Sanctions Screening

Overview

Sanctions screening is the mandatory compliance process of checking all payment parties — individuals, entities, and financial institutions — against government-issued and international watchlists. If a match is found, the payment must be blocked, the account may need to be frozen, and the incident reported to the regulator.

Sanctions are geopolitical tools. Governments use them to restrict economic activity with certain countries, regimes, terrorist organisations, weapons proliferators, and corrupt individuals. For banks, processing a sanctioned transaction — even unknowingly — can result in catastrophic penalties.

Key principle: Sanctions are not optional and not risk-based. If a name is on the list, the payment must stop — no exceptions.

---

Why Sanctions Screening Matters

Real-World Penalty Examples

Bank	Year	Penalty	Reason
BNP Paribas	2014	USD 8.97B	Processing transactions for Sudan, Iran, Cuba
Standard Chartered	2019	USD 1.1B	Iran, Myanmar, Zimbabwe transaction processing
HSBC	2012	USD 1.9B	Mexican drug cartel + sanctions violations
Commerzbank	2015	USD 1.45B	Iran, Sudan payments stripped of identifying info
Westpac	2020	AUD 1.3B	AML/CTF breaches including AUSTRAC violations

Consequences of Non-Compliance

Consequence	Details
Massive fines	Can exceed the bank's annual profit
Criminal prosecution	Senior executives personally liable
Reputational damage	Loss of correspondent banking relationships — effectively cuts off international payments
SWIFT disconnection	Exclusion from the global interbank network
Banking licence revocation	Regulator can shut the bank down
Deferred Prosecution Agreements	Bank placed under US government supervision for years

---

Types of Sanctions

1. Comprehensive Sanctions (Country-Level)

Broad prohibition on all economic activity with a country or regime.

Country / Regime	Sanctioning Body	Scope
Iran	OFAC (US), EU, UN, DFAT (AU)	Near-total ban on transactions
North Korea (DPRK)	OFAC, UN, EU, DFAT	Comprehensive embargo
Syria	OFAC, EU, DFAT	Broad restrictions
Cuba	OFAC	US-specific embargo; others vary
Russia (post-2022)	EU, OFAC, DFAT, UK	Sectoral + targeted
Belarus	EU, UK, OFAC	Senior regime figures + entities
Myanmar	OFAC, EU, DFAT	Military junta + entities

⚠️ Extraterritorial Reach of OFAC: Even a non-US bank can be penalised by OFAC if transactions touch the US financial system (USD clearing, US correspondent banks). This makes OFAC screening globally mandatory.

2. Targeted / Individual Sanctions

Prohibitions on specific named individuals and entities, regardless of their country.

• Specially Designated Nationals (SDN) — OFAC's primary list; includes terrorists, drug traffickers, WMD proliferators, and corrupt oligarchs
• Consolidated UN List — UN-designated individuals and entities
• EU Consolidated List — EU-designated individuals and entities
• DFAT List (AU) — Australia's autonomous sanctions on individuals and entities

3. Sectoral Sanctions

Restrictions on specific sectors of an economy rather than blanket bans — common for Russia sanctions:

Sectoral Sanctions examples (Russia):
├── Energy sector  — Restrictions on financing for oil projects
├── Defence sector — No finance/export of military equipment
├── Finance sector — Restrictions on major Russian banks (debt/equity financing)
└── Technology      — Export controls on dual-use technology

4. Secondary Sanctions

US secondary sanctions penalise non-US entities that do business with sanctioned parties — even if no US person or USD is involved:

Iranian company  ────────────────►  German company (non-US)
(sanctioned)                             │
                                         │ US secondary sanctions risk
                                         ▼
                              German company could be cut off
                              from the US financial system

This is why banks globally — not just US banks — screen against OFAC lists.

