## wpiea2021199-print-pdf

## Source details

**Canonical URL:** [wpiea2021199-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021199-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021199-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021199-print-pdf.pdf.json)

---

### Introduction
- Remittance flows have been historically under-emphasized by macroeconomists relative to FDI and hot money (HM) flows and by development economists relative to aid flows (e.g., ODA).
- Remittances exhibit steadier and more anti-cyclical behavior than FDI and some aid flows, providing critical lifelines during downturns.
- Initial international organization expectation: global remittances to plunge of 20% in 2020 (World Bank, 2020).

### Relevance and Magnitude of Remittances
- Over the last two decades remittance flows to low income countries (LIC) have been larger than FDI flows, with the gap increasing in recent years.
- Remittances can surpass 10 percent of GDP and reach levels above 30 percent for many countries; several highly remittance-dependent countries are middle-income rather than LIC.
- Examples of remittance-dependent and disaster-vulnerable countries: Haiti, Honduras, Philippines, Tonga.
- Selected cited empirical findings:
  - Azizi (2019) and Adams (2011) — effects on poverty.
  - Terrelonge (2014) and Amega (2018) — effects on health.
  - Acharya and Leon-Gonzalez (2014) and Askarov and Doucouliago (2020) — effects on education.
  - Aggarwal et al. (2011), Donou-Adonsou et al. (2020) and Bhattacharya et al. (2018) — effects on financial inclusion.
  - Cazachevici et al. (2020) — positive but small effect on growth, with regional variation.
  - Chami et al. (2005) — hypothesis of negative correlation with per-capita GDP growth via lower work incentives.
  - Barajas et al. (2009) — remittances alleviate poverty and smooth consumption but do not impact growth.
- Evidence that remittances increase after natural disasters (Mohapatra et al., 2012).

### Cost of Remittances — Scale and Distributional Concerns
- The literature on drivers of remittance costs is "scant"; Beck and Peria (2009) is an earlier exception.
- RPW database statistics:
  - Average global cost of remittances stood at 6.8% at end-2019.
  - World Bank estimate: global volume of remittances reached USD714 billion in 2019.
  - Implication: almost USD50 billion in remittances did not reach intended recipients (6.8% of USD714 billion).
  - Average cost of sending remittances to SSA countries stood at 9% at end-2019.
- McKinsey (2016): average cost of a cross-border payment for a US bank is more than 10 times that for a domestic payment.
- High costs disproportionately affect poorest recipients and small-value transfers.

### Data and Research Contribution
- This paper addresses gaps by:
  - Utilizing ten years of RPW data (longitudinal advantage).
  - Expanding cross-section coverage from 119 corridors (Beck and Peria, 2009) to 365 corridors.
  - Delving into remittances ecosystem to identify factors driving costs.
- Two key objectives:
  - Provide realistic assessment of what policymakers can and cannot do to reduce remittance costs.
  - Offer realistic policy advice to lower remittance costs.
- Paper structure (high-level): geography (Section II); mechanics and operational constraints (Section III); database analysis and limitations (Section IV); evolution of costs (Section V); deeper corridor-level analysis (Section VI).

### II. The Geography of Remittance Flows
- Motivation: volume affects costs via economies of scale; larger-volume corridors attract more RSPs and competition.
- Concentration:
  - Top ten sending countries account, approximately, for 60% of world remittance outflows.
  - Top ten receiving countries account, approximately, for 50% of remittance inflows.
- Stylized observations:
  - United States is the leading sending country.
  - India and China occupy the top 2 receiving positions.
  - Unexpected senders: Saudi Arabia and Russia; unexpected receivers: France and Germany.
  - Philippines noted for a government-supported migration culture.
- Corridor-level scale:
  - US-Mexico corridor volume is 70% higher than the second-largest corridor.
  - Among top 10 largest corridors, 8 have the US as origin; 3 have India as receiver; China has one corridor in top 10.
  - US-Nigeria and UK-Nigeria are among top 15, but no other SSA corridors are in the top 50, indicating small SSA corridor sizes.
- Flows by income level (average 2017–2019):
  - LIC receive a tiny fraction of global remittance flows.
  - Most flows go from HIC to LMC; amount to LIC is substantially lower.
  - UMC and HIC receive about half of remittance flows.

### III. The Mechanics of Remittances
- Fundamental distinctions:
  - Cross-border nature and currency differences differentiate remittances from domestic payments.
  - Front-end environment: visible sender/receiver/RSP interactions (internet, mobile, branches, MTO, ATM).
  - Back-end environment: invisible clearing, settlement, FX functions, IT systems and messaging.
- Front-end elements:
  - RSP types: banks, MTO, post office (PO); payment instruments: cash, bank transfer, debit/credit card; account types: regular bank account, e-money account, cash/prepaid options.
- Back-end arrangements:
  - Correspondent bank relationships (CBR); closed-loop systems; interlinked national payment infrastructures (e.g., Directo a México linking FedACH with Mexican RTGS); peer-to-peer/DLT models.
- Stylized payment-chain numerical example:
  - Sender sends USD200 using debit card.
  - Exchange rate used by Agent E: USD/ZMW=20.
  - Receiver expected to get ZMBK 4,000.
  - Domestic and cross-border banking steps involve multiple bank accounts (Bank E1–E4, Bank P1–P3) and payment systems E(a)/E(b), P(a)/P(b).
  - Bank E4 may use different exchange rate than Agent E, creating FX differences.
- Operational factors affecting speed, transparency and cost:
  - Batching policies, domestic payment system efficiency, service hedonic features (speed vs cost), network configuration.
- Observed delays:
  - BIS (2018) survey: some cross-border transactions may take up to seven days.
  - Savings possible via batching and netting.

### Box 1 — Types of Remittances Services (models and CBR architectures)
- Service models:
  - Unilateral: single RSP operates in both jurisdictions (physical presence or virtual).
  - Franchised: RSP contracts agents to capture/disburse; main model for traditional MTOs.
  - Negotiated: selected institutional partners create corridor-focused networks.
  - Open: open networks where capturing RSP can send to any disbursing RSP (requires funds and info to travel together); typical for banks using CBRs.
- Open model shortcomings:
  - Capturing RSP may not know disbursing RSP fee ex-ante; settlement slower and timing unpredictable.
  - Disbursing RSP liquidity management or advance payments create a "speed premium."
  - FX risk and transparency constraints can increase costs; faster payments tend to be costlier.
- Correspondent banking architectures:
  - Traditional CBR: direct correspondent for international payments; more common between developed systems or high-volume corridors.
  - Nested CBR: intermediaries used when direct correspondent is unavailable; makes payments slower, costlier, less transparent.
  - Payable-through accounts: respondent bank allows customers to access correspondent account directly; targeted by de-risking due to higher compliance risks.
- Global policy targets and timeline:
  - G8 L’Aquila Summit (2009): reduce global average cost to 5% in 5 years (5x5 target).
  - G20 endorsement at Cannes (2011); reiterated Brisbane (2014) to reduce global average to 5% (no new deadline).
  - UN SDGs (2015): by 2030 reduce global average cost to less than 3% and eliminate corridors with costs above 5%.
- RPW database (vintage used in paper):
  - RPW initial coverage: 14 sending and 67 receiving countries (120 corridors).
  - Vintage used: 48 sending and 110 receiving countries, covering 365 corridors; data up to third quarter of 2018; sample period Q1 2011–Q3 2018.
  - Sending-country corridor counts: United States 42, United Kingdom 33, Germany 24.
  - Receiving-country corridor counts: India 20, China 16, The Philippines 14.
  - Survey methodology: independent researchers pose as customers; data for major RSPs covering aggregated market share of, at least 80, percent; two amounts surveyed: USD200 and USD500; this paper focuses on USD200.

