TFP and labour productivity
Index, base 100 at Q3 2023. The dashed line is what an annual, hours-only report would show.
Growth accounting
Quarterly output growth split into each input's contribution and the TFP residual. Dots mark total output growth.
Effective use of each input
Share of paid or installed input actually productive.
Input indices
Growth of each input, base 100.
Device status
Simulated IoT devices on site, refreshed every few seconds.
| Device | Type | Input | Status | Reading |
|---|
Event stream
Newest readings first.
Raw records
Unprocessed sensor events, documents and financial entries before they feed the KLEMS indices.
How each input is captured
Each input needs a quantity and a price. Sensors give quantity fast; documents and accounts give price and reconciliation.
| Input | Sensors and live feeds | Documents and records | Financial and historical | Refresh |
|---|---|---|---|---|
| Capital | Telematics, engine-hour meters, IoT on formwork | Equipment logs, hire agreements, work orders | Asset register, depreciation, hire invoices | Live |
| Labour | Geofenced wearables, turnstiles, camera counts | Timesheets, permits, skill certificates | Payroll, overtime, past output per trade | Live |
| Energy | Smart meters, diesel tank sensors | Utility bills, fuel delivery notes | Energy ledger, tariff history | Live |
| Materials | RFID or QR gate scans, load cells, batch-plant telemetry | Delivery orders, bill of quantities, waste reports | Purchase ledger, supplier price indices | Daily |
| Services | Subcontractor check-ins, drone or 360 capture | Subcontracts, progress claims, variations | Accounts payable, retention history | Weekly |
| Output | As-built quantities from scans and BIM | Valuations, certified claims | Revenue, deflated by a construction price index | Monthly |
Lifecycle stage
Simulated TFP growth by stage for the selected project (last four quarters for Build). Pick a stage to filter the ranking table.
Layer 1: is the project ready to perform?
CIS 30, SCORE and CDMI as in the Blueprint. The CIS 30 benchmark of 2.00 m²/man-day is an assumption for this simulation.
KLEMS spider-web
Score = actual ÷ benchmark × 100, inverted for energy and waste where lower is better.
Early warning
Simulated rules on the latest quarter. Thresholds are examples to tune.
Two ways to measure real output
The Blueprint prices measured quantities at base-period rates. The dashboard so far used certified value divided by a price index. Running both gives a cross-check.
Project and contractor ranking
Simulated portfolio ranked by TFP growth over four quarters. Rank is across all stages.
Blueprint calculation (from main slides)
Apartment pilot: contract sum RM314.5m, inputs RM30m, 40m, 10m, 90m and 30m, output +9%.
Materials growth %
Two estimation methods
| Growth accounting | Production function | |
|---|---|---|
| Weights | Observed cost shares | Coefficients estimated from data |
| Data needed | One project, two periods | Many projects or firms over time |
| Output | TFP as one residual | Splits TFP into technological change and technical efficiency |
| Best for | Project tracking and early warning | Sector benchmarking and policy once the pilot database exists |
1Collect the raw data
Scenario: a 12-storey residential block, comparing two quarters. Every box can be edited. Pick a preset to see how a different situation changes the result.
| What you collect | Q2 2026 | Q3 2026 | Unit |
|---|
Q3 cost of the inputs, used for the weights (materials and services use their spend above): Capital RM m Labour RM m Energy RM m
Words you will meet
Price index (deflator): a number that tracks prices over time. Dividing RM figures by it removes price changes, so you compare real work, not inflation.
ln: the natural log button on a calculator. ln(Q3 ÷ Q2) × 100 gives a growth rate that is close to the simple percent change for small moves and adds up cleanly across inputs.
Cost share: how much of total input cost each input takes. Bigger inputs count for more.
Residual: what is left after you subtract everything you can explain. That is TFP.
What the new approach changes
TFP is output growth that inputs do not explain. Better input measurement shrinks the unexplained part and makes what remains more trustworthy.
| Aspect | Traditional | Sensor-and-records KLEMS |
|---|---|---|
| Inputs counted | Mostly labour hours or cost | Five inputs, quantity and price each |
| Capital | Book value of assets | Capital services from metered hours and utilisation |
| Labour | Headcount or paid hours | Present hours adjusted for skill mix and idle time |
| Materials and services | Spend deflated by a general index | Installed quantities at actual unit prices |
| Reporting lag | 12 months or more | 1 to 7 days, quarterly TFP |
| Detail | Company or national | Project, trade and work package |
1. How the percentages are calculated
Growth accounting splits output growth into what the five inputs explain and what is left over. The leftover is TFP (the Solow residual).
