China Will Count Data Assets. The First Growth Jump Needs a Bridge Series.

China's National Bureau of Statistics plans to trial a digital-economy supply-use table, introduce regular accounting for value added in the digital and platform economies, and test the treatment of data as an asset. The program could expose activity that existing statistics miss. It could also produce a headline increase that cannot be compared with the old series.

The distinction matters. A wider statistical perimeter is not fabricated growth. It is a different measurement contract. Unless the bureau publishes a bridge, the first change in measured output will combine real production, price movements, newly capitalized data, revised classifications and better source data.

The first new figure will be a level estimate, not a clean growth rate, unless the old series is rebuilt on the same perimeter.

Problem: one digital economy, several accounting objects

The existing official baseline is precise but narrow. The NBS revised the value added of China's core digital industries to 12.7555 trillion yuan for 2023, equal to 9.9% of GDP. It used the 2021 Statistical Classification of the Digital Economy and Its Core Industries, covering 126 national-industry subcategories in four broad groups. The Fifth National Economic Census supplied much of the underlying evidence.

That series measures value added produced by resident units classified inside the core digital industries. The new statistical reform plan points to three related but non-identical objects.

The plan itself provides no first-release dates for these outputs. It does not specify sector or product coverage, a revision policy, or whether readers will be able to reconstruct totals from published components. It also says China will study and implement the 2025 System of National Accounts, but does not set out how the old core-industry series will map into that framework.

Analysis: the perimeter can move GDP before output moves

A digital supply-use table is broader than an industry total. Under the 2025 SNA's digitalization framework, it separates digitally ordered, digitally delivered and platform-mediated transactions. It also distinguishes domestic and imported supply, then traces use through intermediate consumption, final consumption, fixed investment and exports. Its industry columns include digital infrastructure providers, data- and advertising-funded platforms, digital intermediaries, platform-dependent producers, e-tailers, digital financial providers and other digital-only producers.

That design can improve the accounts. It forces supply to reconcile with use and makes imported digital services visible. It does not automatically reproduce the 2023 core-industry total. One measure classifies producers. The other crosses industries, products and transaction types.

Data assets add a second boundary change. The 2025 SNA classifies data and databases as intellectual-property products when they are organized for efficient access and expected to support production for more than one year, or are held for sale or licensed access. Own-account data assets are generally valued from production costs. Purchased data are valued at market prices. Normal obsolescence reduces both their value and their volume.

Ownership is economic, not merely legal. A long-term license may create a fixed asset for the licensee when it transfers the risks and rewards of use. Repeated short-term payments generally buy a service instead. A license to reproduce can transfer part or all of the original intellectual-property product. These choices decide which unit records investment, depreciation and value added.

The NBS plan does not yet state its choices on those points. Capitalizing work that was previously treated as an expense can raise measured investment and GDP even if the underlying activity has not changed. A new estimate can also rise because platforms enter the business register, imported services are rerouted correctly, or census data replace survey estimates. Each adjustment may improve accuracy. None should be presented as volume growth.

Prices create a separate problem. Software and data are heterogeneous, often produced in-house, and hard to deflate. The 2025 SNA notes that own-account software and data volumes may have to be inferred from input volumes. If China publishes only current-price totals, users will not be able to separate more output from higher labor, cloud or acquisition costs.

Implications: publish the bridge before the headline

The useful release is not a larger number. It is a reconciliation package.

First, the NBS should restate at least the 2023 baseline on the new perimeter, showing which old industries remain, which activities are added and where overlaps are removed. The old 12.7555 trillion yuan figure should either nest inside the new total or sit beside it with a clear reason.

Second, the release should separate current-price change from volume change and identify the deflators used for software, data, cloud services and platform output. A single nominal growth rate would collapse the question the new accounts are meant to answer.

Third, the data-asset account needs a flow bridge: opening stock, new capital formation, transfers, depreciation or obsolescence, revaluations and closing stock. It also needs the capitalization threshold, service-life assumptions and economic-owner rule.

Finally, the bureau should publish enough product and industry components to reproduce the headline, together with source notes and a revision calendar. A supply-use table earns trust through reconciliation, not through the number of cells.

China's reform can make the digital economy more visible without pretending the old accounts were worthless. The test is whether users can distinguish newly observed activity from newly produced activity. Until a common-perimeter history exists, the first jump should be read as a statistical level shift. Growth starts with the second comparable observation.