### Data Quality, RPW Limitations, and Measurement Caveats
- RPW database is "noisy" and requires scrutiny.
- Exact database measurement issues:
  - 3.7 percent of RPW entries show negative exchange rate margins.
  - 23.1 percent show margins equal to zero; RPW metadata: "The 0% in the exchange rate margin does NOT necessarily mean that there is no exchange rate cost, but rather that this cost is not disclosed to the sender at the time of sending."
  - RPW classifies more than 6 percent of entries as not transparent due to exchange rate issues.
  - 1 percent of entries shows zero total remittance cost.
  - Evidence suggests at least one quarter of entries seem to underestimate remittance costs.
- Methodology changes:
  - Methodology changed in Q2 2016 (categories renamed/created/dropped).
  - Survey periodicity: initially semi-annual; since Q2 2013 quarterly; RPW database available from 2011.
- Time-sensitive variables pose challenges (exchange rate selection within day can create negative margins/outliers).
- RPW may fail to capture destination fees; severity and number of affected entries unknown.

### Why Remittance Costs Matter and the SDG Target
- High remittance costs:
  - Increase inefficiency, incentivize informal channels, impair balance of payments statistics, hinder financial inclusion.
  - Reduce resources available for education and health, affecting human capital and growth.
- SDG item 10.c: "By 2030, reduce to less than 3 per cent the transaction costs of migrant remittances and eliminate remittance corridors with costs higher than 5 per cent."
- Aggregation ambiguity: SDG target does not specify simple vs weighted average or weighting scheme; RPW initially used simple averages, since issue 4 (Nov 2012) a weighted average by size of flows has been published (weights do not consider RSP market share).

### Measurement, Aggregation, and Trend Evidence
- Global average cost:
  - Around 9% in 2011; global costs fell slowly thereafter.
  - Mid-2015 trend reversal largely an artifact of changing sample (addition of latecomers from less-developed economies with higher costs).
- Balanced-sample evidence shows stability rather than increase.
- Aggregation effects:
  - Computing global average from corridor-level averages shifts results downward in second half of sample (corridor averages overweight developed corridors with many corridors and RSP).
  - Costs stalled from mid-2015 to mid-2018 in several aggregations.

### Regional, Country, and Corridor Heterogeneity
- SSA is the most expensive region by far but experienced faster falls in costs than other regions; by mid-2018 gap with East Asia and the Pacific (EAP) had drastically reduced.
- Regional grouping by cost:
  - Lower-cost group: Europe and Central Asia (ECA), Latin America and Caribbean (LAC), South Asia (SAR).
  - Middle-cost group: EAP and MENA.
  - SSA standalone highest-cost group.
- Figure 12 (2017Q4–2018Q3 average) observations:
  - Difference between least and most expensive group is almost fivefold.
  - 11 out of 15 most expensive countries are from SSA.
  - 9 out of 15 least expensive are former Soviet Republics.
  - Only five countries (all former Soviet Republics) had average cost below 3% in 2018.
- Inter-country corridor variation:
  - For large receivers (India, Philippines, China) and Nigeria, cheapest vs most expensive sending countries differ by up to 6-fold (India) or almost 3-fold (Nigeria).
- Top-30 most costly corridors: 8 out of 10 costliest are in SSA; many originate in South Africa and Thailand; several involve China and India as receivers.
- Top-30 least costly corridors: all 10 least costly corridors are between former Soviet Republics with Russia as sender; several low-cost corridors have India or MENA senders.

### Provider Types, Payment Methods, Intra-Corridor Dispersion, and FX Margins
- Provider-type averages:
  - Banks are on average much more expensive than MTO; MTO are more expensive than PO.
  - Banks charge 50 percent more than MTO to send to China and Philippines and twice as much to send to India and Nigeria (not shown in figures).
  - PO are often the cheapest provider type but not always.
- Payment instruments:
  - Cash is the most common: around 40 percent of remittances involve cash at both ends.
  - Bank transfers are second most popular and among the most expensive methods.
  - Less than one percentage of remittances are carried out using mobile money; this percentage remained constant across periods described.
- Intra-corridor variation:
  - Large intra-corridor cost dispersion can be as large as inter-corridor differences.
  - Positive correlation between average corridor cost and variability: higher-cost corridors tend to have larger cost dispersion.
- FX margin relevance:
  - For the 30 costliest corridors average FX margin share is about 17 percent.
  - In some corridors FX margin can be as high as 60 percent of total cost.
  - Average FX margin among top-30 FX-margin corridors is essentially the same for MTO and banks, but transparency differs.
- FX margin transparency (Table 2 data):
  - Zero FX margin reported: Banks 19%, MTO 7%, PO 53%.
  - Negative FX margin reported: Banks 2%, MTO 1%, PO 2%.
  - Percentage of banks reporting zero or negative FX margins is almost 3 times higher than MTO, suggesting banks are less transparent and RPW may understate true global average remittance cost.

### Drivers of Remittance Costs (classification and evidence)
- Four broad driver categories: market structure, firm-specific, demand-related, regulatory factors.
- Market structure:
  - Less competitive markets and concentrated banking systems tend to have higher prices.
  - Number of surveyed RSP per corridor shows negative but insignificant correlation with average costs.
  - Number of banks per corridor shows positive correlation with average costs (counter-intuitive).
  - Number of MTOs per corridor shows negative correlation with average costs.
  - Ratio of MTO to banks is strongly negatively correlated with costs, implying competition mainly occurs via specialized MTOs.
- Firm-specific factors:
  - Banks incur higher regulatory/compliance costs; MTOs rely on agents; fintechs operate online.
  - Network and agent structures affect costs; technology is changing dynamics.
- Demand-related factors:
  - Scale matters: remittance-receiving country GDP, number of migrants, and inflow volume negatively correlated with average costs; inflow volume most correlated.
  - Former USSR republics are notable outliers with much lower costs.
- Regulatory factors:
  - Exclusivity clauses by large MTOs act as barriers to entry and curb competition.
  - AML/CFT and de-risking: active correspondents shrunk by 20% and active corridors decreased by 10% from early-2012 to end-2018.
  - Americas (excl. North America) saw largest decline in active correspondents (~30%).
  - Increased correspondent concentration correlates with slower falls or stagnation in global remittance costs around mid-2015 to early-2017.
  - Restrictions limiting remittance provision to banks (IFAD: in 80% of African countries) reduce competition and raise costs.
  - Over-stringent AML/CFT implementation (KYCC, proof-of-address) and excessive compliance expectations increased costs and reduced access.

### Empirical Links Between Cost Components and Other Factors
- Average remittance costs positively correlated with explicit fees and FX margins; fees show tighter link, FX margins strongly associated as well.
- FX margin and exchange rate regime:
  - FX margin appears unrelated to IMF AREAER exchange rate regime classification: average FX margins in fixed vs floating ER regime groups are almost the same.
  - Flexible ER group has lower cost dispersion (2.8 versus 4.1 standard deviation).
- Capital controls on transfers into by immigrants are rare and only marginally associated with average remittance cost.

### Policy Recommendations and Intervention Channels
- Foster competition:
  - Prohibit or eliminate exclusivity clauses that lock agents/markets to single providers.
  - Broaden licenses to allow non-bank RSP (including MMOs, fintechs) to provide remittance services independently.
  - Ensure proportionate, risk-based regulation for non-bank RSP appropriate to their risk profile, avoiding bank-tailored burdens.
- Improve price transparency and consumer protection:
  - Require RSP to disclose cost components they know at time of transfer and clearly state unknown elements.
  - Promote timely remittance cost data collection and dissemination for consumer comparison (e.g., PROFECO weekly statistics in Mexico).
- Reduce barriers to entry and increase contestability:
  - Ensure non-bank RSP fair access to payment system infrastructure; promote interoperability between payment systems.
- Address AML/CFT and de-risking consequences with balanced approaches:
  - Apply proportionate AML/CFT implementation; avoid measures that go beyond FATF recommendations and that result in unnecessary CBR withdrawal.
  - Avoid imposing KYCC or proof-of-address requirements beyond international guidance where inappropriate for remittances.
- Reduce non-pecuniary and hidden costs:
  - Increase geographical coverage of payout locations (leverage postal networks and national post operators; post office global network > 650,000).
  - Encourage post office partnerships in a non-discriminatory, competition-preserving manner; APFSI pilot: PO-delivered remittances cost 37 percent lower than market average; PO market share increased; remittances delivered via PO increased by 60 percent in first three years of pilots.
- Ensure proportionate licensing and avoid excessive compliance requirements that disproportionately raise costs and act as entry barriers.

