A qualification test, not a definition. The OECD’s 2026 report argues that data spaces (in its narrow sense — federated governance + open interoperability + provider-retained usage control) have a comparative advantage over simpler mechanisms only where four conditions hold jointly, and that funding or building one where they do not is paying co-ordination costs for nothing.
“Four specific boundary conditions define the class of problem (use cases) for which data spaces have a comparative advantage over the alternative mechanisms compared in Table 3.1 (e.g. bilateral contracts, centralised platforms and approaches based on PETs. These conditions are (i) three or more parties; (ii) competitive or jurisdictional relationships among them that preclude trusting a single commercial intermediary; (iii) non-excludable collective value that bilateral contracts cannot capture; and (iv) ongoing data flows requiring continuous governance rather than one-off transactions. Where these conditions are absent, simpler data-sharing mechanisms may be sufficient and may not justify the co-ordination costs that data spaces entail.”
⟨Corrected 2026-10-29: the passage previously set here as a blockquote was a paraphrase, not the report’s words. The text above is verbatim §3.2, unclosed parenthesis and all; the substance of the paraphrase was accurate.⟩
Source: §3.2 of OECD, Data Spaces between Co-ordination Challenges, Financial Viability and Public Policy, 2026, restated as policy consideration 1 in §5.1.
The four conditions
| # | Condition | What fails without it |
|---|---|---|
| 1 | Three or more parties | Two parties can contract bilaterally. Governance infrastructure is overhead. |
| 2 | Competitive or jurisdictional relationships precluding a trusted single intermediary | If one actor can be trusted to hold the data, a centralised platform is cheaper and faster. |
| 3 | Non-excludable collective value bilateral contracts cannot capture | If the value is fully appropriable pairwise, bilateral contracts capture it; nothing is left for the commons layer to do. |
| 4 | Ongoing flows requiring continuous governance | A one-off transfer needs a contract, not an institution. |
Necessary, not sufficient. §5.1 adds a second stage: even where all four hold, assess sector-level preconditions — pre-existing digital infrastructure among the intended participants, identifiable anchor firms with sufficient incentive to lead, and a value proposition specific enough that participants can compute their own return. The report’s five domains with structural advantage, derived by grouping 147 Radar initiatives by value proposition, are: supply-chain visibility across competing firms; health data across jurisdictions; multimodal mobility; energy systems with distributed prosumers; smart city services.
Why each alternative mechanism falls short
Table 3.1 of the report (“Comparison of data-sharing mechanisms and enablers”) compares bilateral contracts, centralised platforms, open data/APIs, PETs and data spaces across governance, interoperability, usage control, multi-party scalability and co-ordination cost. The table did not survive PDF→Markdown conversion (columns collapsed; see the conversion caveat on the paper page). The following is reconstructed from the §3.2 prose, which states the same findings in text — ⟨Inference: the table’s cells are not quoted because they are not recoverable; the substance below is directly asserted in the body text.⟩
- Bilateral contracts — enforceable and well understood, but cost scales with the number of pairs and they cannot capture value that is non-excludable across the group. Condition 3 is what they fail.
- Centralised platforms — efficient and fast to deploy, and they solve interoperability by fiat, but they require participants to trust the operator. Condition 2 is what they fail, and Covisint is the canonical demonstration: the failure was institutional, not technical.
- Open data and APIs — low co-ordination cost, but they offer no post-access usage control, so they are unavailable wherever the data are competitively sensitive or personal.
- PETs (federated learning, MPC, TEEs, synthetic data) — they address the confidentiality dimension, and they do it strongly. What they do not do, in the report’s words, is “address the institutional co-ordination problem of who governs, who pays and how collective value is allocated”. See PETs as complement, not substitute and cryptographic trust infrastructure.
- Data spaces — the only mechanism combining neutral governance, open standards-based interoperability and provider-retained usage control. The claim is explicitly about integration, not about any single component: “Governance neutrality is available through public bodies, usage control through bilateral contracts, and open standards through industry consortia. However, no alternative mechanism combines all three.”
The price of that combination is co-ordination cost, which is why the conditions matter: the comparative advantage is real but narrow.
The governance configurations
Table 3.2 of the report maps governance configurations on two axes (roughly: centralised↔federated governance against bilateral↔multilateral relationships). This table was also lost in conversion — only title and source survived. The surrounding prose supports the configuration data spaces occupy: federated governance over multilateral relationships, which is Williamson’s hybrid form — “not a loose amalgam of market and hierarchy but possesses its own disciplined rationale”. The other cells correspond to the alternatives above. ⟨Inference: stated as reconstruction, not as the report’s own cells.⟩
Applying the test at SINE
⟨Inference — this is our reading, not the OECD’s; the report names no SINE-adjacent initiative.⟩ The test is useful precisely because our own candidate cases do not score identically:
- Pathfinder Network — scores on all four. Many parties; direct competitors in the same supply chains; the value of a complete upstream PCF is non-excludable across the chain and cannot be bought pairwise at any sensible cost; flows are continuous because footprints are recomputed every reporting cycle. The OECD would class this as a structural-advantage case (supply-chain visibility across competing firms is its first named domain).
- CCTI / the reinsurance case — scores clearly on 1, 2 and 4; condition 3 is the one to argue, because much of the value of better catastrophe data is appropriable by the party that holds it. Where condition 3 is weak, the OECD’s prediction is that participants will prefer bilateral arrangements and the commons layer will struggle to fund itself — which is a hypothesis we can test rather than a verdict.
- The GIZ agricultural value-chain tender — conditions 1, 3 and 4 hold; condition 2 holds in a different register (the obstacle is less competition between smallholders than the absence of any actor with the legitimacy to hold the data). The Agdatahub failure is the directly relevant precedent — see the bootstrapping problem.
The condition our work most often turns on is 2, and the OECD’s framing of it is sharper than ours has been: the question is not whether an intermediary would behave well but whether participants can be made structurally unable to need to trust that it will.
Cross-references
- OECD, Data Spaces between Co-ordination Challenges, Financial Viability and Public Policy (2026) — the source
- The bootstrapping problem — what happens after a case qualifies
- Data space business models and viability — what happens when it qualifies and still fails
- Coordination failure vs. bargaining failure — the same split in our own vocabulary
- Data Commons · Ostrom’s commons design principles
- Cryptographic trust infrastructure — the mechanism the conditions rule in alongside, not instead of
- IDSA · Catena-X · Gaia-X
Sources
sources/research/2026-10-27-oecd-data-spaces-co-ordination-viability-public-policy/2026-10-27-oecd-data-spaces-co-ordination-viability-public-policy.md
(§3.2, §3.3, §5.1; Tables 3.1 and 3.2 lost in conversion — see caveat above).