## 1. Mobile Money Providers and Remittances Costs

## Source details

**Canonical URL:** [1. Mobile Money Providers and Remittances Costs](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021175-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021175-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021175-print-pdf.pdf.json)

---

### I. Executive summary — Key findings and implications
- Remittances are an important source of foreign currency income in Central America (CA), often exceeding income from tourism and/or foreign direct investment, and represent a critical lifeline for millions of Central American households.
- The paper reviews the U.S. remittance corridors with CA and explores potential for digitalization and fintech to enhance affordability, agility, reliability, and remote accessibility of transfers, aligned with the Bali Fintech Agenda (BFA) (IMF and World Bank, 2018).
- Findings:
  - Fintech advances and further digitalization can expand revenue reach to migrants’ families, contributing to lowering poverty and inequality and raising access to financial services; vulnerable households in rural CA, with young and educated women, are likely to gain the most.
  - Traditional operators provided remittance services at competitive prices by global comparison. In 2019, the average fee for a US$200 transfer to CA was the second lowest worldwide behind South Asia, but average remittance fees remained above the Sustainable Development Goal (SDG) of US$6 (for a US$200 transfer) in all CA countries.
  - Transformation toward fintech in CA has been more parsimonious than groundbreaking; alliances between traditional and new operators enabled more efficient back-end processes for traditional operators and increased access at payout for new operators. Traditional operators continue to dominate CA remittance markets; cost differential between incumbents and new entrants remains small.
  - COVID-19 is accelerating domestic and cross-border digital payments as consumers and governments shift toward digitalization.
  - Digital remittances can promote financial inclusion: reach remote, low-income households, provide faster payments than physical channels, and generate transaction information useful for creditworthiness assessment.
  - Regulatory trade-offs exist between proportionate KYC/AML/CFT requirements and easing digital services; technologies such as digital IDs can help reconcile digitalization and financial integrity.

### II. Digitalization, fintech definitions, and industry structure
- Definitions used:
  - Digital remittances: international money transfer services accessed and/or delivered through digital channels (internet, mobile phone).
  - Digital remittance operators: providers for which digitalized remittances are the main source of income (over 50 percent of revenues).
  - Fintech remittance operators: firms incorporating technology advances that can potentially transform provision of financial services through new business models, applications, processes, and products.
- Industry segments and constraints:
  - Front-end: capturing and disbursing; Back-end: messaging and settlement.
  - Traditional concentration drivers: high fixed costs from physical agent locations; inadequate interoperability; network externalities; better liquidity and FX risk management with higher volumes; AML/CFT and KYC compliance add to fixed costs.
- Fintech enhancements:
  - Front-end: mobile money and online wallets with payment visibility and tracking.
  - Back-end: DLT/blockchain token-based payments as decentralized alternatives to correspondent banking.
  - Transaction benefits: faster (real-time settlement), certain and transparent (rich information disclosure, Big Data/AI), cost-effective (lower processing and liquidity costs, on-demand liquidity, higher STP rates).

### III. Remittances costs, competition, and mobile-money effects
- Global and regional cost findings:
  - The cost of sending remittances in 2019 stood at US$30 billion globally, or 7 percent of total remittances.
  - Over the past decade, the average fee fell by 21 percent in LAC and 17 percent in CA, below the 26 percent world decline.
- By mode:
  - Card-mode declines were widespread in LAC: more than 30 percent over 2011−13 and 20 percent over 2017−2019.
  - Cash remittance fees remained relatively unchanged in LAC and CA over the same period.
- Mobile money RSPs and competition (2016−19):
  - Sample: 354 corridors, 56 with mobile money providers and 298 without.
  - Corridors with mobile money RSP show a leftward shift in cost distribution relative to corridors without mobile money RSP.
  - For a US$200 delivery, differential cost between corridors with and without mobile money services providers was most pronounced for bank transfers and least marked for cash remittances.
  - In corridors with mobile money operators, the cost of traditional RSP providers fell, on average, by 30 cents to every 10-cent reduction in mobile money RSP costs over 2016−19.
  - Even with declines, a sizable cost gap in 2019 remained between traditional delivery modes and mobile remittances for a US$200 delivery.
- Volume-differentiated fees (2016−19):
  - Share of corridors applying volume-differentiated fees declined from 39 to 32 percent for cash corridors and from 37 to 21 percent for non-cash corridors.
- Illustrative operator sample: Guatemala (2019)
  - Sample includes 159 thousand payout points, seven fintech operators, seven non-fintech operators, and six hybrid operators.
  - The average cost of sending remittances to Guatemala was 4.7 percent in 2019 overall.