### Conclusions and Overarching Findings
- Remittance markets are complex, heterogeneous, and multi-layered; analogies with domestic payment systems are simplistic.
- No single magic solution (e.g., fintech alone) will rapidly and substantially reduce costs in poorest regions; structural and endogenous factors limit speed and magnitude of reductions.
- Regularities:
  - Banks generally more expensive than MTO; MTO more expensive than PO.
  - Costs positively associated with number of banks in a corridor but negatively with number of MTOs and the MTO/banks ratio.
  - Higher-cost corridors exhibit higher cost dispersion; active shopping can yield large savings.
  - FX margin is a major component and not clearly related to exchange rate regime.
  - Costs negatively associated with remittance volume, explaining difficulties for poorer countries.
- Policy must be multifaceted: prohibit exclusivity, adopt proportionate AML/CFT application, broaden licensing, enforce transparency, promote payment-system access and interoperability, and leverage postal networks.

### Appendix 1 — Remittance Costs and Exclusivity Agreements: Key Findings
- Central feature: pervasive use of exclusivity clauses by Western Union (WU) and MoneyGram (MG); IFAD (2009) assessed these agreements "effectively 'lock' more than half of all available payout locations."
- Market effects: exclusivity clauses act as barriers to entry, allow incumbents to charge excessive prices, and "lock" agent networks.
- Jurisdictional and regulatory actions cited:
  - Russia mid-2001: exclusivity clause with WU ruled null and void (2003 ruling; WU lost final appeal in October 2004).
  - Ukraine 2004: WU committed to reduce fees by "... no less than 40%".
  - Ethiopia NBE Directive No FXD/30/2006 required non-exclusive agency conditions and prior approval for new agency agreements.
  - Senegal (March 2008) letter warning exclusivity clauses at odds with Competition Law.
  - Central Bank of Nigeria directive (November 2008) forbade approving money transfer agreements with exclusivity clauses.
  - RBI circular "RBI/2010-11/206" (September 2010) ruled inbound exclusivity arrangements illegal.
  - Central Bank of Liberia Directive "CBL/RSD/002/2011" (October 2011) required expunging exclusivity clauses and set fine "not less than L$100,000 [approximately USD1,000] for each day of violation."
  - Bank Al Maghrib (2011) banned exclusivity in Morocco.
  - Rwanda regulation No 06/2012 prohibited exclusivity; repealed by Regulation No 07/2015 allowing exclusivity only if authorized by Central Bank.
  - Tunisia circular N°2013-01 instructed banks to remove exclusivity clauses.
- Multilateral guidance: CPSS and World Bank (2007) recommended discouraging exclusivity conditions.
- Enforcement observations:
  - Discouraging exclusivity without binding prohibition is insufficient; prohibitions must be binding and enforced with penalties.
  - Even when outlawed, agents may remain effectively exclusive; monitoring post-prohibition outcomes is necessary.
- Jurisdictional limits:
  - No remittance-sending country had made exclusivity clauses illegal (as of document).
  - US DOJ Antitrust Division (2004–2005) concluded facts did not support a claim that WU contracting practices harmed US consumers.
  - European Commission preliminary antitrust investigation opened in 2016 remained at preliminary status four years later.
- Market concentration evidence:
  - Watkins and Quattri (2014): in more than two thirds of SSA countries WU and MG controlled at least 50% of market.
  - Since 2011 many former USSR republics face lowest remittance costs globally; Russia has lowest sending costs among G20 peers.
  - Ukraine example shows causal impact: WU agreement produced fee reduction of "... no less than 40%".
- Policy implications:
  - Make prohibitions on exclusivity clauses binding and enforceable with financial penalties.
  - Address market structure and concentration; removing exclusivity lowers prices only if it translates into more competition among MTOs and agents.
  - Use indirect regulatory tools where direct jurisdiction over international MTOs is limited (e.g., Ethiopia's non-exclusive agency requirement).
  - Monitor market outcomes post-prohibition to ensure agents do not remain effectively exclusive.

### Appendix 2 — AML/CFT and the Loss of Correspondent Banking Relationships
- Key market challenges:
  - Most transactions by a couple of MTOs, often less regulated as non-banks, and many transactions in cash.
  - Historically lax identification requirements by MTOs; cash transactions leave little trail.
  - Stricter regulation and enforcement have produced unintended consequences harming remittances as lifelines.
- Enforcement and fines:
  - WU settled a $586 million fine with US FTC and DOJ for criminal violations including failing to maintain effective AML program and aiding wire fraud.
  - In 2019 WU fined €1 million by French regulator for AML rule violations.
  - MG fined $125 million one year after WU FTC/DOJ fine.
  - Large bank fines: HSBC USD1.9 billion (2012); BNP Paribas USD8.9 billion (2017); Deutsche Bank USD630 million (2017).
- Consequences:
  - Banks increased controls and began de-risking (severing CBRs with respondent banks, MTOs, central banks).
  - Compliance focus extended from KYC to KYCC and KYCCC, exceeding FATF requirements in practice.
- Evidence on CBR network:
  - Despite >50% increase in remittance volume, active correspondents decreased by around 20% since 2012.
  - Active corridors decreased by 10% since 2012.
  - CBR network became more concentrated (Gini on active correspondents per corridor increased).
  - Increase in correspondent concentration as of early 2015 coincides with intensification of CBR losses and an interruption in the decrease in global remittance costs six months later.
- Data caveats:
  - SWIFT message volume increased, which might signal scale gains; SWIFT is not the only messaging system and may not reliably represent overall remittances market.
- Countries most affected by loss of CBR (percent change in ratio of banks abroad receiving payment messages and banks receiving payment messages in the country, 2012–2018) — top 40 most dramatic losses (percent changes shown exactly as in source):
  - Timor-Leste - 83.0
  - St Pierre and Miquel - 77.8
  - Mayotte - 71.4
  - Vanuatu - 69.0
  - Seychelles - 67.4
  - Syrian Arab Republic - 67.1
  - Tajikistan - 67.1
  - French Guiana - 65.7
  - Eritrea - 64.1
  - Jamaica - 59.7
  - Bermuda - 58.5
  - Monaco - 56.4
  - Gibraltar - 55.7
  - Cook Islands - 53.2
  - New Caledonia - 51.4
  - Belize - 51.2
  - Cuba - 50.8
  - Brunei Darussalam - 50.7
  - Virquin dsl (GB) - 50.0
  - Solomon Islands - 49.9
  - Yemen - 49.9
  - Faeroe Islands - 48.5
  - Venezuela - 48.5
  - Finland - 48.1
  - Angola - 47.8
  - Ghana - 46.8
  - Tonga - 46.2
  - Andorra - 46.0
  - Vatican City State - 45.1
  - Puerto Rico - 44.3
  - Montenegro - 44.3
  - Trinidad and Tobago - 44.2
  - Libyan Arab Jamahiriya - 44.1
  - Jersey, C.I. - 43.7
  - Isle of Man - 43.4
  - Fiji - 43.2
  - Bolivia - 43.1
  - Morocco - 42.5
  - Mauritius - 41.9
  - Moldova - 40.2
- Distributional effects:
  - All regions affected by CBR withdrawal; poor countries and conflict zones particularly hit (example: Westpac shut down money transfer services to Somalia in 2014).

