SFDR 2.0 Product Categories: Re-Classifying Funds Into Sustainable, Transition and ESG-Basics

The SFDR review does not tidy up Article 8 and Article 9. It replaces them with named categories judged on portfolio thresholds and hard exclusions — which turns a disclosure exercise into a data exercise, and exposes exactly the feeds most asset managers never built.

On 20 November 2025 the European Commission published its proposal to revise the Sustainable Finance Disclosure Regulation (COM(2025) 841 final). The headline is that the informal shorthand everyone has used for six years — “Article 8 fund”, “Article 9 fund” — is on its way out. In its place the proposal introduces three named product categories: Sustainable, Transition and ESG basics. This is still a proposal, working through the ordinary legislative procedure, and the detail will move before it is adopted. But the direction is settled enough to plan against, and the planning is a technology job before it is a legal one.

What actually changes: labels become tests

The current regime is, in practice, a disclosure classification. A product is Article 8 if it promotes environmental or social characteristics, Article 9 if it has sustainable investment as its objective. The distinction has always been soft at the edges, which is why so many funds sit in an ambiguous middle and why the market invented the un-official “Article 8+” tier to describe it.

The proposed categories are different in kind. They are eligibility tests with numbers attached. To use a category name — and, crucially, to use any ESG-related term in the product’s name or marketing — the proposal requires the product to meet defined criteria. The common architecture across the three categories, as drafted, is a minimum 70% of the portfolio invested in line with the category’s strategy, plus a set of mandatory exclusions. Miss the threshold or breach an exclusion and the product cannot carry the label, cannot use the sustainability-flavoured name, and drops into an uncategorised bucket with meaningful marketing restrictions.

That is the shift that matters for the technology function. Under the old regime you could defend a classification with a well-drafted policy statement. Under the proposed one you have to be able to compute, monitor and evidence a portfolio-level figure against a defined denominator, continuously, and prove the exclusions hold. A label you can no longer calculate is a label you can no longer use.

The three categories, in engineering terms

Sustainable is for products investing in activities and undertakings that are already sustainable or contribute to environmental or social objectives — the closest successor to today’s Article 9, with the strictest fossil-fuel and harm exclusions. Transition is the genuinely new idea: products channelling capital towards companies or projects that are not yet sustainable but sit on a credible transition path, evidenced through science-based targets, climate benchmarks or structured engagement. ESG basics is the catch-all for products that apply recognised ESG approaches — best-in-class, ESG outperformance, exclusion screens — without a sustainability or transition objective.

Each category, as proposed, offers deemed-compliance routes that are themselves data-hungry. A product can be treated as meeting the contribution threshold if it holds at least 15% in EU Taxonomy-aligned investments, or if it replicates or is managed by reference to an EU Climate Benchmark (a Climate Transition Benchmark or Paris-Aligned Benchmark). Both routes assume you can source and hold Taxonomy alignment data or benchmark constituent data at position level. Neither is a field most order-management or fund-accounting stacks were built to carry.

Where the data gaps actually sit

Run your existing range against the proposed criteria and the gaps are predictable, because they are the same gaps the current regime let you paper over.

Look-through you do not have. A 70% portfolio threshold has to be computed against something. Funds of funds, mandates holding other pooled vehicles, and multi-asset products need position-level look-through to classify the underlying holdings. Plenty of managers hold this only for the largest lines and estimate the rest — which is fine for a policy statement and not fine for a computed, auditable figure.

Taxonomy alignment as a real feed, not a marketing number. The 15% deemed-compliance route depends on Taxonomy-alignment percentages you can trust and reconcile, not a vendor headline figure applied at fund level. If you cannot trace the number to issuer-level source data with a lineage a supervisor can follow, it will not survive scrutiny.

Transition evidence that is structured, not narrative. The Transition category rewards credible transition paths, which means capturing science-based-target status, benchmark membership or engagement records as structured, queryable fields — not as prose buried in an investment memo that no system can aggregate.

Exclusion screens that run continuously. Mandatory exclusions — controversial weapons, tobacco, serious breaches of the UN Global Compact or OECD guidelines, thresholds on coal and lignite — have to be monitored, not asserted once at launch. That needs a screening feed and a breach workflow wired into pre-trade and periodic compliance, with an evidence trail.

None of this is exotic. It is the familiar pattern in which a regulator turns a claim into a reportable, computed obligation and firms discover their feeds were never built to carry it. The gambling levy did the same thing to operators’ reporting: the moment a number becomes a filing rather than a statement, the question stops being “what is our position” and becomes “can we source, compute and evidence it on demand”.

Why you start now, not at adoption

The tempting response to a proposal still under negotiation is to wait for the final text. That is the wrong instinct here, because the expensive part is not reading the regulation — it is acquiring data you do not currently hold and back-filling it across an existing range. Taxonomy alignment, transition evidence and look-through cannot be conjured at re-classification. They have to be sourced, contracted from data vendors, integrated and reconciled, and that lead time runs in quarters.

The sensible programme is to treat re-classification as a data-plumbing project that happens to have a legal deadline attached, rather than a legal project with a data annex. Map each existing product provisionally against the three categories, mark every criterion you cannot currently compute from a trusted source, and rank the gaps by how long the feed takes to stand up. This is re-plumbing, not re-labelling — the same discipline that CRR3 forced on regulatory reporting teams, where the calculation change was trivial next to the data lineage it demanded.

Assume the thresholds and category boundaries will shift in negotiation, and design the pipeline to be re-parameterised rather than re-built. The one thing that will not change is that the labels have become tests, and a test you cannot compute is a fund you may have to rename or reposition. The managers who discover that at adoption will be the ones explaining to a board why a product lost its name.