Loss-Making Comparables: Keep, Adjust or Exclude?
The framework for loss-making companies in the comparable pool: temporary versus structural losses, the matching rule, missing data, and the documentation the TPO expects.
A comparable company in loss does not compute a PLI — there is no operating profit to divide. The question is what the loss means, because the answer decides the treatment: a company temporarily in loss (start-up ramp, restructuring year, one-off event) is not the same comparable as a company structurally in loss (a business model that does not cover its costs). The Guidelines’ default is the conservative one — a loss-making comparable is excluded unless the tested party is also loss-making in the same cycle — and the framework below is how the exception is argued and documented.
The three categories
| Category | What it is | Default treatment |
|---|---|---|
| Temporary loss | Start-up ramp, a restructuring year, a one-off extraordinary event; the business is expected to reach steady state | Excluded for the loss year; retained where multi-year data shows the steady-state PLI (the loss year documented as the reason for the multi-year view) |
| Structural loss | The business model does not generate an operating profit — the company persists in loss without a ramp to profitability | Excluded — it is not a comparable for a profitable tested party, full stop |
| Tested-party-matched loss | The tested party is loss-making in the same period for the same economic reason | The loss years can be used — but the analysis is on costs and the path to profitability, not on a PLI range; see below |
The categorization is a finding with evidence, not a label: the annual report narrative, the revenue trajectory across the period, the nature of the loss (one-off charge vs operating deficit), and the management commentary together decide the category. “The company is in loss” is not the analysis; “the company is in loss because of X, and X is temporary/structural” is.
The matching rule, applied
The default exclusion exists because a profitable tested party tested against a pool of loss-makers has no range — and a pool from which the loss-makers were silently dropped has a distribution the file cannot explain. The discipline:
- State the count. Of N candidates, M were loss-making in the period. That line belongs in the screening record.
- Categorize each. Temporary or structural, with the evidence for the call.
- Apply the treatment. Temporary → excluded for the year (or retained via multi-year where the period view is the declared method); structural → excluded.
- Show the pool before and after. The distribution with and without the loss-makers, so the effect of the exclusions is visible. A pool that moves materially when the loss-makers come out is a pool that was leaning on them — the file should address that, not hide it.
When the tested party is loss-making
The matching case is the genuine exception, and it is rarer than it looks — most “loss-making tested party” files are loss-making for reasons the pool is not (a one-off, a start-up year, a contract loss):
- Same reason, same cycle. Where the tested party and the comparables are loss-making in the same period for the same economic reason (an industry in distress, a shared start-up phase), the comparison moves off the PLI range: the analysis is on the cost structure, the cost trajectory and the path to break-even, with the comparable set providing the cost base and the cost behaviour. This is a documented departure from the standard range analysis — stated, reasoned, and consistent with the multi-year view.
- Different reason. Where the tested party’s loss is idiosyncratic (its own one-off, its own contract loss), the comparables are the profitable pool, the tested party is outside any range, and the file’s job is the explanation of the deviation — the one-off, the timing, the expected recovery — not a range the tested party cannot be in.
Missing data is not a loss
A different failure mode wears the same clothes: a comparable whose profitability data is missing (segment data absent, the operating line not derivable from the filing). Missing data is not a loss — it is an inability to test, and the treatment is exclusion with the reason “PLI not computable from available data” — which is a defensible, common, and frequently large part of any honest screening. The distinction matters in the matrix: “loss-making” and “data not computable” are different reasons, and a matrix that blends them invites the question of what was actually seen.
The documentation
The screening record carries, per loss-making candidate: the category (temporary / structural / matched), the evidence (the trajectory, the narrative, the one-off identification), the treatment, and the pool’s before/after distribution. In the Local File the loss-maker treatment is a line in the screening summary — the count, the categorization, the effect on the range — because it is one of the choices the TPO re-runs when it reconstructs the pool.
See also
Run the screens as a study, not a spreadsheet
Quartyl applies the method, PLI and screening steps above as a pipeline — and keeps a documented reason for every exclusion.
Related docs
Multi-Year Averaging in TNMM: Simple, Weighted and Period Methods
Single-year vs multi-year data in TNMM: when the OECD requires looking beyond one year, how simple and weighted averages are computed, and when averaging masks a trend.
Read docQualitative Screening: The Company-by-Company Review
How to run the qualitative comparable review: the five things to verify per company, the evidence hierarchy, and accept/reject notes that survive a transfer pricing officer.
Read doc