---

Global Sanctions Lists

Primary Lists (Must Screen Against)

List	Authority	Notes
OFAC SDN	US Treasury Office of Foreign Assets Control	~13,000+ entries; most powerful list globally
OFAC Non-SDN / SSI	US Treasury	Sectoral Sanctions Identifications (SSI) list
UN Security Council Consolidated	United Nations	Mandatory for all UN member states
EU Consolidated Sanctions	European Council	Must screen for EU-connected transactions
UK OFSI Consolidated	HM Treasury Office of Financial Sanctions Implementation	Post-Brexit, UK maintains its own list
DFAT Consolidated	Australian Dept of Foreign Affairs and Trade	Australia's autonomous sanctions
AUSTRAC Guidance	AUSTRAC	AML/CTF risk guidance (supplements DFAT)

Secondary / Supplementary Lists

List	Source	Usage
Interpol Red Notices	Interpol	Wanted persons (not legally binding)
World Bank Debarred	World Bank	Debarred entities from WB projects
FATF High-Risk Countries	FATF	Enhanced due diligence jurisdictions
Transparency International CPI	TI	Corruption Perception Index — risk indicator
Internal Bank Blacklists	Each bank	Internal watchlists from investigations

---

SDN List — Structure Deep Dive

The OFAC SDN list entry contains:

SDN Entry Example:
├── Name:           ALI HASSAN MOHAMMED
├── Aliases:        HASSAN, Ali; MOHAMMED ALI, Hassan
├── Type:           Individual
├── Program:        SDGT (Specially Designated Global Terrorist)
├── DOB:            01 Jan 1975
├── POB:            Yemen
├── Nationality:    Yemeni
├── Passport:       YE123456 (issued Yemen)
├── Address:        Sana'a, Yemen
└── Additional info: Associated with [terrorist org]

Associated entity:
├── Name:           AL-HASSAN TRADING CO.
├── Type:           Entity
├── Program:        SDGT
├── Address:        Sana'a, Yemen
└── ID:             YE-REG-12345

Banks must screen against all fields — name, alias, DOB, address, passport number.

---

What Must Be Screened

In Every Payment

pacs.008 / pain.001 Payment Instruction
├── Debtor
│   ├── Name (and all name variants)
│   ├── Address / country
│   ├── Date of birth (individuals)
│   └── Registration number (entities)
├── Debtor Account (IBAN, account number)
├── Debtor Agent (BIC → country extraction)
├── Instructing Party (if different from debtor)
├── Ultimate Debtor (if present)
├── Intermediary Agent(s) (correspondent banks)
├── Creditor Agent (BIC → country extraction)
├── Creditor
│   ├── Name
│   ├── Address / country
│   └── Registration number
├── Creditor Account
├── Ultimate Creditor (if present)
└── Remittance Information
    └── Scan for entity names embedded in free text

At Customer Onboarding

• Full name + aliases
• Beneficial owners (≥ 25% ownership — see UBO rules in AML)
• Directors and key controllers
• Connected entities and subsidiaries

On List Updates

When OFAC/UN/DFAT publish a new list version, every existing customer in the bank must be re-screened against the new entries.

---

Sanctions Screening Architecture

Payment Instruction
        │
        ▼
┌───────────────────────────────────────────────────────┐
│              SANCTIONS SCREENING ENGINE                │
│                                                        │
│  1. DATA EXTRACTION                                    │
│     Parse all party fields from ISO 20022 message     │
│                                                        │
│  2. NORMALISATION                                      │
│     ├── Remove punctuation: "AL-RASHID" → "AL RASHID" │
│     ├── Expand abbreviations: "Co." → "Company"       │
│     ├── Transliterate: Arabic/Cyrillic → Latin        │
│     ├── Lowercase + trim                              │
│     └── Split compound names                          │
│                                                        │
│  3. MATCHING ENGINE                                    │
│     ├── Exact match (account numbers, BICs, LEIs)     │
│     ├── Fuzzy name match (Jaro-Winkler, Levenshtein)  │
│     ├── Phonetic match (Soundex, Double Metaphone)    │
│     ├── Token / n-gram match                          │
│     ├── Alias expansion (check all known aliases)     │
│     └── Country code match (for embargoed countries)  │
│                                                        │
│  4. SCORING & THRESHOLDING                             │
│     ├── Score 0–100 per candidate match               │
│     ├── < threshold → CLEAR                           │
│     └── ≥ threshold → ALERT                          │
└────────────────────────┬──────────────────────────────┘
                         │
              ┌──────────┴──────────┐
              │                     │
           CLEAR               POTENTIAL MATCH
              │                     │
         Process               Auto-Hold Payment
         Payment               Notify Compliance
                                    │
                          ┌─────────┴──────────┐
                          │                    │
                    FALSE POSITIVE        TRUE MATCH
                          │                    │
                    Document            BLOCK permanently
                    Release             Freeze account
                    Tune rule           File with AUSTRAC/OFAC

---

Name Matching Algorithms

Jaro-Winkler Similarity

Best for short strings (names). Gives extra weight to matches at the start of the string.