*Italic: Source — wpiea2021199-print-pdf (selected sections and appendices summarized).*

### References ___________________________________________________________ 51

### wpiea2021199-print-pdf - References ___________________________________________________________ 51

### Introduction
- Remittance flows have been historically under-emphasized by macroeconomists relative to FDI and hot money (HM) flows and by development economists relative to aid flows (e.g., ODA).
- Remittances exhibit steadier and more anti-cyclical behavior than FDI and some aid flows, providing critical lifelines for many families and countries during downturns.
- The COVID-19 pandemic prompted concerns of a sharp decline in remittances: international organizations initially expected global remittances to plunge of 20% in 2020 (World Bank, 2020).

### Relevance and Magnitude of Remittances
- Panel A (Figure 1) indicates that over the last two decades remittance flows to low income countries (LIC) have been larger than FDI flows, with the gap increasing in recent years.
- Panel B (Figure 1) shows remittances can surpass 10 percent of GDP and reach levels above 30 percent for many countries; several highly remittance-dependent countries are not LICs but middle-income.
- Examples of remittance-dependent and disaster-vulnerable countries include Haiti, Honduras, Philippines and Tonga.
- Empirical literature cited:
  - Azizi (2019) and Adams (2011) — effects on poverty.
  - Terrelonge (2014) and Amega (2018) — effects on health.
  - Acharya and Leon-Gonzalez (2014) and Askarov and Doucouliago (2020) — effects on education.
  - Aggarwal et al. (2011), Donou-Adonsou et al. (2020) and Bhattacharya et al. (2018) — effects on financial inclusion.
  - Cazachevici et al. (2020) — remittances have a positive but small effect on growth, with regionally varying qualitative effects.
  - Chami et al. (2005) — hypothesis that remittances may correlate negatively with per-capita GDP growth due to lower work incentives.
  - Barajas et al. (2009) — remittances alleviate poverty and smooth consumption but do not impact growth.
- Evidence exists that remittances increase after natural disasters (e.g., Mohapatra et al., 2012).

### Cost of Remittances — Scale and Distributional Concerns
- The number of academic studies on drivers of remittance costs is "scant"; Beck and Peria (2009) is a noteworthy earlier exception.
- Remittances are expensive and disproportionately affect the poorest recipients; sending remittances to the poorest countries is substantially more expensive.
- RPW database statistics:
  - Average global cost of remittances stood at 6.8% at end-2019.
  - World Bank estimate: global volume of remittances reached USD714 billion in 2019.
  - Implication: almost USD50 billion in remittances did not reach intended recipients (calculated from 6.8% of USD714 billion).
  - Average cost of sending remittances to SSA countries stood at 9% at end-2019.
- McKinsey (2016) finding: the average cost of a cross-border payment for a US bank is more than 10 times that for a domestic payment.
- The high cost generates a significant loss of resources to recipients, with especially steep burdens on low-income recipients and small-value transfers.

### Data and Research Contribution
- This paper addresses the academic gap on remittance costs by:
  - Utilizing ten years of RPW data (longitudinal advantage over earlier cross-section studies).
  - Expanding cross-section coverage from 119 corridors (in Beck and Peria, 2009) to 365 corridors in the current analysis.
  - Delving deeper into the remittances ecosystem to identify myriad factors driving costs.
- Two key objectives:
  - Provide a realistic assessment of what policymakers can and cannot do to reduce remittance costs.
  - Offer realistic policy advice to lower remittance costs.
- Paper structure overview (high-level):
  - Section II: geography of remittances.
  - Section III: mechanics of remittance transfers and operational constraints.
  - Section IV: analysis of the database, shortcomings and limitations.
  - Section V: evolution of remittance costs across regions, countries and corridors.

_This summary is based solely on the supplied content from the specified PDF content unit._

### Section VI digs deeper and looks at cost dif ference within corridors, among remittance

### wpiea2021199-print-pdf - Section VI digs deeper and looks at cost dif ference within corridors, among remittance

### II. THE GEOGRAPHY OF REMITTANCE FLOWS
- Motivation:
  - Volume is a crucial variable for costs because they are affected by economies of scale.
  - Competition is likely to be higher for larger-volume corridors, which are likely to attract more remittance service providers (RSP).
- Concentration:
  - The top ten sending countries account, approximately, for 60% of the world remittance outflows.
  - The top ten receiving countries account, approximately, for 50% of remittance inflows.
- Stylized country observations:
  - The United States is the leading remittance-sending country.
  - India and China occupy the top 2 receiving positions.
  - Unexpected members in sending and receiving groups include Saudi Arabia and Russia (senders) and France and Germany (receivers).
  - The Philippines is noted for a deeply rooted culture of migration, supported by the government.
- Corridor-level scale:
  - The most relevant scale variable is likely to be the volume at the corridor level rather than at the country level.
  - The US-Mexico corridor (green in Figure 3) is by far the largest one, with a volume 70% higher than the second-largest corridor.
  - Among the top 10 largest corridors, 8 have the US as the origin.
  - Among the top 10 largest corridors, 3 have India (black) in the receiving end.
  - China (red) has just one corridor in the top 10.
  - US-Nigeria and UK-Nigeria corridors are among the top 15 largest corridors, but there are no other corridors involving SSA countries in the top 50, indicating SSA corridors have small size—a key constraint on driving costs down.
- Remittance flows by income level (average 2017–2019):
  - Low-income countries (LIC) receive a tiny fraction of global remittance flows.
  - Most remittance flows go from high income countries (HIC) to lower-middle income countries (LMC); the amount going to LIC is substantially lower.
  - Upper-middle income countries (UMC) and HIC receive about half of the remittance flows.
- Data and caveats:
  - Figure data sources: KNOMAD and The World Bank.
  - The methodology for bilateral remittance estimates has been questioned, and more recent estimates are not available.

### III. THE MECHANICS OF REMITTANCES
- Purpose:
  - Explain and clarify the mechanics behind remittance flows and reveal the operational complexity behind cross-border retail payments.
- Fundamental distinctions:
  - Remittance payments differ from domestic payments because they occur between two jurisdictions and usually involve different currencies.
  - Front-end environment: visible interactions between sender, receiver, and RSP (service channels/access points).
  - Back-end environment: invisible operations (IT systems, messaging, payment processes) that handle clearing, settlement, and FX functions.
- Front-end elements:
  - Service channels/access points: internet, mobile phone, bank branch, money transfer operator (MTO), automated teller machine (ATM).
  - RSP types: banks, MTO, post office (PO), etc. (availability varies by sender).
  - Payment instruments: cash, bank transfer, debit card, credit card.
  - Account types: regular bank account, e-money account; cash or prepaid card options exist if sender lacks an account.
- Back-end arrangements and examples:
  - Correspondent bank relationship (CBR): a correspondent bank holds deposits for a respondent bank and provides payment and other services.
  - Closed-loop (proprietary) systems: capturing and disbursement RSP are the same or belong to the same group (used by most MTO), but third parties are often required for some payment-related services (agents, franchisees, vendors, service providers).
  - Interlinking national payment infrastructures: example cited is Directo a México linking FedACH in the US with the Mexican RTGS system.
  - Peer-to-peer arrangements: direct transactions between end-users, enabled by technologies such as DLT.
- Payment-chain complexity (Figure 6 stylized example and implications):
  - The payment chain involves many clearing and settlement processes from sender payment to receiver disbursement; information and financial flows run separately and at very different speeds.
  - Stylized numerical example:
    - Sender plans to send USD200 and pays with his debit card.
    - Exchange rate used by Agent E equals USD/ZMW=20.
    - Receiver in Zambia is expected to get ZMBK 4,000.
  - Domestic steps (origin country):
    - Sender’s account at Bank E1 is debited USD200 when payment system E(a) is settled; Agent E’s account at Bank E2 is credited USD200.
    - Agent E transfers USD200 to RSP E via payment system E(b); Agent E’s account at Bank E2 is debited USD200 and RSP E’s account at Bank E3 is credited USD200.
  - Cross-border steps (correspondent banks and recipient country):
    - Bank E3 contacts Bank E4 (a global bank) because Bank E3 does not provide international payment services.
    - Bank E4 informs Bank E3 about the exchange rate it will use (which may differ from the rate Agent E provided the sender).
    - Assuming Bank E3 holds sufficient funds at Bank E4, Bank E4 contacts Bank P1 (with whom it has a CBR) and instructs Bank P1 to credit RSP P’s account at Bank P2 by ZMW4,000 via payment system P(a).
    - RSP P requests Bank P2 to transfer funds to Agent P’s account at Bank P3 using payment system P(b).
    - Agent P pays the receiver in cash.
  - Key operational factors affecting speed, transparency, and cost:
    - Operational procedures of institutions (e.g., batching small-value transactions once per day can be cost-effective but introduces delay).
    - Efficiency/speed of domestic payment systems.
    - Hedonic features of the remittances service contracted by the sender (e.g., speed vs cost trade-offs).
    - Configuration of the remittances service network (how sender and receiver service channels are connected).
  - Observed delays and practices:
    - A BIS (2018) survey reported some cross-border transactions might take up to seven days to be executed.
    - Savings can be achieved by batching transactions and netting financial flows from both countries.
- Implications:
  - The separation of rapid information flow from slower financial settlement creates transparency and timing frictions that influence pricing and service features.
  - Small transaction sizes and fragmented corridor structures (especially in SSA) constrain the ability to achieve scale economies and lower costs.

