Executive answer
The transactional net margin method compares the net operating margin earned by one party in a controlled transaction with that of comparable independent parties, using a base such as sales, costs or assets. Article 180, section VI, of the Mexican Income Tax Law recognizes the method.
Frequent use does not make TNMM the default. CUP and other methods should be assessed first. Reliability depends on four decisions: tested party, profit-level indicator, segmentation and comparables. Selecting any of them for convenience can produce a statistically correct range applied to the wrong economics.
The tested party is generally the party to which the method can be applied most reliably and that lacks unique intangibles or difficult-to-compare risks. It may be Mexico or the foreign counterparty. The indicator should fit the function: sales for distribution, costs for manufacturing or services, and assets for asset-intensive activity, subject to facts and data.
TNMM logic
TNMM tests a transactional net result. It tolerates some product differences that would undermine CUP, but still requires functional, accounting and economic comparability. Operating expenses may reflect functions, efficiency, risk or classification differences.
Illustration: a distributor has sales of 100, cost of goods of 70 and operating expenses of 25, producing operating profit of 5 and ROS of 5%. If the comparable range is 3%–7%, the result may fall inside, subject to quality. If the segment includes services or extraordinary cost, 5% may not test distribution.
Do not confuse net margin with taxable income. Define operating items and reconcile to financial statements and returns.
Selecting the tested party
Compare both parties by:
- functional complexity;
- unique intangibles;
- controlled risks;
- available segmented data;
- reliable comparables;
- accounting consistency;
- ability to verify results.
The less complex party is not always smaller. An entity with few employees may own technology; a larger entity may perform routine processes. Residence does not decide.
If the foreign counterparty is selected, obtain sufficient information. Do not accept aggregated numbers from the group without reconciliation. Document currency, accounting, period and segment.
Tested-party matrix
| Criterion | Mexico | Counterparty | Implication |
|---|---|---|---|
| Unique functions | Prefer lower complexity | ||
| Intangibles | Avoid unique contributions | ||
| Risks | Prove control | ||
| Segmentation | Auditable data | ||
| Comparables | Available market | ||
| Accounting | Consistent definitions | ||
| Currency/period | Verifiable bridge |
Record evidence and conclusion. Do not automate a score; the matrix structures judgment.
Profit-level indicators
Return on sales
Operating profit over sales. It often fits distribution when sales reflect the function. Pass-throughs, mix or unrelated revenue may distort it.
Markup on costs
Operating profit over relevant costs. It may fit services or manufacturing. Clean and compare the base; define total, operating or value-added costs.
Return on assets
Profit over operating assets may fit functions in which assets drive value. Valuation, utilization and consistency matter; unrecorded intangibles may distort it.
Berry ratio
Gross profit over operating expenses may be considered for certain intermediaries where expenses reflect value and no additional functions exist. It is not universal.
Other indicators
Every indicator needs an economic connection. Do not select the one that puts the result inside the range.
Selecting the denominator
The denominator should be reasonably independent of controlled pricing, relate to the function and be measured consistently. Where related costs determine both base and consideration, circularity may arise.
For ROS, remove unrelated controlled revenue, taxes and returns. For cost indicators, define categories, pass-throughs and extraordinary items. For assets, address averages, inventory, receivables and idle capacity.
Test sensitivity. If a small reclassification changes the result materially, elevate risk and compare indicators.
Transactional segmentation
Apply TNMM to the transaction or closely linked set. An entity may manufacture, distribute, provide services and finance. Entity-wide testing can offset an out-of-range segment with an inside-range segment.
Build a segmented statement covering income, cost, expense, assets and liabilities. Use direct tracing before keys. Document causation and benefit. Reconcile to statutory statements.
Preserve gross amounts by type for filings even where the analysis aggregates. Aggregation needs an economic relationship.
Segmentation bridge
| Item | Statutory | Direct | Key | Segment | Control |
|---|---|---|---|---|---|
| Sales | Invoices | ||||
| Product cost | Inventory | ||||
| Personnel | Hours | Payroll | |||
| Marketing | Benefit | Campaigns | |||
| Logistics | Volume | Orders | |||
| Administration | Driver | Centers | |||
| Extraordinary | Decision | Approval | |||
| Operating profit | Recalculate |
Preserve version and owner.
If TNMM is applied to full-entity statements without supporting the tested party, indicator and segment, redesign the test before relying on the range.
Comparable search
Define FAR, market, years, codes and screens before results. Review activity, related parties, losses, intangibles, information and events. Do not rely on a short description.
Calculate the same indicator for all. Align items where data permit; reject when an accounting difference is material and unadjustable. Document working-capital and other adjustments.
Sample size does not replace quality. Retain universe, rejections, versions and sensitivity. Refresh independence and results annually under policy.
Working capital
Receivable, inventory and payable differences may affect margin. An adjustment can improve comparability where account definitions, periods and rates are consistent.
Document formula, rate, sign and averaging. Do not adjust away real FAR risks or double-count price effects. Show before and after.
