Method

Methodological transparency, data support, willingness to be challenged

Data limits are stated, not hidden. A first number published and later corrected is preferable to a perfect number never released. This page brings together, in one place, the sources used across all Vespri reports and the limits of each — instead of leaving them scattered in the appendices of individual issues.

§ 01 — Principles

Three editorial rules

01

Verifiable sources

Public statistical and institutional bodies, public company financial statements, job-posting aggregators and recruiting portals with accredited data access — always within each source's terms of use.

02

Stated limits

Every report includes a methodology sheet: what the data captures, what it doesn't, and where a figure should be read as a lower bound, not an absolute value.

03

Iteration, not perfection

Publish, then expand: more countries, more companies, more sectors in later versions — with revisions always tracked.

§ 02 — The sources

Four sources, four different reliability profiles

Employment

Institutional restructuring monitoring

A public monitoring database on European corporate restructuring: 31,427 events, 2002-2026, 29 countries (EU-27, Norway, United Kingdom). It records the year of announcement, not the year jobs actually disappear — a multi-year plan weighs in full on the announcement year. UK coverage stops after January 31, 2020 (Brexit). Sector classification is heuristic, based on company-name recognition and spot-checked: roughly 90% of all events concern non-engineering sectors and are excluded from the analysis.

Finance

Public company financial statement data

81 listed European stocks, 7 sectors, 2022-2025 history — a structural limit of free access to this type of data, not extendable beyond 4 fiscal years without a paid provider. Any Sector×Year cell with fewer than 3 companies with valid data is excluded from charts and rankings, to prevent a single stock from passing as a sector average.

Skills

Accredited job-posting aggregator

A job-posting aggregator with API data access: breadth across seven sectors, four geographic areas, dozens of companies — but only a truncated snippet of the description. Percentages from this source are a lower bound on the real prevalence of each skill, not an exact share.

Skills

Corporate recruiting portals

Direct access to the public endpoints exposed by the recruiting platforms of the companies analyzed: full posting text, but only for the 27 companies — out of 93 total — that expose an accessible endpoint. 5,887 of the 16,386 total postings, averaging 4,199 characters per posting.

§ 03 — Stated limits

What to keep in mind

Limits collected here explicitly

Employment The database covers 29 European countries, not Serbia (where Stellantis has a strategic plant in Kragujevac) nor other non-EU countries. The sector classification is not an official code in the database: it is an attribution by company-name recognition, corrected multiple times over the course of the analysis.
Finance — margins and leverage Automotive leverage is structurally inflated by the captive finance companies (customer financing) present at Volkswagen, Stellantis, BMW and Mercedes-Benz — not comparable at equal operating risk with a sector like Defense. The Energy sector aggregates upstream oil&gas, integrated utilities and a pure renewables player: structurally different businesses, so the aggregated figure is indicative, not a homogeneous average.
Skills — job postings The Country field remains partly unresolved in the full-text data (~90% "Other/N.A.") due to a structural source-side limit — which is why the analysis is aggregated by sector across all countries together, not split by geography. The classification is heuristic (keyword matching): a starting point to be refined, not an official figure. Academic background, seniority and role function are captured only when the posting states them explicitly — their absence does not mean the requirement doesn't exist. Coverage is not uniform between the two sources (93 companies on the posting aggregator, 27 on full text): the deeper subset is not necessarily representative of the entire sector.

Questions about the data, before citing it

Full methodology available on request for anyone citing a Vespri figure in a report, an internal note or an article.