### IV. U.S.−CA corridor stylized facts and modes of transfer
- Scope: CA comprises Belize, Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, and Panama.
- Aggregate corridor flows:
  - Remittances from CA migrants to the United States represent 82 percent of the total U.S.−CA corridor inflows; the U.S.−CA corridor had an inflow of US$24 billion in 2018 (7 percent of its GDP), making it the second largest corridor after U.S.−Mexico.
  - Within the U.S.−CA corridor, Guatemala accounts for 35 percent of total remittances, followed by the Dominican Republic and El Salvador at 20 percent each.
- Operators and modes:
  - RSP in LAC increased from an average of 14 providers per country in 2010 to 20 in 2016.
  - Mobile money remittances in 2019: about 0.15 percent of all transfers for LAC (versus 14.3 percent in SSA and 1.2 percent in EAP).
  - Origination: 80 percent of migrants send remittances through agent-based interactions and pay with cash.
  - Delivery heterogeneity: almost half received via bank accounts or cashed out at bank branches; the other half via traditional chain stores and MTOs allied with local retail stores or transfer companies.
  - Overall digitalization has been slow pre-COVID; COVID-19 shifted preferences toward digital remittance and payments (examples: Zelle payments value and volume in the U.S. grew around 60 percent during the first half of 2020; Tigo Money’s digital transactions in LAC almost doubled during Q2 2020).

### V. Empirical analysis of remittance-fee drivers (firm perspective)
- Sample and scope:
  - Focus on three operators: one traditional MTO, Western Union, and two digital operators, Xoom and Remitly.
  - Sample covers 31 corridors (of which 15 LAC corridors) originating from the U.S.
  - Modes considered: card-to-cash and bank transfer-to-cash for US$200 and US$500 remittances.
- Reduced-form pricing equation (preserved specification):
  - ln(ffs)i = α0 + α1 ln(amount)i + α2 ln(fee competitor)i + α3 ln(remittances)i + α4 Xi + εi
- Key regression findings:
  - Fees are differentiated by remittance amount and mode; evidence of fee differentiation (or lack of) is weak for cash-to-cash and transfer-to-cash in LAC.
  - Fees are responsive to competitors’ fees: elasticity with respect to competitor fees is significant and large in LAC, implying price sensitivity across firms.
  - Some evidence that higher remittance corridor volumes are associated with lower fees (negative ln(remittances) coefficients in several specifications).
  - Greater financial development and deeper financial institutions are associated with lower remittance fees in several specifications, though results are not robust across all firms and indices.
  - Specific quantitative finding: an improvement in the financial system development index by 10 percent could allow for a reduction in remittance fees from 2 to 3 percent across the three firms.
- Sample fee levels (selected table entries for US$200 transfers, 2019):
  - LAC totals: Xoom 9.0 Remitly 5.6 Western Union 9.8 (Total row).
  - Country examples (Xoom Remitly Western Union):
    - Argentina 13.5 7.1 20.0
    - Guatemala 7.0 6.6 5.5
    - El Salvador 13.7 8.5 9.8
    - Honduras 13.7 5.2 10.0
    - Mexico 7.0 2.0 5.5
  - Non-LAC examples:
    - Bangladesh 5.7 4.0 5.5
    - China 12.5 4.0 28.5
    - India 7.0 4.0 5.5
- Caveats highlighted:
  - Inclusion of a larger number of MTOs, financial institutions, and fintech operators would allow more systematic differences in price setting to emerge.
  - Time-varying data would better capture firms’ responses to new entrants or competitor price changes.
  - Data limitations restrain sample size for subregions, precluding a dedicated CA analysis in regressions.
  - Important control variables (regulation, payment network extent, payment system efficiency) are excluded and may bias estimates.

