Size Filter: The Turnover and Asset Thresholds in Screening

Definition

The size filter is the screening bound that keeps the comparables in the tested party’s size band — the turnover and total-asset thresholds (the floor and the ceiling) below which a candidate is treated as too small to be comparable and above which it is treated as too large. The reason size is a comparability factor: scale changes the [PLI] (/docs/glossary/profit-level-indicator) for economics that have nothing to do with the transfer price — the purchasing leverage (the larger buyer’s lower input cost, the volume discount), the fixed-cost absorption (the larger operation’s overhead spread across more revenue, the operating margin’s scale component), and the finance access (the larger entity’s cheaper debt, the [working capital] (/docs/glossary/wc-adjustment) structure). A candidate at 20× the tested party’s turnover is not the tested party with a different price — it is a different economics, and its margin carries a scale component the tested party does not have. The filter’s practice form: a band around the tested party (the floor and ceiling as multiples of its turnover/assets — the search design fixing the bounds, typically on the order of 0.5×–5× or 1×–10×, the exact multiple a design decision), applied in the quantitative screening sequence, with the bounds documented in the study (the [accept-reject matrix] (/docs/glossary/accept-reject-matrix) records every candidate dropped at the bound, with the bound value stated).

The size filter, in one application:
  1. The tested party's size (the turnover, the total assets — the band's centre)
  2. The bounds (the floor and ceiling — the multiples fixed in the search design)
  3. The candidates (each candidate's turnover/assets, per period)
  4. The decision (inside the band → to the next screen; outside → the recorded reject)

The element	The content
The measure	Turnover (the revenue scale) and total assets (the balance-sheet scale) — the two proxies; the turnover is the primary, the assets the check
The bounds	The floor and ceiling as multiples of the tested party (the design decision — the 0.5×/5× or 1×/10× band, the stated rationale where the band is asymmetric)
The period rule	The candidate’s size per financial year (a candidate inside the band in FY24 but outside in FY25 is a period completeness question, not an automatic reject)
The record	Every bound-reject in the matrix with the candidate’s size and the bound value — the specific number, not “too small”

The working read (the quantitative screening guide): the size filter is the first quantitative screen, and its discipline is symmetry of treatment — the band excludes both tails (the too-small and the too-large), because a floor-only filter (keeping everything above a minimum) quietly lets the scale economics in from the top. The examination reads a floor-only size bound as the pattern that keeps the large, high-margin comparables (the scale component flatters the range) while the ceiling is never applied. The comparable set the filter produces should be centred on the tested party’s scale — and the [regional vs local] (/docs/benchmarking/regional-vs-local-comparables) interaction matters: in a thin home market, the band is the tension (the local population inside the band may be too small for the [statistical significance] (/docs/glossary/statistical-significance) floor), which is a design decision — widen the geography, or state the thin-market justification — not a silent loosening of the bounds.

Example

An Indian contract manufacturer: turnover ₹180 cr, total assets ₹95 cr. The search design fixes the band: turnover 0.5×–5× (₹90 cr – ₹900 cr), assets 0.5×–5×. The candidate screen:

Candidate	Turnover	In band?	Decision
C1	₹140 cr	Yes	→ next screen
C2	₹62 cr	No (below the 0.5× floor)	Reject — size floor (stated: ₹62 cr < ₹90 cr)
C3	₹1,150 cr	No (above the 5× ceiling)	Reject — size ceiling (stated: ₹1,150 cr > ₹900 cr; the scale economics)
C4	₹210 cr in FY24, ₹55 cr in FY25	FY25 below the floor	Flag — the period rule: the FY25 collapse is itself a comparability event, reviewed in the qualitative screen

The C3 reject is the instructive one: the large manufacturer’s OP/OC is higher than the tested party’s (the fixed-cost absorption, the purchasing leverage) — and an unceilingled filter would have kept it, pulling the range up with a scale component the tested party does not have. The band, applied to both tails, is what keeps the [comparable set] (/docs/glossary/comparable-set) at the tested party’s scale.

See also

• Benchmark Search Design
• Quantitative Screening
• Comparable Set
• Working Capital Adjustment

FAQ

What multiples are “normal” for the size band? There is no prescribed multiple; the practice band is 0.5×–5× or 1×–10× of the tested party’s turnover (and the same order on assets), the exact choice a design decision recorded in the search design. The rationale to state: why this band (the scale economics — the leverage and absorption arguments — say the comparables should be of a comparable order, and the band is the operational form). Asymmetric bands (a 0.5× floor with a 20× ceiling, or vice versa) need the asymmetry explained — a band that is generous in the direction that flatters the range is the examination’s pattern, and the stated rationale is the defence.

Is the size filter an adjustment or a screen? A screen — it decides membership (the candidate in or out of the comparable set), not a value correction. The working capital adjustment, by contrast, corrects a member’s PLI value for a known, quantifiable difference. The distinction is the discipline: scale is treated as a comparability failure (the candidate is a different economics — excluded), while working capital is treated as a differential (the same economics, a known distortion — corrected). A candidate whose size is outside the band is not “adjusted down” into the set — it is rejected at the bound, with the number recorded.

What if the whole home market is larger (or smaller) than the tested party? Then the regional vs local decision carries the design: either the geography is widened (the regional population has the scale match) or the thin-market justification is stated (the home market’s structure makes the band unsatisfiable — the best available population, with the limitation disclosed in the file). The failure mode is the silent band-loosening (the bounds changed to fit the available population, with no record) — the documented alternative (the widened geography or the stated limitation) is the defensible one, and the [benchmarking mistakes checklist] (/docs/benchmarking/benchmarking-mistakes) treats the unstated bound change as a named mistake.