*Italic: Source: wpiea2021199-print-pdf (selected sections II and III).*

### Box 1. Types of Remittances Services

### Box 1. Types of Remittances Services

### Types of remittances service models
- Unilateral service model
  - RSP offers service without involving other actors as capturing or disbursing agents.
  - Possible when the RSP is physically present at both jurisdictions or when the network is virtual (computers and mobile phones).
  - Example: global banks with physical presence in many countries (ICICI Bank branch in the UK sending remittances to India; Standard Chartered sending remittances from London to its branch in Zambia).
  - Limitation: constrained ability to deliver services where it is not economically viable to be physically present—often precisely where migrants originate.

- Franchised service model
  - RSP enters a legal contract with agents to offer remittances services on the RSP’s behalf.
  - RSP provides back-end infrastructure (messaging, settlement) and front-end needs (marketing).
  - Typical agents: gas stations, liquor stores, supermarkets, post offices, foreign exchange bureaus.
  - Main model for traditional MTOs (Western Union, MoneyGram, Ria, UAE Exchange) providing great capillarity.

- Negotiated service model
  - RSP engages selected institutions in other jurisdictions to create mutually beneficial networks of physical access points.
  - Common for RSPs focused on specific corridors; partners are likely not competitors and share comparative advantages.
  - Examples: bilateral arrangements between banks, credit union schemes, transfer services or schemes by postal organizations (e.g., La Poste with Algerie Poste; DBS (Singapore) with DBO (Philippines)).

- Open service model
  - Capturing RSP uses an open network accessible to any RSP to send funds to the disbursing RSP; no direct relationship between capturing and disbursing RSP (and their agents).
  - Requires transaction information and funds to travel together for settlement.
  - Typical for banks using correspondent bank relationships or interlinked national payment systems; non-bank RSPs can access as bank customers.
  - Strength: allows banks to send funds to virtually any other bank worldwide, reaching even smallest corridors.
  - Shortcomings:
    - Capturing RSP does not know ex-ante the disbursing RSP’s fee; final remittance cost may not be known until completion.
    - Payment settlement is slower and timing is difficult to predict compared with models where fees are known ex-ante.
    - Lack of synchronization between payment and information flows can be managed by disbursing RSPs with liquidity management, working capital lines, or providing liquidity to agents—but paying before receiving funds incurs extra costs (opportunity cost) or operational risks, prompting a “speed premium.” The faster the remittance payment the costlier it is likely to be.
    - Foreign exchange risk: example given where agreed exchange rate of 20 moves to 19, increasing required USD and causing potential FX premium charged by intermediary agents.
    - Transparency constraints: capturing RSP cannot know disbursing fees ex-ante, making transparent pricing and speed information challenging; disbursing fee may be the only revenue source for disbursing RSP/agent, increasing remittance cost.

### Correspondent banking architectures and implications
- Traditional CBR
  - More likely between countries with more developed financial systems or where cross-border payment volumes are high.
  - Example: Nigerian respondent bank has a CBR with a New York correspondent bank which directly engages the Brazilian customer bank’s correspondent in Miami (as depicted in preceding figures).

- Nested CBR
  - Occurs when a respondent bank lacks a correspondent that can directly engage the beneficiary bank’s correspondent; intermediaries are used.
  - Example: Beninese customer’s bank uses a Nigerian counterpart which has a CBR with a New York bank that has the relationship with the Brazilian correspondent.
  - Consequences: payment becomes slower, costlier and less transparent.

- Payable-through (pass-through / pass-by) accounts
  - Similar to nested CBR but respondent bank allows its customers to access the correspondent account directly.
  - Targeted by global banks’ de-risking strategies due to higher compliance risks.

### Global policy targets and timeline
- G8 L’Aquila Summit (2009): commitment to reduce the global average cost of remittances to 5% (the 5x5 target), down from "around 10% prevailing at the time", in 5 years.
- G20 endorsement at Cannes Summit (2011).
- G20 reiterated commitment at Brisbane Summit (2014) to “... take strong practical measures to reduce the global average cost of transferring remittances to five percent...”; no new deadline set following unmet target.
- Inclusion in United Nations Sustainable Development Goals (SDGs) (2015): objective that by 2030 the global average cost of remittances should be lower than 3% and no corridor with costs above to 5% should exist.

### Remittance Prices Worldwide (RPW) database — scope, methodology, and vintage used
- Origin and purpose
  - World Bank launched the RPW database in September 2008 to promote reductions in remittance costs.
  - Initial coverage: 14 sending and 67 receiving countries, totaling 120 corridors.
  - The database has continuously grown and "today it approaches 400 corridors."

- Database vintage and coverage used in this paper
  - The vintage used included 48 sending and 110 receiving countries, covering 365 corridors.
  - RPW database vintage in figures refers to data up to the third quarter of 2018.
  - Sending-country corridors count: United States 42, United Kingdom 33, Germany 24.
  - Receiving-country corridors count: India 20, China 16, The Philippines 14.
  - The RPW database vintage used in this paper goes from the first quarter of 2011 to the third quarter of 2018.

- Survey methodology highlights
  - Data collected by independent researchers posing as customers; collection for each corridor occurs on the same day to control for exchange rate effects.
  - For each corridor, data are collected for major RSPs, including primary MTO, banks and post office where available.
  - Surveyed RSPs in each corridor aim at a representative sample totaling an aggregated market share of, at least 80, percent.
  - Two amounts surveyed for price comparisons: the local currency equivalent of USD200 and USD500.
  - This paper focuses on remittance costs associated with remitting USD200.

- Frequency and timing
  - Although the RPW survey initiated in 2008, the database is available from 2011.
  - The survey is carried out every quarter, though the specific month within the quarter varies by corridor.
  - The number of surveyed corridors increased over time; some corridors’ sample periods begin later (example: UAE-Sudan began second quarter of 2016).