### VI. Remittances and financial inclusion in CA — empirical evidence
- Probit regressions: two specifications
  - Household-level Remittance-receivers_i = α + βX_i + μ_i using IADB household surveys (2016) for GTM, SLV, DR, HND.
  - Individual-level Account_i = α + βX_i + μ_i using Findex 2017 (1000 respondents per country).
- Who receives remittances (selected marginal effects / coefficients preserved):
  - Woman-headed households: more likely to receive remittances by 13, 8 and 7 percentage points in Guatemala, El Salvador and Dominican Republic, respectively (probit coefficients Head Female: GTM 0.13***; SLV 0.08**; DR 0.07**; HND 0.14*).
  - Rural households more likely to receive remittances by 4 and 8 percentage points in Guatemala and El Salvador, respectively (Head Rural: GTM 0.04***; SLV 0.08***).
  - Head with primary education are less likely to receive remittances relative to those with middle and university schooling (Head Prim-educ negative and significant across countries).
  - Head unemployed increases probability of receiving remittances (Head Unemployment: GTM 0.06***; SLV 0.18***; DR 0.18***; HND 0.14***).
  - Neighbors receiving remittances increases household’s chance (Neighbors positive and significant).
- Financial inclusion patterns:
  - Odds of being financially included higher for male, older, urban individuals; lower for less educated and younger individuals.
  - Individuals with primary education are less likely to have an account by 39 and 32 percentage points in Guatemala and the Dominican Republic, respectively (Primary-Educ: GTM -0.39***; DR -0.32***).
  - Being unemployed reduces chances of having an account by 16 percentage points in Guatemala and El Salvador, and by 12 and 18 percentage points in Dominican Republic and Honduras, respectively (Unemployment negative and significant across countries).
  - Owning a mobile phone increases likelihood of having an account in all four countries (Owns Mobile phone: GTM 0.14***; SLV 0.20***; DR 0.19***; HND 0.21***).

