The third consecutive year of losses, explained internally by a difficult market and purchase prices set by the group. The accountant sees no problem: the loss is real. For ANAF, the combination of repeated losses and significant intra-group transactions is not an observation, it is a declared selection criterion.
A company can record losses without breaching the arm's length principle. The problem arises when the losses are recurring, the transactions with related parties are significant, and the functional profile described in the transfer pricing file does not explain why this particular entity bears the negative result.
ANAF has publicly stated that selection for transfer pricing inspections is made through risk analysis, and the criteria expressly include companies with a significant volume of intra-group transactions that record recurring losses or reduced profitability compared with comparables. The result of these checks, again according to the data published, has been a reduction in reported tax losses running into billions of lei, a substantial part of it coming from transfer pricing adjustments.
The whole analysis comes down to one question: what risks did the entity assume. A full-risk manufacturer or distributor, one that sets its own prices and carries the stock, credit and market risk, can lose money when those risks materialise, and the loss is the natural consequence of its profile. An entity described in its own file as having limited functions and reduced risk, remunerated through stable and moderate margins, should not, under the logic of risk allocation, bear losses.
This is where the contradiction the inspection exploits arises: the documentation presents a limited-risk distributor remunerated through a predictable margin, while the financial statements show an entity absorbing all of the group's shocks. One of the two descriptions is false. If the functional profile is real, the remuneration does not observe the arm's length principle and must be adjusted. If the remuneration reflects reality, then the entity in fact carries risks the file does not recognise, and the functional analysis is wrong.
The practical consequence is that the defence does not start from explaining the loss, but from checking the consistency between the file and the facts. The contracts, the pricing decisions, who carries the stock and non-payment risk, who decides on investment and campaigns, who bears the returns. A file that describes a profile reality does not confirm becomes, at inspection, evidence against the taxpayer.
Unpaid receivables at the entity that carries the credit risk, impaired stock at the one that carries the stock risk, exchange-rate losses at the one that bears them contractually. This holds up, provided the risk belongs to the entity, not the group, and the materialisation is documented.
Losses from the market-penetration phase are recognised, but within limits: a reasonable duration, a strategy documented at the time, an expected benefit for the entity bearing the cost. A strategy invoked retroactively, in the fifth year, is no longer a strategy.
This does not protect automatically. An external event does not turn a limited-risk entity into one that bears any negative impact; the losses can stay local only if the treatment reflects the economic reality and the actual allocation of risks.
The strongest argument, where it exists: comparable independent companies that recorded losses in the same period and in the same market. It moves the discussion from the entity's profile to the reality of the sector, ground on which the comparability study can be defended.
When the inspection finds that the result does not observe the arm's length principle, the adjustment recalculates the entity's profitability at the market level, usually at the median of the comparison range. The effect on a loss-making company is double and, for that reason, disproportionate: the carried-forward tax loss, which would have reduced tax in later years, is cancelled, and additional corporate income tax is assessed for the period under review, together with late-payment charges (accesorii: interest and penalties).
On top of this comes economic double taxation: the profit attributed to the Romanian subsidiary has already been taxed at the level of the related party in the other state. Eliminating it does not happen automatically; it requires a corresponding adjustment in the other state or, where that state refuses, the mutual agreement procedure. The adjustment mechanism and how the sums are built are dealt with separately, in the analysis on transfer pricing adjustments.
For groups that have left losses in Romania year after year, the exposure is calculated over the whole limitation period, not just the last year. This is where the sums that seem disproportionate to the subsidiary's size come from.
The functional profile in the file against the operational reality: who sets the prices, who carries the stock, who bears non-payment, who funds the campaigns. Contradictions are resolved now, by correcting the file or the contracts, not during the inspection.
Broken down by element: the intra-group purchase price, volumes, fixed costs, impairments, exchange rate, events. Each element is attributed to a risk, and the risk to an entity.
The decisions that produced the loss, taken at the time: the market-entry strategy, the business plan, the group's pricing decisions, the correspondence. One document from the year of the loss is worth ten explanations given now.
Searching, within the same comparability study, for independent companies that lost money in the same period. If the market lost money, the entity has an argument; if only the entity lost money, it has a problem.
The plan showing the return to profitability and the timeframe. A loss with a horizon is a stage; a loss with no horizon is a model for allocating profit outside Romania.
When the analysis shows the remuneration does not support the profile, correcting the pricing policy for the future, and possibly a voluntary adjustment for the past, costs less than an adjustment to the median imposed at inspection.
The complete file, with an updated functional analysis and a section dedicated to explaining the losses. The time limit for submitting it is short; the moment the file is requested decides the balance of the entire inspection.
The reality of the loss is not disputed. What is disputed is that this particular entity bears it, given its functional profile.
Nearly half of the additional tax assessed in inspections of related-party transactions comes from taxpayers who had the file. What matters is what it says and whether reality confirms it.
An explanation with no contemporaneous document does not hold up. The decisions that produced the loss must be documented in the year they were taken.
The carried-forward tax loss is lost too, for the entire limitation period, and the adjusted profit has already been taxed in another state.
Not automatically, but it does mean selection for a check. ANAF has publicly stated that its risk-analysis selection criteria include companies with a significant volume of intra-group transactions that record recurring losses or reduced profitability compared with comparables. A loss is not an irregularity; an unexplained loss, at an entity presented as having limited risk, is.
Because it determines what risks have been allocated to the entity. A full-risk manufacturer or distributor can bear losses when the risks materialise. An entity described in the file as having limited functions and reduced risk, remunerated through stable margins, should not bear losses; if it does, either the profile in the file does not match reality, or the remuneration does not observe the arm's length principle.
Those that correspond to risks actually assumed by the entity and that have materialised: unpaid receivables at whoever carries the credit risk, impaired stock at whoever carries the stock risk, price or exchange-rate fluctuations at whoever bears them contractually. There are also situations of market entry, restructuring or extraordinary events, but these require separate documentation, a limited duration and a plan for returning to profitability.
Not automatically. An external event does not turn a limited-risk entity into one that bears any negative impact. The costs or losses can stay, in whole or in part, at the local level only if that treatment reflects the economic reality and the actual allocation of risks, not merely the group's wish to leave them there.
By recalculating the tax result at the market level of profitability, usually at the median of the comparison range. The effect is double: the carried-forward tax loss is eliminated and additional corporate income tax is assessed. According to the data published by ANAF, the tax losses of the companies checked have been reduced by sums running into billions of lei.
It is documented before the inspection, not during it: the functional analysis showing the risks actually assumed, evidence that they materialised, comparables recording similar losses in the same period, the business plan and the path to profitability. A real loss, documented at the time, is defensible; the same loss, explained retroactively, rarely is.
Informative material, updated on 18 September 2026. It does not constitute legal or tax advice; individual situations must be assessed on their own facts.
Related analyses are grouped under transfer pricing. For the assistance provided to companies, see tax for companies.
An initial discussion checks whether the functional profile in the file withstands confrontation with the operational reality, breaks down the cause of the loss by risk, and sets out what should be documented now and what should be corrected for the future.