// Jaro-Winkler example
double score = jaroWinkler.apply("Mohammed Ali", "Mohammad Ali");
// Returns ~0.944 — high confidence match

// Threshold typically set at 0.85 for sanctions
boolean isAlert = score >= 0.85;

Levenshtein Distance

Counts minimum edits (insert, delete, substitute) to transform one string to another.

int distance = levenshtein.apply("HASSAN", "HASAAN");
// Returns 1 (one substitution) — likely a match

Double Metaphone (Phonetic)

Encodes names by how they sound, not how they're spelled.

String[] codes = metaphone.doubleMetaphone("Mohammed");
// Returns ["MHM", "MMT"] — same codes as "Mohamed", "Muhammed"
// Catches transliteration variants

Token Set Ratio

Handles names in different word order.

// "Ali Hassan Mohammed" vs "Mohammed Ali Hassan"
// Standard fuzzy: low score (different order)
// Token set ratio: high score (same tokens, different order)
double score = tokenSetRatio("Ali Hassan Mohammed", "Mohammed Ali Hassan");
// Returns 1.0 — same tokens

Match Threshold Strategy

Threshold	Behaviour	Trade-off
100% (exact only)	Minimal false positives	Easily evaded by minor spelling variants
85–90%	Balanced	Industry standard for high-volume screening
70–80%	More alerts	High false positive rate; analyst workload ↑
60–70%	Very sensitive	Unusable false positive rate

Banks typically run multiple thresholds — 85% for auto-hold, 70% for audit log only.

---

Country Risk and Embargo Screening

Beyond name matching, the country in a payment must be checked:

public class CountryRiskScreeningService {

    public CountryRiskResult screen(String bicOrIban) {
        String countryCode = extractCountry(bicOrIban);  // e.g., "IR" from BKIRIRIA

        if (comprehensiveEmbargo.contains(countryCode)) {
            // e.g., IR (Iran), KP (North Korea), SY (Syria)
            return CountryRiskResult.block("Comprehensive embargo: " + countryCode);
        }

        if (sectoralSanctions.contains(countryCode)) {
            // e.g., RU (Russia) — depends on transaction type
            return CountryRiskResult.review("Sectoral sanctions apply: " + countryCode);
        }

        if (fatfHighRisk.contains(countryCode)) {
            // FATF grey/black list countries
            return CountryRiskResult.enhancedDueDiligence(countryCode);
        }

        return CountryRiskResult.clear(countryCode);
    }

    private String extractCountry(String bic) {
        // BIC format: BBBB-CC-LL-BBB (CC = country)
        return bic.length() >= 6 ? bic.substring(4, 6).toUpperCase() : "XX";
    }
}

High-Risk / Embargoed Country Categories

Category	Examples	Treatment
Comprehensive embargo	Iran, DPRK, Syria, Cuba (US)	Block all transactions
Sectoral sanctions	Russia, Belarus	Block based on transaction type/sector
FATF Black List	Iran, DPRK	Enhanced due diligence + report to regulator
FATF Grey List	Pakistan, UAE (past), Philippines	Increased monitoring
High-risk DFAT	Countries on DFAT concern list	Enhanced scrutiny

---

False Positive Management

False positives are the biggest operational challenge in sanctions screening. A bank processing millions of payments per day may generate thousands of false alerts daily.

