Yongjian Ke

Digitising construction: policy, capability and the building site

· 5 min

Construction is among the least digitised industries in the world, and governments have spent a decade trying to change that. Over several years I have worked with Prof Jingxiao Zhang and colleagues at Chang'an University on four studies that approach the problem from different heights. One asks what governments are actually offering. One asks what determines whether firms respond. Two show what digitisation looks like by the time it reaches a road under construction. Read together they describe a chain, and they also show where the chain has not yet been joined up.

What governments are actually offering

The first study examined 82 Chinese digital transformation policies issued between 2017 and 2022, sorting their contents into three families of instrument: supply-side measures such as technical support, funding and training; environment-side measures such as planning targets, tax relief and credit; and demand-side measures such as public procurement, pilot projects and promotion (Zhang, Chen, Ballesteros-Pérez, Ke, Gong, & Ni, 2023).

The mix was uneven. Roughly 40 per cent of instrument use was supply-side and 36 per cent environment-side, leaving around 24 per cent on the demand side. Government procurement was the least-used instrument of the twelve examined. Scoring the quality of 25 representative policies produced a respectable average, but the same weaknesses kept recurring: short validity periods, vague statements of what the industry actually needs, and a narrow band of incentives that leaned on pilots and talent programs while rarely reaching tax relief or subsidy. Policy quality also declined from the eastern provinces westward, tracking economic development.

The practical implication is straightforward. A government that funds technology but does not buy digitally delivered work leaves firms carrying the commercial risk of adopting it on their own.

Why firms adopt, or do not

The second study asked what determines whether firms move at all. It used topic modelling over the research literature to surface candidate factors, then had experts weigh how those factors influence one another (Li, Han, Zhang, Philbin, Liu, & Ke, 2022). The factors sorted into five groups: enterprise resources, enterprise capability, enterprise spirit, macro environment and industry environment.

Enterprise capability dominated, carrying roughly twice the weight of the macro environment, and within it, the capacity for technological innovation was the single largest factor. The policy environment mattered, but it ranked third.

Subsidies can buy equipment. They cannot buy the capability to use it well.

Set beside the first study, that produces an awkward pairing. The instrument governments reach for most often is supply-side support, which acts on resources. The binding constraint sits in capability, which is slower to build and much harder for a subsidy to create. This finding rests on expert judgement over a modest sample of literature, so it is better read as a considered ordering than as a measurement, but the direction is worth taking seriously.

What it looks like on a road

The remaining two studies start from interviews with practitioners, 22 in one and 25 in the other, and both surfaced the same complaint. Project data is not missing. It is stranded. A road project generates survey control points, design drawings, asphalt production records, haulage routes, paving temperatures, compaction passes and laboratory tests, and each set sits with whoever collected it. Nobody can connect a batch of asphalt to the truck that carried it, to the stretch of road it was laid on, to the test result for that stretch.

The response in both cases was a single platform rather than another isolated tool. A BIM model provides the spine, BeiDou positioning tracks pavers, rollers and trucks, sensors capture temperature and vibration, and 5G moves the data as it is produced. One study covers the whole process across measurement, design, construction and inspection, demonstrated on Cambodia's Phnom Penh–Sihanoukville Expressway, the first expressway in that country (Chen, Zhang, Ni, Skitmore, Ballesteros-Pérez, Ke, Zuo, & Sun, 2022). The other goes deeper into the pavement itself, pairing a data management system with five monitoring subsystems for production, mixing, transport, paving and rolling (Zhang, Zhu, Liu, Zuo, Ke, Philbin, Zhou, Feng, & Ni, 2023).

What this buys is traceability and early warning: an alarm when a mix temperature or compaction reading falls outside tolerance, while the material is still workable, rather than a finding weeks later when the only remedy is rework. The pavement study also reports a construction cycle around 30 per cent shorter and substantial cost savings. Those are self-reported figures from a single project with no comparison site, so I would read them as a description of what the system does rather than as a measured effect.

The chain, and where it breaks

Taken in order, the four studies trace a path. Policy is skewed towards supply. What actually determines uptake is firm capability. Where capability exists, the result is a delivery system built around data rather than paper.

There is a return path as well, and it is the more interesting one. The platforms generate exactly the implementation evidence the policy study admits it lacks. Evaluating policy documents tells you what governments intended; it says nothing about what changed on site. Project data of this kind could close that loop, and testing whether particular instruments shift delivery performance is the obvious next study.

A second thread runs through the set without being examined. The policy and capability work is entirely about China. The working systems were demonstrated on a Belt and Road project in Cambodia. Capability formed under one set of institutional conditions is being exported to another with different regulation, labour markets and client expectations. What survives that transfer, and what quietly does not, is a question none of the four papers takes up.

Journal papers discussed on this page