Methodology

We built the headline number (€20.5 billion) bottom-up from public data so any journalist or analyst can recreate it. This page documents every input, every formula, and every assumption we made — including the ones we'd like to improve in next year's edition.

Scope

We model 2025 calendar-year activity in the seven largest EU labour markets — Germany, France, Italy, Spain, the Netherlands, Poland and Sweden — which together cover approximately 78% of EU27 employment. We then gross the result up to the EU27 by dividing by 0.78.

We do not model the UK, Switzerland, Norway or other non-EU labour markets in this edition.

Definitions

• Ghost job: a publicly visible posting that has been live for > 60 days without a recruiter action (response, advance, reject) on any application. Three sub-classes: paused, phantom, evergreen.
• Ghosted applicant: a candidate who applied to a real, non-ghost posting and never received any response — automated rejection included.
• Opportunity cost: the value of output not produced because a real (fillable) role stayed open longer than the country benchmark time-to-fill.

The three cost buckets

1. Recruiter wasted hours

recruiter_eur =
  ghost_postings
  × apps_per_posting
  × recruiter_min_per_app / 60
  × recruiter_cost_per_hour

2. Candidate wasted hours

candidate_eur =
  ( ghost_postings × apps_per_posting
  + real_postings × apps_per_posting × ghosted_applicant_share )
  × candidate_min_per_app / 60
  × labour_cost_per_hour

3. Opportunity cost

opportunity_eur =
  real_unfilled_postings
  × extra_days_to_fill
  × output_per_filled_day
  × 0.18  // attribution to ghost-job crowding [9]

Country totals are summed, then multiplied by 1/0.78 to project to EU27. Result: €20.5B.

Inputs by country

Every figure below is published in the open dataset (CSV download on the report page). Sources are numbered against the list at the bottom.

Limitations & assumptions

• We use a 60-day no-action window to classify ghost jobs. A 30-day window raises the headline by ~22%; a 90-day window lowers it by ~14%. We chose 60 days as the figure most defensible across seasonal hiring cycles.
• Application volumes per posting come from ATS benchmarks (Greenhouse, Workable). We do not have visibility into companies that don't use a commercial ATS, so true volumes may be slightly higher.
• We attribute only 18% of opportunity-cost extra days to ghost-job crowding. The other 82% is everything else (skills shortage, visa friction, slow hiring committees) — out of scope here.
• Candidate time-per-application (mean 24–28 minutes) is self-reported in survey data. AI-assisted CV tailoring will likely lower this in 2027.
• Industry shares are weighted from public posting data and are indicative — granular industry-level recruiter cost data does not exist for every country.

Right of reply

If you represent a company, vendor, or government body and believe one of our inputs is materially wrong, write to research@tenperzent.com with the source. We update the dataset on a rolling basis and document every revision.

Sources

1. [1] Eurostat — Labour Force Survey — employment, unemployment, vacancy rates by country (accessed 2026-04-22)
2. [2] Eurostat — Hourly labour costs — annual data (accessed 2026-04-22)
3. [3] Eurostat — Job vacancy statistics — quarterly data (accessed 2026-04-22)
4. [4] Indeed Hiring Lab — Indeed Hiring Lab — time-to-fill and posting longevity (accessed 2026-04-18)
5. [5] LinkedIn Economic Graph — LinkedIn Economic Graph — share of postings open >60 days (accessed 2026-04-18)
6. [6] Greenhouse — Greenhouse Source-of-Hire Report 2025 — applications per requisition (accessed 2026-04-15)
7. [7] Workable — Workable Hiring Pulse 2025 — recruiter screening time per application (accessed 2026-04-15)
8. [8] Glassdoor Economic Research — Glassdoor Workplace Trends — candidate response rates (accessed 2026-04-12)
9. [9] OECD — OECD Employment Outlook 2025 (accessed 2026-04-10)
10. [10] Greenhouse — Greenhouse Candidate Experience Report 2025 — share of applicants who never receive a response (accessed 2026-04-09)
11. [11] Revelio Labs — Revelio Labs — share of public postings classified as 'evergreen' or 'phantom' (accessed 2026-04-08)
12. [12] European Central Bank — ECB Statistical Data Warehouse — average GDP per hour worked, EU27 (accessed 2026-04-07)

License: CC BY 4.0. Cite as: TenPerZent (2026). The €20.5B Ghost Economy.