Cross-border cooperation and evidence

Every serious cross-border financial crime case eventually collides with the same structural problem: the evidence is somewhere else, and somebody else's law governs how you can get it. Officials who have only worked domestic files consistently underestimate how much this single fact shapes case strategy, which charges are viable, which timeline is realistic, and which cases quietly die not because the evidence doesn't exist but because it could not be obtained lawfully and in time.

Start with the fundamental distinction I drill into every cohort: intelligence is not evidence. The Egmont GroupEgmont GroupThe network of national Financial Intelligence Units enabling secure, rapid FIU-to-FIU exchange of financial intelligence, subject to strict onward-use and consent restrictions., the informal network of 170-plus Financial Intelligence Units founded in 1995, exists to let FIUs exchange financial intelligence with each other rapidly, FIU-to-FIU, without going through a formal mutual legal assistance channel. A South African FIC request to a Mauritian FIU counterpart under the Egmont Secure Web can produce a useful lead — an account number, a beneficial-owner name behind a shell, a pattern match to a known typology — often within days or weeks. But Egmont Principles for Information Exchange explicitly restrict how that intelligence can be used: it is for intelligence and analytical purposes, not as evidence in criminal proceedings, and re-dissemination to a third party (including, often, to the country's own prosecuting authority for direct evidential use) requires the consent of the originating FIU. Practitioners who forget this and try to tender Egmont-derived material directly at trial find it excluded, or find the originating FIU refusing all future cooperation because the condition was breached. The correct discipline is to use Egmont intelligence to build the investigative picture and to identify precisely which underlying records you then need to obtain through a channel that will produce admissible evidence.

That evidential channel, in the overwhelming majority of cases, is mutual legal assistance, a formal request made under a bilateral Mutual Legal Assistance Treaty (MLAT), a multilateral instrument (UNCAC Chapter IV for corruption-linked cases; the Council of Europe Convention on Mutual Assistance in Criminal Matters and its protocols for European partners; the Palermo Convention for transnational organised crime), or, absent any treaty, on the basis of reciprocity or comity under domestic law. The request is made central-authority to central-authority; typically Ministry of Justice to Ministry of Justice, rather than prosecutor to prosecutor, and it must satisfy the requested state's domestic procedural requirements, not merely the requesting state's own standard of relevance. This is the single most common cause of MLAT failure: a request drafted the way a domestic production order would be drafted, without regard to what the requested state's statute actually demands (dual criminalityDual criminalityA common MLAT precondition requiring that the conduct underlying the request be a criminal offence in both the requesting and requested state., a specified minimum penalty threshold, particularised facts rather than a fishing expedition, and — in civil-law systems especially — a formal legal characterisation of the offence under the requesting state's law).

The delay problem is real and it is structural, not incidental. Every credible study of MLAT practice, the World Bank/UNODC StAR Initiative's own practitioner guides prominent among them; records median response times in the twelve-to-eighteen-month range for financial-crime MLATs, with a meaningful tail extending past three years for complex, multi-jurisdictional requests routed through registries in secrecy-friendly intermediary jurisdictions. That delay is driven by a predictable set of causes: the requested state's central authority is under-resourced and processes requests in strict queue order regardless of urgency; the request is technically deficient (wrong legal basis cited, facts insufficiently particularised, dual-criminality analysis missing) and is returned for correction, restarting the clock; the request seeks material subject to bank secrecy or professional-privilege carve-outs in the requested state that require a judicial order rather than administrative production; or the requested state's own domestic proceedings (a linked confiscation action, a competing request from another state) take priority. I tell officials to build MLAT lead-time into the case plan from day one, not as an afterthought once the domestic evidence has run out, because by the time you realise you need the foreign bank records, you may already be inside the limitation period or facing a bail-review deadline the response cannot beat.