False Positive Rate Benchmarks

Industry	Typical FP Rate	Target
Retail banking	95–99% of alerts are false positives	< 95%
Correspondent banking	90–97%	< 90%
Trade finance	85–95%	< 85%

Reducing False Positives

Techniques:
├── Whitelisting — Known good entities (e.g., "Shell" the oil company vs "Shell" the person)
├── Rule tuning — Adjust thresholds per list / name type
├── Context scoring — Weight by address, DOB, account number match
├── Entity disambiguation — Resolve common names using LEI/registration numbers
├── Suppression rules — Suppress alert if same entity already reviewed + cleared
└── Post-review feedback loop — Analyst decisions train the model

Analyst Workflow for False Positive Disposition

Alert Received
     │
     ▼
Analyst Reviews:
  ├── Does nationality/address/DOB match?
  ├── Is this a common name? (e.g., "Mohammed Ali" — millions worldwide)
  ├── Additional identifiers available? (passport, LEI, registration)
  ├── Context: what is the transaction for?
  └── Has this entity been cleared before?
     │
     ├── CLEARLY DIFFERENT PERSON → False Positive
     │     Document: name, why cleared, analyst ID, timestamp
     │     Add to suppression list (if safe to do so)
     │
     ├── UNCERTAIN → Escalate to Senior Analyst
     │     Request additional customer information
     │     Apply de-risking measures
     │
     └── CONFIRMED MATCH → True Hit
           Escalate to Compliance Manager
           Block payment / freeze account
           File with AUSTRAC/OFAC

---

Watchlist Maintenance and List Updates

Sanctions lists are updated frequently — OFAC alone may update the SDN list multiple times per week:

List Update Workflow:
1. New list published by OFAC/UN/EU/DFAT
2. Compliance team notified (automated)
3. New list ingested into screening engine (automated, < 1 hour)
4. Delta screening run against:
   ├── All existing customers (customer screening)
   ├── All pending/queued payments
   └── Recent transactions (look-back period, e.g., 30 days)
5. New alerts generated for newly added entries
6. Audit log of list version used for each screening event

@Service
public class WatchlistMaintenanceService {

    @Scheduled(fixedDelay = 3_600_000)  // Every hour
    public void checkForListUpdates() {
        for (SanctionsList list : SanctionsList.values()) {
            String latestVersion = listProvider.getLatestVersion(list);
            String currentVersion = listVersionRepository.getCurrent(list);

            if (!latestVersion.equals(currentVersion)) {
                List<SanctionsEntry> delta = listProvider.getDelta(
                    list, currentVersion, latestVersion);

                // Update local list
                listRepository.applyDelta(list, delta);
                listVersionRepository.update(list, latestVersion);

                // Re-screen impacted customers
                screeningJobQueue.submit(new DeltaScreeningJob(delta));

                auditLog.info("Sanctions list updated: {} from {} to {}",
                    list, currentVersion, latestVersion);
            }
        }
    }
}

---

Sanctions Evasion Red Flags

Banks must be alert to attempts to circumvent sanctions controls:

Red Flag	Description
Stripping	Removing identifying information from SWIFT MT messages (e.g., removing Iranian bank BIC)
Name misspelling	Deliberate typos to avoid fuzzy matching (e.g., "Khaminei" for "Khamenei")
Shell company routing	Using non-sanctioned shell companies owned by sanctioned individuals
Third-country routing	Routing payments through a non-sanctioned jurisdiction to obscure origin/destination
Proxy ownership	Sanctioned individual owns < 50% but controls the entity through nominees
Mixed jurisdiction	Legitimate entity registered in a sanctioned country (e.g., a branch in Iran)
PayID masking	Using a phone number or email PayID to hide account details
Correspondent layering	Using chains of correspondent banks to obscure the sanctioned party

Payment Stripping (Historical SWIFT MT)

Original MT103:
:52A: BKIRIRIA  ← Iranian bank BIC
:59: /IR987654321 John Smith

Stripped MT103 sent to US correspondent:
:52A: (field removed)
:59: John Smith

US correspondent bank sees no Iranian connection → processes
→ Bank later fined for knowing about the stripping

ISO 20022 makes stripping much harder — structured XML fields are harder to remove without breaking validation.

---

Correspondent Banking and De-Risking

Banks are reducing their correspondent banking relationships to avoid sanctions exposure — known as de-risking:

Bank A (small, high-risk jurisdiction)
  │
  │ Applies for USD correspondent account
  │
  ▼
Major US Bank (potential correspondent)
  │
  │ Assesses:
  ├── Does Bank A have robust sanctions controls?
  ├── Is Bank A's home jurisdiction high-risk?
  ├── What are Bank A's customers like?
  └── Is the compliance cost worth the revenue?
  │
  ▼
If risk too high → De-risk: decline the relationship

This de-risking trend creates financial exclusion for smaller/developing-country banks — a global policy challenge.

