CARF and DAC8: Five Tests for Your Reporting Readiness

4 October 2026 · Niko Sievert

A reporting file can be formally valid and still be systematically wrong: a service may have been overlooked when defining the reporting scope, a tax determination about a user may never have been reviewed, or a transaction feed may be missing. Where the same flawed rule is applied repeatedly, the problem extends beyond individual records. Purely technical file validation offers no assurance against this.

The following five tests address precisely these risks. They ask not only whether a file is produced, but also which errors it might conceal. They are proposals for internal controls, not a legally prescribed sequence of tests. They can be repeated whenever products, data pipelines or rules change.

The framework in brief

CARF is the OECD standard for the automatic exchange of information on crypto-assets. In the EU, DAC8 — Directive (EU) 2023/2226 — introduces corresponding due diligence and reporting obligations, principally in Annex VI to the amended Directive 2011/16/EU on administrative cooperation. DAC8 draws on the crypto-asset services covered by the MiCA Regulation and also covers operators without MiCA authorisation. What matters for classification as a reporting provider is whether the services effect exchange transactions for or on behalf of reportable users.

One difference from the basic CRS model is particularly important for a readiness check: the crypto-asset reporting obligation under DAC8 generally also covers reportable users who are tax resident in the same EU Member State as the reporting provider. A reporting population drawn from CRS practice that is limited exclusively to foreign tax residents therefore falls short.

The first DAC8 reports generally concern the 2026 calendar year; some other CARF jurisdictions start later. The national implementation governs the specific obligations and deadlines. In parallel, the CRS has been expanded to include, among other things, certain e-money products and central bank digital currencies. A CRS review does not, however, replace a review of crypto-asset reporting.

Test 1: Define the scope of the provider, service and product

Failure scenario: A wallet gains an integrated swap function. The product goes live, but the documented scope determination remains unchanged. The new service is not included in the data extraction. If it is subject to reporting, the associated reportable transactions are missing from the report.

Why this is systemic: The error arises before the reporting logic itself — in deciding which services feed into reporting at all. Controls that start only with the reporting population already defined inherit this blind spot.

Guiding question: Does the documented scope determination still match what is actually offered — and which entity effects the exchange for whom? Neither the absence of custody nor the label ‘DeFi’ determines the reporting obligation on its own. For products at the boundary, the connection to the CRS must also be clarified.

Test 2: Track user determinations throughout the lifecycle

Failure scenario: A user reports a move abroad; the controlling person of a corporate customer changes. The KYC system records the change, but no review of the tax classification is triggered.

Why this is systemic: A complete KYC file does not establish that the tax classification is correct. If the link to the tax review is missing, the same root cause persists with every comparable change. The procedures for new and pre-existing users must each work as well.

Guiding question: When information is missing, contradictory or changes, where is that detected, who decides, and does the decision actually reach the reporting process? A new user, a pre-existing user and a case involving changed circumstances offer different perspectives on this.

Test 3: Recalculate one user's results independently

Failure scenario: A mapping rule assigns transfers to the wrong transaction type. The file is formally correct and the totals look plausible — but the classification is still wrong.

Why this is systemic: Transaction information is generally aggregated by crypto-asset and category. A total does not expose the classification error. The same rule repeats it for every affected event.

Guiding question: For a reportable user and a specified period, can you start with the original events and, independently of the reporting module, arrive at the same result? If not, does the cause lie in the underlying data, their valuation or their classification? A sound reconciliation answers this question for the portion examined, not for the entire reporting population.

Test 4: Challenge the completeness of the underlying population

Failure scenario: A transaction feed was never connected to the reporting pipeline. The report and the underlying population agree — because the same feed is missing from both.

Why this is systemic: A reconciliation based on the same incomplete data confirms agreement while leaving the gap invisible. Even an error-free calculation therefore does not answer whether all relevant sources have been included.

Guiding question: How do you know, for a selected category and period, that your underlying population is complete — and what supports confirmation from outside the reporting function? Discrepancies should be traced to their cause, not absorbed into a balancing item.

Test 5: Run through the process, including a correction

Failure scenario: An incorrect source record is corrected. The change reaches the source database but is not reflected in the next draft report.

Why this is systemic: Correcting the source does not by itself remove the error from the report. Without reliable downstream processing, errors already identified remain. A technically accepted draft proves neither that the scope is correct nor that the input data are complete.

Guiding question: Is a correction deliberately introduced during a test run properly reflected in the affected processing steps and the newly generated draft — and can you subsequently demonstrate who decided what, and on what basis?

From sampling to targeted improvement

The value of these tests lies in identifying recurring causes behind individual anomalies. A missing feed, an incorrect mapping rule and a broken update process each call for a different response. The tests provide starting points for that investigation without replacing a full compliance review.