Where public data lack segmentation, disclose the limitation. A mechanical adjustment does not guarantee improvement.
Result and range
Calculate at full precision, order comparables and apply Article 180 statistics. Avoid early rounding. Distinguish annual and multiple-year ranges.
An inside-range result does not prove materiality, benefit, deduction or treaty access. An outside result requires recalculation, reconciliation, FAR review and cause analysis before adjustment.
Connect the result with policy. If the target is ROS, translate it into a purchase or service price without circularity.
Distributor scenario
Mexico distributes products, makes routine sales decisions and owns no brand. It may be the tested party with ROS where segmented data exist. Compare marketing, inventory, credit and warranty.
If it uniquely develops the market or controls strategic price, the profile may not be routine. Do not apply a standard ROS without recognizing contribution.
Separate losses by launch, obsolescence and decisions. Determine risk control.
Manufacturing or service scenario
A contract manufacturer may use markup on costs where the base represents its function. Compare capacity, materials, assets and quality. A service center also needs benefit, catalog and cost evidence.
Pass-through items can distort the denominator. Determine treatment from facts and comparables. Idle capacity and extraordinary costs require analysis.
Compare TNMM with cost plus. A net indicator may tolerate gross differences, but not an incomplete base.
Testing a foreign party
Request segmented statements, reconciliation, accounts, currency, method, comparables and functional evidence. Verify that the result excludes other jurisdictions or dealings.
Define access and retention. A conclusion dependent on data unavailable in Mexico is weak. Agree on a global calendar.
Assess whether foreign-market comparables are more reliable and whether economic circumstances differ. Document the judgment.
Operational monitoring
Budget the indicator and drivers. Compare actual, forecast and range monthly. Separate price, volume, mix, cost, currency and extraordinary variance.
Define thresholds and owners. A dashboard does not replace the report, but detects deviations. Update in-year prices if agreement and rules allow.
At year-end, document true-up, taxes, counterparty and recalculated indicator. Fix the cause to avoid repetition.
Common errors
- Selecting Mexico automatically.
- Using the full entity without segmentation.
- Choosing the indicator by outcome.
- Mixing accounting definitions.
- Including pass-throughs without analysis.
- Ignoring tested-party intangibles.
- Selecting comparables by industry rather than function.
- Applying working-capital adjustments mechanically.
- Hiding losses through averages.
- Failing to translate the method into policy.
Quality control
The TNMM Design Review delivers delineation, tested-party matrix, indicator, segment, search, adjustments, range and monitoring. An independent reviewer reproduces calculations and bridges.
Tax approves method; operations FAR; accounting segmentation; leadership changes; global team foreign data. Preserve versions and limitations.
Design record and refresh triggers
Prepare a concise memorandum for every segment documenting the dealing, tested party, alternatives, indicator, denominator, period, comparables, adjustments and result. Link each decision to evidence and a reviewer. Distinguish unavailable data from unreliable data: the first may be a control gap; the second a technical limitation.
Define triggers to reopen the design: new line, intangible acquisition, decision-right change, personnel migration, ERP change, persistent loss, new internal comparable or restructuring. Refreshing numbers is insufficient when FAR changes.
Compare the tested party and indicator annually even if they remain the same. Documenting stability is preferable to assuming it. When either changes, prepare a quantitative and operating bridge and set an effective date without retrospectively selecting the design that produces the best outcome.
Auditability test
Take one range observation and reproduce its source, screens and indicator. Take the Mexican result and trace revenue, cost, allocations and adjustments to the trial balance. Then select a dealing from the subledger and confirm that it enters the segment. These tests address accuracy and completeness.
Ask a reviewer outside the preparation team to recalculate profit from the available file. Record manual dependencies, broken formulas and global access needs. Response time shows whether the model can withstand an information request.
Related topics
- PT-017 through PT-020: FAR, comparability, method and range.
- PT-022 and PT-023: resale price and cost plus.
- PT-079: limited-risk models.
TNMM design package
The package should explain why the tested party is less complex, has reliable data and lacks contributions preventing comparison. Evaluate both parties before choosing. Then connect the indicator to the activity: sales for distribution, costs or assets where they better reflect functions and risks. Do not select a PLI by habit.
Prepare segmented statements, allocation rules and reconciliation to the ledger. Identify operating, non-operating, exceptional and pass-through items consistently with comparables. Test sensitivity to working capital, periods, filters and reclassifications. An accounting difference may move the range more than the market does.
Retain the search, rejections, calculations, review and approval. Operations validates facts and accounting validates segmentation. The final output should reproduce numerator, denominator and comparable set while explaining why alternatives produced less reliable results.
Sources and verification date
- Mexican Income Tax Law, current text, Articles 179 and 180.
- OECD Mexico Transfer Pricing Country Profile.
- OECD Transfer Pricing Guidelines.
Sources checked on August 2, 2026. Reliability depends on the tested party, indicator, segment, comparables and data.
Request a TNMM Design Review to validate the tested party, indicator, segment, comparables, range and operating translation.