---

Regulatory Reporting — True Sanctions Hit

When a confirmed sanctions hit is found:

Immediate Actions (Within Hours)

1. Block the payment permanently
2. Freeze the customer's account (if the customer is sanctioned)
3. Escalate to Compliance Manager and Legal
4. Do NOT tip off the customer — this is a criminal offence

Regulatory Reporting

Jurisdiction	Authority	Report	Deadline
Australia	AUSTRAC + DFAT	SMR / Freezing notice	Immediately / ASAP
US	OFAC	OFAC Report	Within 10 business days
UK	OFSI (HM Treasury)	Freezing / asset report	Within 14 days
EU	National FIU	Per member state rules	Varies

Record Keeping

• All screening events must be logged (payment ID, parties screened, list version, score, decision)
• Minimum retention: 7 years in Australia
• Records must be producible on demand for regulatory examination

---

Sanctions vs AML vs Fraud

Dimension	Sanctions	AML/CTF	Fraud
Purpose	Block geopolitically prohibited entities	Detect financial crime patterns	Prevent monetary loss
Legal obligation	Mandatory — no risk-based discretion	Mandatory — risk-based approach	Risk management
Lists used	Government watchlists (OFAC, UN, DFAT)	Internal TM rules, FATF guidance	Internal blacklists, fraud networks
Can tip off customer?	❌ Criminal offence	❌ Criminal offence	✅ Yes (contact to verify)
Report to	AUSTRAC, OFAC, OFSI, DFAT	AUSTRAC (SMR)	Internal only
Reviewer	Compliance officer	AML analyst	Fraud analyst
Threshold approach	Zero tolerance	Risk-based	Risk-based
False positives	High volume — need analyst triage	Moderate	High — automated clearing common

---

Java Spring — Full Screening Implementation

@Service
@Slf4j
public class SanctionsScreeningService {

    private final WatchlistRepository watchlistRepository;
    private final NameNormalisationService normaliser;
    private final MatchingEngine matchingEngine;
    private final PaymentHoldService holdService;
    private final ComplianceAlertService alertService;
    private final ScreeningAuditRepository auditRepository;
    private final CountryRiskService countryRiskService;

    @Value("${screening.fuzzy.threshold:0.85}")
    private double fuzzyThreshold;

    @Value("${screening.exact.account-numbers:true}")
    private boolean screenAccountNumbers;

    public ScreeningResult screen(PaymentInstruction instruction) {
        List<ScreeningSubject> subjects = extractAllSubjects(instruction);
        List<ScreeningAlert> alerts = new ArrayList<>();

        for (ScreeningSubject subject : subjects) {

            // 1. Country-level check (fast — before name matching)
            if (subject.hasCountry()) {
                CountryRiskResult countryRisk =
                    countryRiskService.screen(subject.getCountry());
                if (countryRisk.isBlocked()) {
                    return escalateHit(instruction, subject, countryRisk.getReason());
                }
            }

            // 2. Account number / BIC exact match
            if (screenAccountNumbers && subject.hasAccountIdentifier()) {
                List<WatchlistMatch> exactMatches =
                    watchlistRepository.findExactByIdentifier(
                        subject.getAccountIdentifier());
                if (!exactMatches.isEmpty()) {
                    return escalateHit(instruction, subject,
                        "Exact account/BIC match: " + subject.getAccountIdentifier());
                }
            }

            // 3. Fuzzy name matching
            String normalisedName = normaliser.normalise(subject.getName());
            List<WatchlistCandidate> candidates =
                watchlistRepository.findCandidates(normalisedName);

            for (WatchlistCandidate candidate : candidates) {
                double score = matchingEngine.score(normalisedName,
                    candidate.getNormalisedName());

                if (score >= fuzzyThreshold) {
                    alerts.add(ScreeningAlert.builder()
                        .subject(subject)
                        .candidate(candidate)
                        .score(score)
                        .matchType(MatchType.FUZZY_NAME)
                        .build());
                }
            }
        }