### VII. Supply, demand, regulatory conditions, and country case studies
- Technology and uptake:
  - Smartphone use widespread among Hispanic migrants; mobile and 3G network coverage in CA match LAC coverage for most CA countries.
  - Consumer digital uptake moderate due to behavioral inertia, informal employment at origination, and inadequate financial literacy at receiving end.
  - At origination only 20 percent of migrants use online banking platforms; on receiving end over two thirds in Guatemala, Honduras, and Nicaragua and over a third in El Salvador and Dominican Republic have yet to experience the mobile internet.
- Regulatory environment (GSMA MMRI assessments):
  - CA scores poorly on authorization, transaction limits, and KYC regulations; fares well on infrastructure, agent network, and consumer protection dimensions.
  - Country-specific weaknesses:
    - Dominican Republic and Nicaragua score poorly on authorization requirements, including proportionality of capital requirements.
    - El Salvador and Nicaragua score poorly on infrastructure and investment environment (sector-specific taxation, ID verification infrastructure, interoperability infrastructure, utilization of interest income).
    - Guatemala scores poorly on consumer protection (complaint procedures, providers’ disclosure, safeguards for customer funds).
    - Several CA countries score poorly on transaction limits and proportionality of KYC and AML/CFT requirements by user risk profile.
- Country case studies:
  - Guatemala (2019 stylized facts):
    - Remittances inflows US$10½ billion (over 13 percent of GDP; 90 percent from the U.S.).
    - Remittances account for about 45 percent of recipient households’ income (one in ten households).
    - Digital origination is nascent (0.3 percent of total); cash pick-up dominates (89 percent).
    - Banks play a large role in payout (91 percent of total payout, of which 80 in cash); 28 bank branches per 1,000 km2.
    - Tigo Money: as of June 2020, 4.1 percent of Tigo’s customer base had active mobile financial services; receiving cross-border payments through Tigo’s e-wallet represents about 2 percent of total remittance inflows.
    - Only 44 percent of population has an account with a financial institution or mobile money provider.
    - Regulatory gaps: mobile financial services regulated as services channelled through supervised financial institutions; lack of dedicated MFS framework; mobile money providers lack direct access to national payments and settlement system.
  - Honduras (2019 stylized facts):
    - Remittances US$5.4 billion (22 percent of GDP; about 85 percent from the United States).
    - Remittances mainly via traditional MTOs, followed by banks; virtually all transactions originated in cash.
    - Tigo Money licensed in 2019; Tengo electronic wallets created in 2013.
    - In 2019, 17.6 percent of Tigo Money transactions and 8.8 percent of Tengo transactions were to receive remittances, but together accounted for an estimated 1 percent of total remittances.
    - 2019 average cost of sending remittances from the U.S. to Honduras was about 3¾ percent for a $200 transfer (World Bank), slightly above the SDG target.
    - Senders reported average costs of 2 to 3 percent during the last two years (Central Bank of Honduras’ remittances survey).
    - Digitally-originated remittances declined about 45 percent over the last decade; card-funded transactions remain least cost-effective.
    - Recipient financial exclusion: only 17 percent of remittance recipients had a checking account; 92 percent had a mobile phone; nearly two-thirds used the internet.
    - Regulatory constraints: 2016 mobile money regulation is a welcome step but not well defined; high requirements for agents and authorized transactional centers; high minimum capital requirement 30 million Lempiras (about US$1.2 million) could be a barrier; transactional limits set at 30,000 Lempiras a month (around US$1,250); virtually no interoperability between mobile money and banks’ payment systems.

### VIII. COVID-19 impacts on digital remittances and payments
- Pandemic effects:
  - Western Union and MoneyGram recorded around 100 percent expansion of digital channels during Q2 2020, accounting for around 22 percent of customer-to-customer revenues (versus 14 percent before the pandemic).
  - Governments’ social protection responses promoted digital payment adoption:
    - Honduras launched Bono Único: vouchers of 4,000 Lempira to more than 70,000 households delivered electronically (mobile phone text message).
    - Guatemala provided Bono Familia: cash transfers of up to 2,600 quetzals through electronic transfers and tokens to vulnerable households.

### IX. Policy implications and recommendations
- General recommendations:
  - Expand use of mobile financial services and promote adequate regulation to make digital financial services more cost-effective.
  - Maintain use of digital means for social welfare payments beyond the pandemic to sustain adoption.
  - Regulators and public authorities should ease transfers while safeguarding customer protection, financial stability, and integrity.
  - Facilitate entry of new operators and broaden collaborations between telecommunication and fintech companies, banks, and traditional MTOs to expand digital financial services and inclusion.
- Risk-sensitive regulatory principles:
  - Continuous risk assessment: monitor product risks targeted at underserved groups and gradually ease KYC and CDD as warranted.
  - Risk mitigation via product design: entry-level mobile financial products should limit features or adopt progressive CDD; transactional limits could be used depending on CDD scope and customer risk profiles; additional features (credit scoring, lending, insurance) could be unlocked after a low-risk track record under simplified KYC/CDD.
  - Appropriate consumer protection: establish initial capital requirements; separate customers’ funds from providers’ capital in trusts; encourage interoperability to promote competition, reduce fixed costs and unlock economies of scale.
- Country-specific regulatory advice:
  - Guatemala: regulate e-money issuance to provide a level playing field; widen scope and reach of operations (e.g., interoperability); promote risk-based CDD.
  - Honduras: refine existing regulations to broaden use of mobile financial services, improve interoperability, and adopt proportionate risk-based customer-due diligence.

