06-reference/research

psf leverage ratio disclosed filings

2026-10-09·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
professional-servicesleverageuk-llpprimary-sourcesdeloitte

Nobody publishes a leverage ratio, but UK LLP statutory accounts make one computable - and at Deloitte it has fallen four years running

The question

"Does any professional-services firm disclose an actual leverage ratio or partner-to-associate trend anywhere - listed-consultancy S-1s, UK LLP statutory accounts, or audit transparency reports?"

Promoted from [[2026-07-27-psf-pyramid-economics-post-agent-analyst-tier]], which rated its own central claim low-to-moderate confidence precisely because no measured leverage series was known to exist. The job here is to settle observability, not to re-argue the flattening thesis.

What we already know (from the vault)

What the web says

Convergences and contradictions

Synthesis for RDCO

The honest answer has three parts, and only the first is unqualified. No professional-services firm anywhere publishes a labelled leverage ratio or a partner-to-associate trend. Not in SEC filings, where the phrase returns zero hits across a quarter-century of 10-Ks and the only grade-level disclosure belongs to a defunct micro-cap. Not in the Big Four's voluntary annual reviews, which publish partner counts and people counts in separate tables and never divide one by the other. The metric is not withheld so much as never constructed - these firms report revenue per partner and profit per partner, which are the money ratios, and they do not report the staffing ratio that Maister calls the master variable.

The second part is the actual find: UK limited-liability-partnership statutory accounts make the ratio computable, and that is a legal guarantee rather than a firm's discretion. SI 2008/1913 Schedule 1 paragraph 66 compels every large and medium UK LLP to state its average number of members for the year, and the companion employee-numbers note - at Deloitte, split into fee-earning and non-fee-earning on both headcount and FTE bases - supplies the denominator. That converts the parent brief's question from inference to observation for every Big Four UK firm, every magic-circle law firm, and every UK LLP consultancy above the small-company threshold, going back as far as the filings run. The filings are free, un-authenticated, and at Companies House the member-level LLAP01 and LLTM01 event filings carry effective dates, so the partner count is reconstructible at daily granularity and not just annually. For the one firm I actually read, the direction is unambiguous and monotonic: employees per equity member fell from 38.9 to 32.2 across FY23-FY26 on a headcount basis, 37.1 to 30.0 on an FTE basis.

The third part is why that number should not be quoted as evidence for the agent thesis. The decline is driven more by partners being added than by juniors disappearing, which is the signature of a firm protecting its promotion incentives through a demand trough, not of a shrinking apprenticeship base. The window is exactly the 2023-25 consulting downturn the parent brief named as the confound, so one firm over four years cannot separate cycle from structure. And the member boundary is gameable in the direction that would manufacture a false flattening signal: Deloitte's 707 salaried partners sit in the employee count, the true all-partner ratio for FY25 is closer to 17 than 34, and no statutory series exists for the salaried tier. The measurement problem has moved from "no data" to "data with a named, bounded confound," which is a real upgrade and still short of proof.

For the founder's live bet this is a credibility asset rather than a strategy input. phData is the main bet and he is a Principal Consultant / Data Solutions Architect there, not a seller of services, so nothing here changes what he sells. What it changes is the quality of what he can say. The "pyramid is flattening" claim is currently made in public almost entirely off trade-press headcount stories; being able to cite a filed statutory note, name the paragraph of the statutory instrument that compels it, show a four-year monotonic series, and then volunteer the salaried-partner confound that most people quoting the trend have not noticed is a sharper move than having a stronger opinion. That is the "agents in production, measured honestly" register the credibility bet runs on. The cheap, concrete action is vault hygiene: the parent brief's "zero PSF disclosures" finding needs a correction block pointing here, the same way its Maister-formula correction works.

Why this is in the vault

It overturns a specific stated finding in [[2026-07-27-psf-pyramid-economics-post-agent-analyst-tier]] ("no large PSF has disclosed a leverage ratio... the measurement does not exist yet"), which needs a correction block, and it supplies the primary-source citation set - SI 2008/1913 Sch 1 para 66, Deloitte LLP's FY26 Note 4, Companies House LLP event filings - that any future RDCO writing on post-agent professional-services economics should use instead of trade-press headcount stories, per the credibility-for-phData-sales bet.

Open follow-ups

Related

Sources

Primary, read directly

Secondary / shape-only, not independently read

Blocked or not reached

Vault