        // Audit every screening event
        auditRepository.save(ScreeningAuditRecord.builder()
            .paymentId(instruction.getId())
            .subjectsScreened(subjects.size())
            .alertCount(alerts.size())
            .listVersions(watchlistRepository.getCurrentVersions())
            .screenedAt(Instant.now())
            .result(alerts.isEmpty() ? "CLEAR" : "ALERT")
            .build());

        if (alerts.isEmpty()) {
            return ScreeningResult.clear(instruction.getId());
        }

        // Hold payment and notify compliance
        holdService.hold(instruction.getId(), HoldReason.SANCTION_ALERT);
        alertService.raiseAlerts(alerts);

        log.warn("Sanctions alert on payment {}: {} alerts raised",
            instruction.getId(), alerts.size());

        return ScreeningResult.potentialMatch(instruction.getId(), alerts);
    }

    private ScreeningResult escalateHit(PaymentInstruction instruction,
            ScreeningSubject subject, String reason) {
        holdService.holdPermanently(instruction.getId(), HoldReason.CONFIRMED_SANCTION);
        alertService.raiseConfirmedHit(instruction, subject, reason);
        log.error("CONFIRMED SANCTIONS HIT on payment {}: {}",
            instruction.getId(), reason);
        return ScreeningResult.confirmedMatch(instruction.getId(), reason);
    }

    private List<ScreeningSubject> extractAllSubjects(PaymentInstruction instruction) {
        List<ScreeningSubject> subjects = new ArrayList<>();

        // Parties
        subjects.add(ScreeningSubject.fromParty(instruction.getDebtor(),
            SubjectRole.DEBTOR));
        subjects.add(ScreeningSubject.fromParty(instruction.getCreditor(),
            SubjectRole.CREDITOR));

        // Agents / BICs
        subjects.add(ScreeningSubject.fromBic(instruction.getDebtorAgentBic(),
            SubjectRole.DEBTOR_AGENT));
        subjects.add(ScreeningSubject.fromBic(instruction.getCreditorAgentBic(),
            SubjectRole.CREDITOR_AGENT));

        // Optional parties
        if (instruction.hasUltimateDebtor()) {
            subjects.add(ScreeningSubject.fromParty(instruction.getUltimateDebtor(),
                SubjectRole.ULTIMATE_DEBTOR));
        }
        if (instruction.hasUltimateCreditor()) {
            subjects.add(ScreeningSubject.fromParty(instruction.getUltimateCreditor(),
                SubjectRole.ULTIMATE_CREDITOR));
        }

        // Intermediary agents
        instruction.getIntermediaryAgents().forEach(agent ->
            subjects.add(ScreeningSubject.fromBic(agent.getBic(),
                SubjectRole.INTERMEDIARY_AGENT)));

        return subjects.stream()
            .filter(ScreeningSubject::isScreenable)
            .collect(toList());
    }
}

---

Screening in the Payment Lifecycle

                ┌────────────────────────────────────┐
                │    SCREENING TOUCHPOINTS            │
                ├────────────────────────────────────┤
  Onboarding ──►│ 1. KYC onboarding screen           │
                │    (customer + UBOs + directors)    │
                ├────────────────────────────────────┤
  Outbound   ──►│ 2. Pre-payment screen               │
  Payment       │    (all parties in pain.001)        │
                ├────────────────────────────────────┤
  Inbound    ──►│ 3. Inbound payment screen          │
  Payment       │    (all parties in pacs.008)        │
                ├────────────────────────────────────┤
  Scheduled  ──►│ 4. Periodic re-screen              │
  Batch         │    (existing customer base)         │
                ├────────────────────────────────────┤
  List Update──►│ 5. Delta screen on list change     │
                │    (re-screen against new entries)  │
                └────────────────────────────────────┘

---

Related Concepts

• aml_kyc.md — Runs alongside sanctions; different purpose and lists
• fraud.md — Third compliance check in the same pipeline
• outbound.md — Sanctions check before payment is submitted
• inbound.md — Sanctions check on received pacs.008
• fis.md — FI-level (BIC/country) sanctions screening
• payment_exceptions.md — Sanctions holds as exceptions
• swift.md — SWIFT gpi and sanctions compliance