Simplifying a report does not fix what the report was not doing

A sustainability report under the first CSRD cycle is assembled roughly the way the rules intended. A reporting function gathers data from the operating units, has it checked by an assurance provider, and files the result against a deadline set in legislation. The document is public and filed on time. Then the next cycle opens. The rules say a great deal about what goes into that document and almost nothing about what anyone does with it afterward.

The EU Council has given final approval to the Omnibus I package, cutting CSRD scope and trimming what remains. Advisory firms across Europe have a what-this-means-for-you explainer in circulation, and the trade coverage has settled on red tape as the frame. The argument that followed divides in two: the cut went too far and investors have lost information they were starting to rely on, or the cut was too small and the remainder still costs more than it returns.

The burden was real. For a mid-size firm with no sustainability function, the first cycle meant hiring, and an assurance bill arriving whatever the size of the business. Trimming scope for companies in that position is defensible on its own terms, and the case for it was a serious one. That should be granted without hedging.

Both sides of the volume argument assume the reports were wired to the outcomes they were adopted for. Neither case states that assumption out loud. Institutional sociology has been examining it since the 1970s.

Two kinds of gap

Meyer and Rowan, writing in 1977, described organizations taking on formal structures because their environment expects them, while the work inside carries on under its own logic, and treated that gap as a sensible adaptation to conflicting demands. Westphal and Zajac later put a measurable case underneath it: a study of stock repurchase programs across large US corporations over six years in the late 1980s and early 1990s, testing which firms announced a program and then never bought the shares. Nobody in that story has to be accused of bad faith. The announcement was the product, and the purchase would have added nothing to an effect already obtained. The non-implementation was measurable, which is what makes the case worth citing rather than merely plausible.

That pattern is policy-practice decoupling, where a structure is announced and then left unused. Bromley and Powell’s review of the accumulated literature names a second form, and that one describes corporate reporting. They call it means-ends decoupling: the policy is adopted and the practice genuinely happens, while the connection to the outcome it was adopted for never appears. Nothing here requires anyone to fake a report. The data gets collected and the assurance gets bought. The document exists and anyone can download it. The absent step is the one where a figure inside it changes what somebody does, which is the condition an earlier piece on quarterly portfolio reviews described from the other side of the same problem.

What the trim reduces

Cutting scope lowers the cost of the activity. Whatever connects a disclosure to a decision was built inside the company, and an amendment to a reporting regime reaches none of it. A company that reported without altering its operations gets a smaller bill for the same conduct. A company that had already tied its disclosures to capital allocation keeps that arrangement, because it built the arrangement itself and the regime only ever set a floor. Neither the original regime nor the amendment asks a company to state which decision any disclosure feeds, and that is the one requirement that would have tested the link. Simplification sorts companies by their compliance spend, and it leaves open the question of whether that spend was returning anything.

Which disclosures to keep

There is a stronger objection to this, and it runs through an argument made on this site about forbidden local exceptions: banning a deviation moves it to wherever the rule has no reach, because the conditions that produced it are unchanged. Withdrawing an obligation has a comparable effect at the other end. Conduct carries on once the disclosure covering it stops, and the record goes. For a company that had wired its disclosures into its own decisions, that record carried something an outsider could use, and the objection holds at full strength. Where the reports were decoupled, the record was a record of activity, and the activity had already come loose from the outcome. The loss is uneven, which makes it hard to price and easy to argue about from either side.

So there is a decision in front of anyone who runs a reporting function, and two defensible answers to it. The first is to report to the new floor and no further, treating disclosure as a compliance cost and taking the saving the legislators meant to hand over. That is coherent, and for a company with thin margins and a small finance team it may be the only answer available. The second is to hold the collection where it is and change what it is for: for each disclosure the company was producing, name the person who would act differently if the number moved. Keep the disclosures that pass that test and stop the ones that fail, whether or not the law still asks for them. That is the question worth putting to any framework, including the ones collected on our Insights page.

The second answer is the better one, for a reason that has little to do with sustainability. A company that cannot answer the test for its emissions data is unlikely to answer it for its operating reports either, and the emissions data is a set the regime already obliged it to build. The money that went into building it has been spent.