated sweet spot:
| Diagnostic Condition | Points Off | Engineering Reason |
|---|---|---|
| Extrapolated model | -25% | The model exceeds 10,000 distinct shapes, 40,000 elements, or 350,000 m² (larger than any calibration model). |
| Infrastructure facility in file | -15% | Bridges and civil alignments have distinct curve-geometry authoring dynamics not covered in building rounds. |
| Installations (MEP) model | -10% | Experts exhibit the highest variance and disagreement on mechanical/electrical/plumbing routing. |
| No triangle count (geometric detail) | -10% | The file contains boundary representations without explicit mesh surface counts. |
| No relation incidences | -10% | Missing containment or aggregation trees make coordination effort harder to estimate. |
| No measured surface area | -5% | Surface area had to be estimated using class volume heuristics. |
| Estimated from percentage | -5% | A portion of property set coverage had to be statistically interpolated. |
Economic certainty vs. authoring effort certainty
From a financial appraisal perspective, decoupling the labor quantity from the labor rate is a deliberate accounting principle:
1. User-customized rates have lower rate variance
When you enter your own verified contractual rate—for example, €95/hr from a specific architectural agreement or internal payroll ledger—the rate variance for your organization is actually zero. You know exactly what an hour of BIM specialist time costs you. Penalizing the valuation confidence because you supplied an accurate rate would be statistically wrong.
2. Market benchmark rates carry market variance
When using the default €85/hr rate, the estimate reflects a European median blended rate (accounting for junior modellers, senior coordinators, software seat amortisation, and office overhead). While regional market conditions vary, the uncertainty of what BIM labor costs in a city is separate from how many hours the file requires to rebuild.
3. Auditability under IAS 38 and IFRS 13
When recording a BIM model on the balance sheet as an intangible asset (IAS 38) or reporting fair value (IFRS 13), financial auditors demand traceability:
- Effort justification: The auditor verifies the engineering hours with the ForensicBIM audit report and statistical confidence score.
- Rate justification: The auditor verifies the hourly rate against the company's actual payroll, contractor invoices, or framework agreements.
Keeping the two components clearly separated prevents conflating engineering uncertainty with procurement terms.
How can you improve the statistical confidence score?
If your report shows a Medium or Low confidence score (below 65%), you cannot raise it by tweaking the hourly rate. Instead, improve the quality and completeness of the IFC file:
- Ensure complete spatial hierarchy: Verify that elements are assigned to storeys (
IfcRelContainedInSpatialStructure) and systems (IfcRelAssignsToGroup). - Export explicit geometric representations: Ensure your authoring tool exports faceted Breps or tessellated geometries with verifiable face/triangle counts.
- Include measured quantities: Export standard quantity sets (such as
Qto_WallBaseQuantities) with explicit gross and net surface areas. - Segment massive federations: If the model is marked as extrapolated (over 40,000 objects), export individual discipline models or building phases rather than an entire city block in one container.
Related guides and methodology
- What's in the €85 Hourly Rate? — Full breakdown of direct labor, employer taxes, software subscriptions, and corporate overhead.
- Valuation Methodology Whitepaper — In-depth explanation of the 121 expert estimates, 43 calibration models, and the 11 valuation rungs.
- General Quality Audit Methodology — Standards compliance overview across buildingSMART, IFRS 13, and ISO 19650.