Some structural fixes exist and are worth knowing. Joint Investigation Teams (JITs), used extensively within the EU under the 2000 EU Convention on Mutual Assistance and Council Framework Decision 2002/465/JHA, and increasingly by ad hoc arrangement outside the EU (Eurojust has supported JITs with African and Latin American partners in major corruption and trafficking cases), dispense with the request-and-response formality altogether: investigators and prosecutors from both states operate under a single team agreement, sharing evidence directly and in real time, with admissibility questions negotiated up front in the team agreement rather than litigated case by case afterward. Where a JIT is available, it should almost always be preferred over sequential MLAT requests for cases with an ongoing evidence-gathering phase. Spontaneous exchangeSpontaneous exchangeThe unprompted transmission of information by one state's authority to a foreign counterpart, without a prior formal request. is the other underused tool: many treaties and the FATF standards permit — and in some regimes require — a state that comes across information relevant to another state's investigation, without being asked, to transmit it unprompted. FIUs and tax administrations that build a culture of spontaneous exchangeSpontaneous exchangeThe unprompted transmission of information by one state's authority to a foreign counterpart, without a prior formal request. (flagging a suspicious pattern to a counterpart the moment it is noticed, rather than waiting for a formal request) materially shorten the intelligence-to-action cycle across the network, and it costs the sending state very little.

The last practical point concerns sequencing evidence with charge strategy. Because a laundering charge is autonomous of the predicate in most modern statutes, prosecutors facing a slow-moving MLAT for predicate-offence evidence (say, records proving a bribe was paid abroad) can often proceed on the laundering count using domestic evidence of the unexplained pattern of receipt and concealment, while continuing to pursue the foreign evidence for the predicate as a parallel, non-blocking track. This is not a technicality, it is often the only way to keep a prosecution alive within domestic limitation periods when the foreign channel is, realistically, running on its own eighteen-month clock regardless of anyone's urgency.

Tax administrations have their own, largely separate, cooperation architecture from the criminal-justice channels covered in the previous lesson, and confusing the two costs officials real time. The OECD/Council of Europe Convention on Mutual Administrative Assistance in Tax Matters, together with the network of bilateral double-tax agreements and Tax Information Exchange Agreements (TIEAs), provides for Exchange of Information on Request (EOIREOIRExchange of Information on Request; a treaty partner responds to a specific request about a named taxpayer, assessed via Global Forum peer review.): a tax administration with an open enquiry asks a treaty partner for specific, foreseeably relevant information, bank records, ownership information, accounting records — about a named taxpayer. The Global Forum on Transparency and Exchange of Information for Tax Purposes peer-reviews every member's EOIREOIRExchange of Information on Request; a treaty partner responds to a specific request about a named taxpayer, assessed via Global Forum peer review. framework and legal capacity to respond, and its ratings (Compliant, Largely Compliant, Partially Compliant, Non-Compliant) function much like FATF Mutual Evaluation ratings do for AML — as the diagnostic baseline donors and FSRBs use to prioritise technical assistance.

EOIREOIRExchange of Information on Request; a treaty partner responds to a specific request about a named taxpayer, assessed via Global Forum peer review. is reactive: you must already know, or strongly suspect, that a particular taxpayer holds assets or income in a particular partner jurisdiction before you can ask. The single biggest structural advance of the last fifteen years was to complement EOIREOIRExchange of Information on Request; a treaty partner responds to a specific request about a named taxpayer, assessed via Global Forum peer review. with Automatic Exchange of Information (AEOI) under the OECD Common Reporting Standard (CRS), agreed in 2014 and now operative among more than 120 jurisdictions. Under CRS, financial institutions in a participating jurisdiction identify accounts held by non-resident individuals and entities and report account balances, income and (since a later refinement) certain gross proceeds to their domestic tax authority, which then transmits that data automatically, on a bulk annual basis, to the account-holder's state of tax residence, without any prior request. CRS converted offshore account-holding from a fact a tax administration had to discover through investigation into a fact that, for CRS-reportable accounts, now arrives on a data feed every September. For a revenue authority in a developing country that has signed the Multilateral Competent Authority Agreement (MCAA) on CRS and built the receiving IT infrastructure, this is transformative; it turns risk-based case selection from guesswork into a data-matching exercise against filed tax returns.