### Data limitations and measurement issues
- RPW is survey-based and prone to measurement errors: reporting and recording errors, exacerbated by high granularity.
- Time-sensitive variables pose challenges:
  - Exchange rate margin depends on the actual exchange rate used (average, closing, etc.), which may differ from the rate used by the RSP when setting price.
  - In developing countries exchange rates can vary meaningfully within the same day, potentially explaining negative exchange rate margins found in the database and occasional large negative margins (outliers).
- Potential for non-random errors:
  - Non-response and inaccurate reporting are potentially serious problems; remittance pricing strategies are often not transparent and some firms have less incentive to fully disclose prices.
- Methodology changes:
  - The methodology changed in the second quarter of 2016: new categories created, others renamed, and one dropped.
- Survey periodicity:
  - Initially semi-annual; since the second quarter of 2013 the survey became quarterly.

*Source: wpiea2021199-print-pdf - Box 1. Types of Remittances Services*

### Section III explained, it might not be possible f or a RSP to disclose its f inal price when it

### wpiea2021199-print-pdf - Section III explained, it might not be possible f or a RSP to disclose its f inal price when it

### Data quality, RPW limitations, and measurement caveats
- The RPW database is “noisy” and requires scrutiny before making inferences.
- 3.7 percent of RPW database entries show negative exchange rate margins, raising accuracy questions.
- 23.1 percent show margins equal to zero; the RPW metadata notes that “The 0% in the exchange rate margin does NOT necessarily mean that there is no exchange rate cost, but rather that this cost is not disclosed to the sender at the time of sending.” and that “… this cost is not known by the sender at the time of sending”.
- The RWP database classifies more than 6 percent of the entries as not transparent due to exchange rate related issues.
- One percent of the entries shows zero total remittance cost.
- Evidence suggests at least one quarter of the entries seem to underestimate remittance costs.
- RPW may fail to capture fees charged at the destination; it is not possible to know how many entries are affected by measurement errors nor their severity.

### Why remittance costs matter and the SDG target
- High remittance costs increase inefficiency, incentivize informal channels, impair balance of payments statistics, and can hinder financial inclusion.
- Remittances finance education and health; higher costs reduce resources available for human capital and growth.
- SDG item 10.c: “By 2030, reduce to less than 3 per cent the transaction costs of migrant remittances and eliminate remittance corridors with costs higher than 5 per cent.”
- The SDG target is framed as a global average cost plus a corridor ceiling, but does not specify whether the average should be simple or weighted, the weighting scheme, or the level of aggregation.
- The first three RPW reports provided simple averages; since issue 4 (November 2012) a weighted average by size of flows has been published. The World Bank acknowledges that weights do not consider RSP market share.

### Measurement, aggregation, and trend evidence
- Figure 10 (described): global average cost around 9% in 2011; global costs falling slowly; mid-2015 trend reversal is largely an artifact of changing sample (latecomers from less developed economies facing higher costs).
- Using a balanced sample (same group of countries over time) shows stability rather than increase.
- Aggregation matters: computing global average from corridor-level averages shifts results downward in the second half of the sample; corridor averages implicitly overweight developed corridors with many corridors and RSP (e.g., top 7 receiving countries with most corridors have the same weight as the 33 SSA countries in RPW).
- Panels indicate costs stalled from mid-2015 to mid-2018 in several aggregations.

### Regional, country, and corridor heterogeneity
- Sub-Saharan Africa (SSA) is the most expensive region by far but has experienced faster falls in costs than other regions; by mid-2018 the gap to East Asia and the Pacific (EAP) had been drastically reduced.
- Regional grouping: lower cost group = Europe and Central Asia (ECA), Latin America and Caribbean (LAC), South Asia (SAR); middle cost group = EAP and MENA; SSA standalone highest-cost group.
- Figure 12 (2017Q4–2018Q3 average): top-15 most expensive vs least expensive countries:
  - Difference between least and most expensive group is almost fivefold.
  - 11 out of the 15 most expensive countries are from SSA.
  - 9 out of 15 of the least expensive are former Soviet Republics.
  - Only five countries (all former Soviet Republics) had average cost below 3% in 2018.
- Inter-country corridor variation (Figure 13): for large receivers (India, Philippines, China) and Nigeria the cheapest vs most expensive sending countries differ by up to 6-fold (India) or almost 3-fold (Nigeria).
- Top-30 most costly corridors: 8 out of 10 costliest corridors are in SSA; many costliest corridors originate in South Africa and Thailand; several costly corridors involve China and India as receivers.
- Top-30 least costly corridors: all 10 least costly corridors are between former Soviet Republics with Russia as sender; several least costly corridors have India as sender (SAR) and many originate in MENA countries.

### Provider types, payment methods, intra-corridor dispersion, and FX margins
- Provider type averages:
  - Banks are on average much more expensive than MTO; MTO are more expensive than post offices (PO).
  - Banks charge 50 percent more than MTO to send to China and Philippines and twice as much to send to India and Nigeria (not shown).
  - PO are often the cheapest provider type but not always.
- Payment instruments:
  - Cash is the most common payment method: around 40 percent of remittances involve cash at both ends.
  - Bank transfers are the second most popular method and have consistently been among the most expensive methods.
  - Less than one percentage of remittances are carried out using mobile money; this percentage remained constant across the two periods described.
- Intra-corridor variation:
  - Large intra-corridor cost dispersion exists; dispersion can be as large as inter-corridor differences.
  - Positive correlation between average corridor cost and variability: higher-cost corridors tend to have larger cost dispersion, implying potential large savings if migrants shop around.
- FX margin relevance:
  - For the 30 costliest corridors the average FX margin share is about 17 percent.
  - In some corridors the FX margin can be as high as 60 percent of total cost.
  - Panels comparing banks and MTO show average FX margin among top-30 FX-margin corridors is essentially the same for MTO and banks, but transparency differs (see Table 2).
- FX margin transparency (Table 2):
  - Zero FX margin reported: Banks 19%, MTO 7%, PO 53%.
  - Negative FX margin reported: Banks 2%, MTO 1%, PO 2%.
  - The percentage of banks reporting zero or negative FX margins is almost 3 times higher than MTO, suggesting banks are less transparent and that RPW may understate true global average remittance cost.

### Drivers of remittance costs (classification and evidence)
- Four broad categories: market structure, firm-specific, demand-related, regulatory factors (with some overlap between categories).
- Market structure:
  - Less competitive markets and concentrated banking systems tend to have higher prices.
  - Number of surveyed RSP per corridor shows a negative but insignificant correlation with average costs; breaking down by RSP type reveals:
    - Number of banks per corridor shows a positive correlation with average costs (counter-intuitive).
    - Number of MTOs per corridor shows a negative correlation with average costs.
    - The ratio of MTO to banks is strongly negatively correlated with costs, suggesting competition mainly occurs via specialized MTOs.
- Firm-specific factors:
  - Business models differ: banks provide many services (regulatory/compliance costs may be higher); MTOs historically had fragmented physical networks but broader geographical coverage via agents; fintechs are usually online.
  - Network and agent structures affect costs; technology is changing these dynamics.
- Demand-related factors:
  - Scale matters: remittance-receiving country GDP, number of migrants, and volume of remittance inflows are negatively correlated with average costs. Volume of inflows is most correlated.
  - Panels show negative correlations; outliers include former USSR republics which have much lower costs.
- Regulatory factors:
  - Exclusivity clauses by large MTOs act as barriers to entry and curb competition.
  - AML/CFT and de-risking (loss of correspondent banking relationships, CBR) have reduced CBR and active corridors: from early-2012 to end-2018 active correspondents shrunk by 20% and active corridors decreased by 10%.
  - Declines in CBR were geographically widespread; Americas (excl. North America) saw largest decline (~30%).
  - Increased concentration of correspondents correlates with slower falls or stagnation in global remittance costs around mid-2015 to early-2017.
  - Restrictions on institutions allowed to offer remittance services: IFAD notes in 80% of African countries regulators restrict remittance payments to banks; such restrictions limit competition and lead to partnerships that raise costs.
  - Over-stringent AML/CFT implementation (e.g., KYCC, proof of address requirements) and excessive compliance expectations have increased costs and reduced access.

