Choosing PLI, Years and Benchmarking Parameters

The parameters step is the study’s configuration record. Every value you set — the PLI, the tested party margin, the years, the filters, the percentiles — is stored on the study in a single parameters block, and every run is reproducible from it. The parameters are not a runtime preference: a reviewer reads them to check the search was designed correctly, and the report carries them as the record of what the study did.

The PLI

The profit level indicator is the ratio the study computes for the tested party and for every comparable. The available PLIs and their formulas:

PLI	Formula
OP/OC (default)	Operating Profit / Cost
OP/Sales	Operating Profit / Revenue
TP/Sales	Operating Profit / Revenue — the same quantity as OP/Sales, under a label the report carries for the tested-party side
EBIT/Sales	Operating Profit / Revenue
EBITDA/Sales	EBITDA / Revenue
Berry Ratio	Gross Profit / Operating Expenses
EBIT/Total Assets	Operating Profit / Total Assets

The choice follows the tested party’s function — routine service and distribution functions point to an operating-profit base, asset-heavy functions to a total-assets base. The reference for each ratio is in the PLI reference, and the method-level logic for choosing between them is in How to Choose a Method.

Three properties the design guarantees:

• The PLI is computed the same way everywhere. The screening’s extreme-value gate, the statistics and the report all read the selected PLI’s own ratio — gross profit over operating expenses under the Berry Ratio, operating profit over total assets under EBIT/Total Assets — each against a band calibrated for that quantity. Changing the PLI therefore changes both the range and the accept-reject outcome; a study’s verdict cannot be read out of context.
• The PLI has to be computable from your file. Each PLI declares the dump columns it needs, and a run whose file cannot supply them is blocked at the wizard rather than failing quietly at the statistics.
• The tested party margin is a parameter, not a result. The tested party’s own PLI value (default 10%) is what the range is measured against; it is recorded on the study, not re-derived at review time.

The method

There is no method dropdown in the wizard, deliberately. What you configure is a margin benchmark: a PLI, a tested party, a population, and a range at chosen percentiles — the transactional net margin shape of a benchmark. The TNMM guide covers the method itself; Transfer Pricing Methods covers how it sits against CPM, CUP, and the transactional methods. Where a method is named — in the master report — it is recommended by the report engine’s decision engine from the derived FAR characterization, deterministically and on the record, not typed in by whoever configured the run.

One boundary on what the numbers mean: the platform computes the range and the tested party’s position within it, and the resulting TP adjustment is the computed quantity. It does not run comparability adjustments over the pool — there is no working-capital adjustment, no functional-segregation adjustment, no other quantified strip-out of a comparable’s financials. Where those adjustments matter to your position, they are your analysis, outside the tool.

The years and the averaging

• Single Year (Latest) uses only the most recent detected year: one PLI per company, no consolidation.
• Multi-Year (Golden Rule — Aggregate Raw Financials) — the default — consolidates across the detected years, and the consolidation is the Golden Rule: the raw financials are aggregated (sum of numerators over sum of denominators), never the per-year ratios averaged against each other. Older studies may carry the legacy simple value; the engine treats it exactly as it treats the Golden Rule option, so the computation does not change with the label.

Which is what keeps a three-year pool and a three-year tested party commensurable. The mechanics and the consistency requirement are in Multi-Year Averaging and Multi-Year Data.

The search scope

Parameter	Default	Meaning
Jurisdiction	India	The compliance context, and the arm’s-length range convention where the platform carries a curated rule for it
Revenue min / max	0 = unset	The size scope of the comparable population, read in the chosen currency and scale unit (actual, thousand, million, crore)
Maximum RPT %	25	Rejects a company whose related-party transactions exceed this share of its own revenue
Employee cost min %	0 = unset	Rejects companies whose employee cost falls below this share of revenue — a cost-structure filter on the pool
Comparable pool	—	The companies in the file you uploaded: no parameter narrows it by geography or industry code
Lower / upper percentile	25 / 75	The bounds of the arm’s length range — the interquartile range by default

A filter left at its sentinel value (0, or the upper cap) means unset — the filter is not applied, not “filter at zero”. Leaving scope filters unset gives the widest defensible pool; tightening them is how a thin or noisy pool is fixed, and it is the first place a reviewer looks when the range looks wrong.

The jurisdiction changes how the range is struck, not which companies are in it. Curated conventions exist for a small set of jurisdictions — India under Rule 10CA moves to the 35th–65th percentile once the set has six or more entries, with the median as the target; the UAE, USA, UK, Singapore and Germany prescribe the interquartile range. Everywhere else the OECD interquartile default applies, and the wizard’s “apply jurisdiction convention” control only appears where there is a rule to apply.

Statistics need at least two accepted companies; a scope that screens the pool below that produces no range, and the result says so rather than computing one from a single company. Between two and three the percentile band is an interpolation artefact, so no band is published at all: the arithmetic mean is reported as the point estimate instead, the same fallback a statutory mean framework uses.

Why the parameters are stored on the study

• They drive the run. The pipeline reads the parameters block: the PLI determines what is computed, the filters determine who is screened, the percentiles determine what the range is.
• They are the record. The report’s strategy section prints the parameters as configured; the auditor’s first question — “what did the search look for?” — is answered from the study itself.
• They bound the workflow. The parameters stay editable until the study is archived, but re-running the pipeline on a changed set is only possible while the study is in Draft, Study Initialized, Rejected or Failed. Once a study has passed review, a changed configuration is a new study rather than an edit — and archiving freezes the whole record.

The creation flow that collects these fields is in Creating a Study.