The headline tiles come from these: TFP growth, last 4 quarters is the sum of the four quarterly TFP growths. Output growth that is true efficiency gain is that sum divided by the sum of four quarters of gY. Published statistics often use the Tornqvist form, which averages the cost shares of the two periods. This console uses fixed shares to stay simple.
2. How each quantity is measured
| Input | Quantity Xi and cost |
|---|---|
| Capital | Machine-hours by asset type, weighted by rental price per hour. Cost = rental or imputed rent. |
| Labour | Hours worked by skill group, weighted by wage. Cost = payroll. |
| Energy | kWh and litres of fuel. Cost = utility and fuel bills. |
| Materials | Delivered or installed quantities, or purchases deflated by a price index. |
| Services | Subcontract spend deflated by a services price index, or certified work volume. |
Output Y is gross output: the certified value of work delivered, deflated by a construction price index. Because energy, materials and services are inputs, gross output is the right measure.
3. Simple worked example
Round numbers for one quarter.
| Item | Growth g | Share s | s × g |
|---|---|---|---|
| Capital | 1.0% | 14% | 0.14 |
| Labour | 1.5% | 30% | 0.45 |
| Energy | 1.0% | 4% | 0.04 |
| Materials | 2.0% | 36% | 0.72 |
| Services | 2.0% | 16% | 0.32 |
| Weighted inputs | 1.67 | ||
| Output | 3.00% | 3.00 | |
| TFP | 1.33 |
TFP = 3.00 − 1.67 = 1.33 points, so 44% of output growth is efficiency. Labour productivity growth is 3.0 − 1.5 = 1.5%, which would credit labour with 1.5 points when only 1.33 is true TFP.
4. Live example from this dashboard
5. Case studies and published KLEMS work
Real studies that apply the same framework to construction. The figures on this dashboard are simulated and are not drawn from them.
| Study | Scope | What it found |
|---|---|---|
| Abdel-Wahab and Vogl, Construction Management and Economics | Construction in Europe, the US and Japan, 1971 to 2005, using EU KLEMS | Labour productivity growth slowed in most countries. The gap between construction and all industries is largely explained by construction's weak TFP. |
| Ruddock and Ruddock (2011) | UK construction, EU KLEMS growth and productivity accounts | Rising ICT capital and better use of labour skills, with a relatively high TFP rate for the UK. |
| Sweden versus Europe (KTH) | Eight European countries, TFP approach with KLEMS data | Tests whether Swedish construction really lags by comparing total factor, capital and labour productivity across countries. |
| Ireland CSO, Productivity in Ireland | KLEMS decomposition of gross output, Irish construction | In 2023 construction gross output fell 4.9%, mainly from materials (-2.7) and services (-1.7). Across 2011 to 2020, KLEMS MFP explained relatively little of growth, which came mostly from materials and services. |
| India KLEMS database (RBI) | 27 industries including construction, 1980 to 2011 | Shows how a national KLEMS database is built so results are comparable with EU KLEMS and World KLEMS. |
| Belgian regions growth accounting | Construction in three Belgian regions, 1980 to 2006 | Applies growth accounting to regional construction and notes that TFP also absorbs measurement error in inputs. |
What to take from them. First, national studies work at industry level and years apart, which is the gap project-level sensor capture fills. Second, TFP is a residual and picks up input measurement error, so better capture of machine hours, labour time and installed quantities makes it more reliable. Third, KLEMS on gross output gives a smaller MFP than a value-added measure, so compare like with like.
Where the output figure comes from
Output Y is the real (price-adjusted) value of work delivered in the quarter. It is built from project documents, then cross-checked against the accounts and site measurements.
| Source | Field to extract | Role |
|---|---|---|
| Interim payment certificates and progress claims (quantity surveyor) | Gross value of work done this period, including approved variation orders, before retention and advance recovery | Primary nominal output |
| Contract WIP schedule and income statement | Revenue recognised for the period (percentage of completion) | Reconciles to the certificates |
| BIM progress model, laser scans, drone survey | Installed quantities multiplied by bill-of-quantities unit rates | Independent check on claimed progress |
| Construction price index (national statistics or a project rate index) | Index level for the quarter, same base year for every quarter | Converts nominal RM to real output |
From raw records to percentage and index
One project, two quarters (Q2 and Q3 2026). Illustrative figures. Every step runs left to right: raw record, real quantity, growth, weighted contribution.