CRS, however, has a well-known and deliberately engineered gap: it covers traditional financial accounts (bank deposits, custodial accounts, certain insurance and investment products) but not crypto-assets held outside a traditional financial institution, and it has known loopholes around certain investment entities and residence-by-investment arrangements that the OECD itself has flagged and periodically tightened. The OECD's response is the Crypto-Asset Reporting Framework (CARFCARFCrypto-Asset Reporting Framework, the OECD's 2022/23 automatic-exchange standard for crypto-asset transactions, first exchanges scheduled for 2027.), finalised in 2022 and formally adopted into the CRS/DAC framework package in 2023, requiring crypto-asset service providers, exchanges, certain wallet providers, and OTC brokers — to report transactions in crypto-assets on the same automatic-exchange model as CRS covers traditional accounts. First exchanges under CARFCARFCrypto-Asset Reporting Framework, the OECD's 2022/23 automatic-exchange standard for crypto-asset transactions, first exchanges scheduled for 2027. are scheduled for 2027, tracking calendar-2026 data in early-adopting jurisdictions, and the framework is being incorporated into EU law via DAC8, which amends the EU's existing Directive on Administrative Cooperation to require crypto-asset reporting on a comparable timetable. Officials advising ministries or drafting domestic legislation now should treat CARFCARFCrypto-Asset Reporting Framework, the OECD's 2022/23 automatic-exchange standard for crypto-asset transactions, first exchanges scheduled for 2027. implementation — the domestic legal basis compelling crypto-asset service providers to collect and report the data, and the IT capacity to receive and match it, as the most consequential AEOI development of this decade, on a par with CRS itself when it launched in 2014-2017.

The exchange-of-information toolkit does not stop at request-and-response and automatic exchange. Spontaneous exchange under Article 6 of the Multilateral Convention lets a tax administration transmit information it believes will be of interest to a treaty partner even absent a request; for example, where an audit uncovers evidence that a taxpayer's counterparty in another jurisdiction appears to be under-reporting. Simultaneous tax examinations allow two or more administrations to examine, in parallel and on a coordinated timetable, the tax affairs of a taxpayer or corporate group operating across their borders, comparing findings as the audits proceed rather than after the fact, this is the standard tool for cracking a Zambian mining group's transfer-pricing structure where both the Zambian revenue authority and a counterpart jurisdiction's authority hold half the picture. Joint audits go a step further, with officials from both administrations physically or virtually working the same file together, agreeing facts and even a single joint assessment position, an approach ATAF has actively promoted among African revenue authorities and which the OECD's Forum on Tax Administration has formalised in joint-audit guidance.

The ATAF/OECD Tax Inspectors Without Borders (TIWB) programme deserves particular attention for a developing-country audience because it is capacity-building rather than merely cooperative: experienced auditors are seconded, on a cost-shared basis, to work directly alongside domestic auditors on live, real transfer-pricing and international tax cases, transferring skills in the course of actual casework rather than through classroom training. TIWB's own published results record billions of dollars in additional tax assessed across partner administrations since the programme's 2015 launch, and several African revenue authorities — including ZRA in Zambia and the Ghana Revenue Authority — have used TIWB placements specifically to build in-house transfer-pricing audit capacity on extractive-sector transactions. My own advice to any revenue authority weighing where to invest scarce international-cooperation capacity is to build EOIREOIRExchange of Information on Request; a treaty partner responds to a specific request about a named taxpayer, assessed via Global Forum peer review. and CRS-matching competence first, because that infrastructure is prerequisite to using TIWB placements and simultaneous examinations productively, a joint audit is only as good as the domestic team's ability to independently verify what the foreign data reveals.

This lesson is deliberately practical, because the gap between the law of international cooperation and its actual practice is almost entirely a drafting-and-relationship problem. Officials in developing-country administrations frequently tell me the same story: they sent a request, months passed, nothing came back, and they concluded the counterpart jurisdiction was uncooperative. In my experience reviewing these files, the more common explanation is that the request itself was deficient in one of a small number of predictable ways, and the requested authority, under-resourced, working through a queue, with no incentive to chase clarification; simply let it sit rather than reply asking for a fix.