### Empirical links between cost components and other factors
- Average remittance costs are positively correlated with both explicit fees and FX margins; fees show a tighter link but FX margins are strongly associated as well (corridor-level evidence).
- FX margin appears unrelated to exchange rate regime classification (IMF AREAER polar classification): average FX margins in fixed vs floating ER regime groups are almost the same; flexible ER group has lower cost dispersion (2.8 versus 4.1 standard deviation).
- Capital controls on transfers into by immigrants are rare and have only marginal association with average remittance cost.

### Policy recommendations and intervention channels
- Foster competition:
  - Prohibit or eliminate exclusivity clauses that lock agents/markets to single providers.
  - Broaden licenses to allow non-bank RSP (including MMOs, fintechs) to provide remittance services independently of banks.
  - Ensure proportionate, risk-based regulation for non-bank RSP appropriate to their risk profile, avoiding imposing bank-tailored burdens.
- Improve price transparency and consumer protection:
  - Require RSP to disclose cost components they know at time of transfer and clearly state unknown elements.
  - Promote collection and dissemination of timely remittance cost data for easy consumer comparison (e.g., PROFECO weekly statistics example in Mexico).
- Reduce barriers to entry and increase contestability:
  - Ensure non-bank RSP fair access to payment system infrastructure; promote interoperability between payment systems.
- Address AML/CFT and de-risking consequences with balanced approaches:
  - Apply proportionate AML/CFT implementation; avoid measures that go beyond FATF recommendations and that result in unnecessary CBR withdrawal.
  - Avoid imposing KYCC or proof-of-address requirements beyond international guidance where inappropriate for remittances.
- Reduce non-pecuniary and hidden costs:
  - Increase geographical coverage of payout locations to reduce travel/time burdens on beneficiaries; leverage postal networks and national post operators as extensive rural coverage (post office global network > 650,000; IFAD evidence).
  - Encourage post office partnerships in a non-discriminatory, competition-preserving manner (APFSI pilot results: PO-delivered remittances cost 37 percent lower than market average; PO market share increased; remittances delivered via PO increased by 60 percent in first three years of pilots).
- Ensure proportionate regulatory licensing and avoid excessive compliance requirements that disproportionately raise costs and act as entry barriers.

### Conclusions and overarching findings
- Remittance markets are complex, heterogeneous, and multi-layered; analogies with domestic payment systems are simplistic and potentially misleading.
- No single magic solution (e.g., fintech alone) will rapidly and substantially reduce remittance costs in the poorest regions; structural and endogenous factors limit the speed and magnitude of cost reductions.
- Some regularities emerge:
  - Banks are generally more expensive than MTO; MTO more expensive than PO.
  - Costs are positively associated with the number of banks in a corridor but negatively associated with the number of MTO and the MTO/banks ratio, highlighting that competition is mainly through MTOs.
  - Higher-cost corridors exhibit higher cost dispersion; active shopping can yield large savings.
  - FX margin is a major component of total cost and is not clearly related to exchange rate regime.
  - Costs are negatively associated with remittance volume, which explains challenges poorer countries face in lowering costs.
- Policy action is necessary but must be multifaceted: prohibit exclusivity clauses, adopt proportionate AML/CFT application, broaden licensing, enforce transparency, promote payment-system access and interoperability, and leverage postal networks to reduce both tag and hidden/non-pecuniary costs.

*Italic: Source — wpiea2021199-print-pdf (IMF working paper content provided).*

### REFERENCES

### wpiea2021199-print-pdf - REFERENCES

### Bibliographic coverage
- Compilation of scholarly and institutional references on remittances, correspondent banking, financial inclusion, and related policy debates, including works by Acharya and Leon-Gonzalez (2014); Adams (2011); Aggarwal, Demirgüç-Kunt, and Pería (2011); Amega (2018); Askarov and Doucouliago (2020); Azizi (2019); Barajas et al. (2009); Beck and Peria (2009); Bhattacharya et al. (2018); BIS (2020); Bettin, Presbitero and Spatafora (2014); Cazachevici, Havranek and Horvath (2020); Cecchetti and Schoenholtz (2018); Chami, Fullenkamp and Jahjah (2005); Cenfri (2018); Cloteau and Ansón (2011); CPMI (2016); CPSS and World Bank (2007); Donou-Adonsou, Pradhan and Basnet (2020); Erbenová et al. (2016); Lubambu (2014); Frankel (2011); FSB (2018); G8 and G20 communiqués; GFMF (2007); IFAD (2009, 2016, 2017, 2020); IFAD and World Bank (2015); IMF (2008, 2017, 2018); McKinsey (2016); Mohapatra, Joseph and Ratha (2012); Moré (2015); OECD/Scalabrini Migration Center (2017); Orozco and Yansura (2017); Sayeh and Chami (2020); Terrelonge (2014); The Guardian (2013); UN (2015); Watkins and Quattri (2014); World Bank (2007, 2014, 2016, 2018a, 2018b, 2020, 2021).

### Appendix 1 — Remittance costs and Exclusivity Agreements: key findings
- Central feature: pervasive use of exclusivity clauses by the two largest MTOs: Western Union (WU) and Money Gram (MG); IFAD (2009) assessed that “These agreements effectively ‘lock’ more than half of all available payout locations.”
- Market effect: exclusivity clauses act as a barrier to entry, allow incumbents to charge excessive prices, and “lock” agent networks (banks, foreign exchange bureaus, post offices).
- Early precedent: mid-2001 Russia—WU instructed correspondent Russian banks to leave a cash-to-cash transfer network citing item 4.2.8 (an exclusivity clause); Russlavbank filed a complaint under the Russian Ministry of Antimonopoly Policy.
- Russian ruling: September 2003 regulator ruled clause violated Art. 6 of the Federal Law “On Protection of Competition”; WU lost its final appeal in October 2004; exclusivity clause included in contracts with more than 200 Russian banks was ruled null and void.
- Ukraine case: 2004 Antimonopoly Committee investigation led to an agreement in which WU committed to reduce its fees by “... no less than 40%” in remittances sent from overseas to Ukraine.
- Regulatory actions in sending and receiving countries:
  - National Bank of Ethiopia Directive No FXD/30/2006: article 3.2.1 required RSPs to “arrange non-exclusive conditions when making agency agreements between them”; article 3.2.4 required RSPs to obtain NBE approval before new agency agreements.
  - Senegal (March 2008): Ministry of Finance letter warning exclusivity clauses were at odds with Competition Law (non-binding instrument).
  - Central Bank of Nigeria directive “BSD/DIR/CIR/GEN/VOL. 2/017” (November 2008): “will no longer approve any money transfer agreements ... that contain exclusivity clauses”; finding that those clauses “constitute a restraint on competition and unnecessarily increase the cost of money transfer services to the users”.
  - Reserve Bank of India circular “RBI/2010-11/206” (September 2010): ruled exclusivity arrangements for in-bound cross-border money transfers illegal, concluding they “limit competition, distort pricing and thereby act as a deterrent to a payment system fro m achieving the desired goal of efficiency, affordability and being ubiquitous”.
  - Central Bank of Liberia Directive “CBL/RSD/002/2011” (October 2011): required licensed financial institutions “having exclusivity clauses ... [to] ... expunge these clauses from their agreements...” and set a fine of “... not less than L$100,000 [approximately USD1,000] for each day of violation...”
  - Bank Al Maghrib (2011): banned exclusivity agreements in Morocco.
  - National Bank of Rwanda regulation N o 06/2012 (June 2012) prohibited exclusivity agreements for payment service providers; Regulation N o 06/2012 was repealed by Regulation N o 07/2015, which in its Article 32 stated “Contracts with exclusivity are not permitted by this regulation unless they are authorized by the Central Bank.”
  - Central Bank of Tunisia circular N°2013-01 (January 2013): instructed banks to remove exclusivity clauses at the request of the Secretary of State for Emigration.
- Multilateral guidance: CPSS and World Bank (2007) “General Principles for International Remittances Services” recommended that “Competition can be assisted by discouraging RSPs from imposing exclusivity conditions on agents”.
- Enforcement and effectiveness observations:
  - Discouraging exclusivity clauses without binding prohibition is insufficient; voluntary abandonment is unlikely because exclusivity is central to WU and MG business models.
  - Non-binding instruments (e.g., letters) are ineffective; prohibitions must be binding and enforced, including through financial penalties (example: CBN).
  - Even when exclusivity clauses are outlawed, market structure can persist: agents may continue to work with a single MTO, limiting the competitive impact (The World Bank, 2018b).
- Jurisdictional limits and antitrust actions:
  - No remittance-sending country has made exclusivity clauses illegal (as of the document).
  - US DOJ Antitrust Division investigation (2004–2005) concluded in 2005 that “... the facts do not support a claim that Western Union’s contracting practices with its agents in the United States have caused harm to consumers.”
  - European Commission opened a preliminary antitrust investigation in 2016 into WU exclusivity clauses; four years later the investigation remained at preliminary status.
- Market concentration and outcomes:
  - Watkins and Quattri (2014) reported that in more than two thirds of SSA countries WU and MG controlled at least 50% of the market.
  - Since 2011 many former USSR republics face the lowest remittance costs in the world; Russia has the lowest sending costs among its G20 peers.
  - The Ukraine example provides evidence of causal impact: agreement with WU produced a fee reduction of “... no less than 40%”.