| Item | Raw data and where it is extracted | Raw Q2 | Raw Q3 | Real Q2 | Real Q3 | Growth g | Cost share s | s × g |
|---|---|---|---|---|---|---|---|---|
| Output | Certified gross value of work, from payment certificates. Divided by construction price index 112.0 (Q2) and 113.4 (Q3), base Q3 2023 = 100 | RM 18.40m | RM 19.52m | 16.43 | 17.21 | 4.667% | - | - |
| Capital | Engine-hours exported from the equipment telematics platform, summed over the fleet | 3,820 h | 3,930 h | 3,820 | 3,930 | 2.839% | 13.9% (RM 2.34m) | 0.394 |
| Labour | Gate and wearable logs reconciled to timesheets, hours weighted by skill group | 41,200 h | 42,150 h | 41,200 | 42,150 | 2.280% | 30.1% (RM 5.07m) | 0.686 |
| Energy | Smart-meter kWh and diesel litres from tank sensors and bills, converted to gigajoules | 1,420 GJ | 1,505 GJ | 1,420 | 1,505 | 5.814% | 4.0% (RM 0.68m) | 0.235 |
| Materials | Purchase ledger matched to delivery orders. Divided by materials price index 108.0 and 109.5 | RM 5.78m | RM 6.05m | 5.35 | 5.53 | 3.186% | 35.9% (RM 6.05m) | 1.145 |
| Services | Subcontractor invoices matched to certified claims. Divided by services price index 111.0 and 112.2 | RM 2.55m | RM 2.70m | 2.30 | 2.41 | 4.641% | 16.0% (RM 2.70m) | 0.744 |
| Weighted inputs (sum of s × g) | 3.204 | |||||||
| TFP growth = 4.667 − 3.204 | 1.463% | |||||||
The steps behind each cell
Output: 19.52 ÷ 1.134 = 17.21 and 18.40 ÷ 1.120 = 16.43, so gY = ln(17.21 ÷ 16.43) × 100 = 4.667%. Total input cost for the shares is 2.34 + 5.07 + 0.68 + 6.05 + 2.70 = RM 16.84m. Overhead and margin are left out of the shares.
Percentage and index
TFP growth 1.463% means output grew 1.463 points faster than the five inputs together explain.
TFP index = 118.2 × (1 + 1.463 ÷ 100) = 119.9, taking Q2 at 118.2.
Labour productivity growth = 4.667 − 2.280 = 2.387%. Labour looks 2.387 points better, but only 1.463 points is TFP.
Efficiency share = 1.463 ÷ 4.667 = 31% of output growth. These feed the Overview tiles and the Raw data quarterly dataset.
How the quantity surveyor values the work
The certified figure is measured progress priced at contract rates. It is an interim valuation, usually monthly.
1. Start from the contract price breakdown: a bill of quantities (item, quantity, rate), a schedule of rates, or milestones for lump sums.
2. Measure progress on site from drawings, inspections and the site walk, as a quantity or a percentage of each item.
3. Price it: completed quantity × contract rate.
4. Add approved variations and day-works at agreed rates, and any price adjustment or materials on site the contract allows.
5. Period value = cumulative value to date − cumulative value in the previous certificate.
6. The contract administrator certifies it. Retention and advances come off afterwards, so the gross valuation is the output figure.
| Item | Basis | Value |
|---|---|---|
| Concrete slab | 800 of 1,200 m³ × RM 450 | RM 360,000 |
| Roofing (lump sum) | 40% of RM 2,000,000 | RM 800,000 |
| Approved variation | Complete | RM 20,000 |
Watch for: percentage completion on lump sums is a judgement, so cross-check with scans or BIM. Front-loaded rates overstate early progress. Unfixed materials can be double counted with the materials input, so pick one treatment. Variations raise value without raising efficiency, so track them separately. Cost-plus and design-and-build contracts value progress by milestone or actual cost.
Checks before the numbers are trusted
Use the same price index base for every quarter. Match claims to the period the work was done, not the month they were certified. Compare certified value with installed quantities times unit rates and investigate gaps above a few percent. Make sure subcontract spend is not also counted in labour or materials. Flag records with estimated or missing sensor data (see the quality column in Raw data) before they enter the calculation.