The first discipline is to write the request for the reader, not for yourself. A central authorityCentral authorityThe designated national body (typically within a Ministry of Justice) that receives, vets and transmits formal mutual legal assistance requests. officer in the requested state, who has never heard of your case and does not share your domestic legal vocabulary, must be able to read the request and immediately answer three questions: what specific information or evidence is sought; why it is relevant to an identified offence under the requesting state's law, and whether the requested state's own law permits it to be produced (including whether the conduct described would be an offence there too, if dual criminalityDual criminalityThe requirement, in most MLA regimes, that the conduct underlying the request also constitute an offence under the requested state's law. applies). Requests that assume the reader already understands the case, or that describe the underlying scheme in narrative prose without extracting the specific facts that satisfy each legal element, are probably the most common cause of a request being returned for clarification rather than actioned, and every round of clarification restarts months of delay.

Particularise, do not narrate. State precisely which accounts, which account-holder names and any known aliases, which date ranges, which document types (account-opening documentation, signatory records, transaction ledgers for a specified period, correspondent-bank SWIFT records) are sought, and tie each item explicitly to a fact you need it to prove. A request asking broadly for "all banking records relating to Mr X" invites refusal on fishing-expedition grounds in almost every requested-state's domestic law; a request asking for "the account-opening file and transaction ledger for account ending 4471 at Bank Y for the period 1 January 2023 to 30 June 2024, relevant to establishing the receipt and onward transfer of funds alleged to be proceeds of the bribery offence particularised in paragraph 4" will very rarely be refused on that ground.

Address dual criminalityDual criminalityThe requirement, in most MLA regimes, that the conduct underlying the request also constitute an offence under the requested state's law. explicitly rather than leaving it to be inferred. Set out, side by side, the elements of the requesting state's offence and a plausible equivalent offence under the requested state's law (most central authorities can identify likely equivalents from published guidance or prior cooperation, and where in doubt this is worth a preliminary informal enquiry before the formal request is even sent). Where the requested state's law imposes a minimum-penalty threshold for MLA eligibility, confirm and state that the alleged offence clears it.

Use the informal pre-request channel. Almost every functioning MLA relationship allows an informal enquiry — an email or call between investigators, or between liaison magistrates/prosecutors where such postings exist — to test feasibility, confirm the correct legal basis and central-authority contact point, and flag any procedural quirks (a requirement for the request to be in the requested state's official language, a requirement for supporting documents to be apostilled or otherwise authenticated, a preference for a particular template) before the formal request is drafted. Officials who skip this step and simply fire a formal request into the central-authority mailbox lose, on average, one full clarification cycle compared with those who make the informal call first, in a system where a single clarification cycle can cost three to six months, that call is not optional courtesy, it is case management.

Sequence urgency correctly. Where assets are at risk of dissipation, most MLA frameworks and many domestic statutes allow a parallel, expedited provisional-measures requestProvisional-measures requestAn expedited parallel request seeking to freeze or preserve assets pending the outcome of the substantive evidence request. (freezing or preservation) alongside the substantive evidence request; do not wait for the full evidentiary package to be assembled before asking the requested state to freeze the account, because by the time the substantive request is answered the funds may be gone. Clearly label the request as urgent, explain concretely why (a specific, imminent transaction or maturity date, not a generic urgency claim), and, where the relationship allows it, follow up the written request with a direct call to the named contact, repeatedly, respectfully, and on a fixed cadence, because silence in these systems usually means the file is in a queue, not that it has been refused.

Build and maintain the relationship independent of any single case. The FIUs, central authorities and revenue administrations that get the fastest responses are, almost without exception, the ones that show up to ESAAMLG, Egmont, ATAF and Global Forum meetings consistently, second staff to regional and international placements, and respond promptly and well to other states' incoming requests. Reciprocity in this system is informal but real: a central authorityCentral authorityThe designated national body (typically within a Ministry of Justice) that receives, vets and transmits formal mutual legal assistance requests. that has a backlog will, when forced to prioritise, prioritise the counterpart that has itself been a reliable and responsive partner. I tell every cohort the same thing on this point: your response time to other people's requests is, in a very real sense, an investment in your own future response times, and treating incoming MLA requests as a low-priority administrative chore is a strategic error with a delayed but very real cost.