### Policy implications and recommendations (as stated in the source)
- For durable reductions in remittance costs:
  - Make prohibitions on exclusivity clauses binding and enforce them, including by applying financial penalties.
  - Recognize that simply discouraging exclusivity is insufficient; regulators should adopt enforceable measures.
  - Address market structure and concentration because removing exclusivity clauses will only lower prices if it translates into more competition among MTOs and agents.
  - Use tools that can indirectly regulate international MTOs when direct jurisdiction is limited (example: Ethiopia’s requirement for non-exclusive agency agreements and prior approval).
  - Monitor post-prohibition market outcomes to ensure agents do not remain effectively exclusive to one MTO.

*Content derived from the "REFERENCES" section and Appendix 1 ("Remittance costs and Exclusivity Agreements") of wpiea2021199-print-pdf.*

### APPENDIX 2

### APPENDIX 2 — AML/CFT and the Loss of Correspondent Banking Relationships

### AML/CFT challenges in the remittances market
- The remittances market faces three key challenges for monitoring, identifying, and repressing illegal financial activities:
  - Most transactions are carried out by a couple of MTO, which usually face less regulation as non-banks, and most of those transactions occur in cash.
  - Identification requirements by MTO have been historically lax, and cash transactions leave little trail.
  - More stringent regulation and stricter enforcement, while necessary to curb illegal activities, have produced unintended consequences that harm remittances’ role as a lifeline for many poor families in developing countries.
- Since 2001, regulators have tightened “know your client” (KYC) requirements and enhanced enforcement driven by AML/CFT concerns.

### Enforcement actions and compliance cost increases
- High-profile enforcement and penalties prompted MTOs to strengthen compliance and monitoring:
  - WU settled on a massive $586 million fine imposed by the US Federal Trade Commission (FTC) and the DOJ after admitting “... criminal violations including willfully failing to maintain an effective anti-money laundering program and aiding and abetting wire fraud.”
  - In 2019 WU was fined in Europe in €1 million by the French regulator for AML rule violations.
  - MG was fined $125 million one year after the FTC and DOJ fined WU for failing to implement the comprehensive fraud prevention program mandated by the 2009 FTC order and for violating a 2012 deferred prosecution agreement with the Department of Justice.
- Large international bank fines raised compliance expectations and led banks to de-risk:
  - In 2012 HSBC agreed to pay USD1.9 billion in fines for channeling Mexican drug money.
  - In 2017 BNP Paribas admitted guilt for doing business with countries under US sanctions and agreed to pay a USD8.9 billion fine; it was also temporarily barred from dollar clearing.
  - Deutsche Bank was fined in 2017 in USD630 million for its role in a Russian money-laundering scheme.
- Consequences of enforcement and rising compliance costs:
  - Banks boosted internal controls and began pulling away from some customers, sectors, countries and regions (de-risking).
  - The compliance focus expanded from KYC to KYCC (know your customer’s customer) and even to KYCCC (know your customer’s customer’s customer), exceeding Financial Task Force requirements.

### Impact on correspondent banking relationships (CBR) and remittance market structure
- Mechanisms of impact:
  - Banks severed CBR not only with respondent banks but also with MTOs and some central banks, partly because MTOs’ lax controls made them particularly risky clients.
  - Increasing compliance costs made some CBR financially unattractive.
- Evidence of changing CBR network structure:
  - Despite more than 50% increase in the volume of remittances, there has been a steep decrease in the number of active correspondents (around 20%) since 2012.
  - There was a smaller but still relevant drop of 10% in the number of active corridors since 2012.
  - The CBR network has become increasingly concentrated (Gini coefficient on the number of active correspondents per corridor increased; calculation on a constant population of corridors).
  - The increase in active correspondent concentration as of early 2015 coincides with the intensification of CBR losses that took place 6 months earlier and with an interruption in the decrease in the global costs of remittance 6 months later.
- Caveats and uncertainties:
  - The effect of a more concentrated CBR network on remittance costs is not unambiguous:
    - Continued increase in total SWIFT message volume indicates possible gains of scale for correspondent banks.
    - SWIFT is not the only messaging system; many remittance transactions are carried out by MTOs that do not rely on SWIFT.
    - SWIFT data might not provide a reliable picture of the overall remittances market.

### Countries most affected by loss of CBR (2012–2018)
- The following list reports changes in the ratio of the number of banks abroad receiving payment messages from the stated country, and the number of banks receiving payment messages in the stated country, in percent. Most of the countries listed are poor and small states.
- Countries and percent changes (top 40 most dramatic losses):
  - Timor-Leste - 83.0
  - St Pierre and Miquel - 77.8
  - Mayotte - 71.4
  - Vanuatu - 69.0
  - Seychelles - 67.4
  - Syrian Arab Republic - 67.1
  - Tajikistan - 67.1
  - French Guiana - 65.7
  - Eritrea - 64.1
  - Jamaica - 59.7
  - Bermuda - 58.5
  - Monaco - 56.4
  - Gibraltar - 55.7
  - Cook Islands - 53.2
  - New Caledonia - 51.4
  - Belize - 51.2
  - Cuba - 50.8
  - Brunei Darussalam - 50.7
  - Virquin dsl (GB) - 50.0
  - Solomon Islands - 49.9
  - Yemen - 49.9
  - Faeroe Islands - 48.5
  - Venezuela - 48.5
  - Finland - 48.1
  - Angola - 47.8
  - Ghana - 46.8
  - Tonga - 46.2
  - Andorra - 46.0
  - Vatican City State - 45.1
  - Puerto Rico - 44.3
  - Montenegro - 44.3
  - Trinidad and Tobago - 44.2
  - Libyan Arab Jamahiriya - 44.1
  - Jersey, C.I. - 43.7
  - Isle of Man - 43.4
  - Fiji - 43.2
  - Bolivia - 43.1
  - Morocco - 42.5
  - Mauritius - 41.9
  - Moldova - 40.2

### Observations on distributional effects
- All regions have been affected by CBR withdrawal, but poor countries and conflict zones have been particularly hit.
- Example: In 2014 Westpac announced it would shut down its money transfer services to Somalia due to fears over breaching AML/CFT laws; Somalia is one of the most remittance-dependent countries.

*Source: APPENDIX 2, wpiea2021199-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021199-print-pdf.pdf_