*Source: wpiea2021175-print-pdf*

### 1. Mobile Money Providers and Remittances Costs ___________________________________________________ 15

### 1. Mobile Money Providers and Remittances Costs

### Figures
- 1. Cross-Border Money Transfer Volume by Category ________________________________________________ 11
- 2. Supply, Demand and Regulatory Conditions of Remittances Digitalization in CA _______________ 25

### Annexes
- I. Impact of Remittances on Households’ Income _____________________________________________________ 37
- II. Probit Models: Data and Regression Results ________________________________________________________ 41
- III. Mobile Money Remittances: Regional Coverage ___________________________________________________ 45
- IV. Key Drives of Remittances’ Fees: A Firms Perspective _____________________________________________ 46

*Source: wpiea2021175-print-pdf - 1. Mobile Money Providers and Remittances Costs ___________________________________________________ 15*

### REFERENCES _____________________________________________________________________________________________33

### wpiea2021175-print-pdf - REFERENCES _____________________________________________________________________________________________33

### List of Acronyms
- AI  Artificial Intelligence
- AML  Anti-Money Laundering
- API  Application Programming Interface
- ATM  Automated Teller Machine
- CA  Central America
- CFT  Combating the Financing of Terrorism
- CPMI  Committee on Payments and Market Infrastructures
- DLT  Distributed Ledger Technology
- EAP  East Asia and Pacific
- ECA  Europe and Central Asia
- EMDE  Emerging Market and Developing Econonomies
- GSMA  Global System Mobile Association
- KYC  Know Your Customer
- LAC  Latin America and the Caribbean
- LIC  Low Income Countries
- MENA  Middle East and North Africa
- MMRI  Mobile Money Regulatory Index
- MNO  Mobile Network Operator
- MTO  Monetary Transfer Operator
- NA  North America
- P2P  Person-to-Person
- SA  South Asia
- SSA  Sub-Saharian Africa

### I. EXECUTIVE SUMMARY — Key findings and implications
- Remittances are an important source of foreign currency income in Central America (CA), often exceeding income from tourism and/or foreign direct investment, and represent a critical lifeline for millions of Central American households.
- The paper reviews the U.S. remittance corridors with CA and explores potential for digitalization and fintech to enhance affordability, agility, reliability, and remote accessibility of transfers, aligned with the Bali Fintech Agenda (BFA) (IMF and World Bank, 2018).
- IMF engagement: the Fund collaborates with the Committee on Payments and Market Infrastructures (CPMI) on a global roadmap to improve cross-border infrastructure and payments at the request of the G20; participation includes CPMI-IOSCO working group on digital innovations.

Findings:
- Fintech advances and further digitalization in the remittance industry can be expected to expand revenue reach to migrants’ families, contributing to lowering poverty and inequality and raising access to financial services; vulnerable households in rural CA, with young and educated women, are likely to gain the most.
- Traditional operators provided remittance services at competitive prices by global comparison. In 2019, the average fee for a US$200 transfer to CA was the second lowest worldwide behind South Asia. Nonetheless, average remittance fees remained above the Sustainable Development Goal (SDG) of US$6 (for a US$200 transfer) in all CA countries. Evidence suggests greater competition could further decrease remittance fees.
- Transformation toward fintech in CA has been more parsimonious than groundbreaking; alliances between traditional and new operators have enabled more efficient back-end processes for traditional operators and increased access at payout for new operators. Traditional operators continue to dominate CA remittance markets; cost differential between incumbents and new entrants remains small.
- COVID-19 is accelerating domestic and cross-border digital payments as consumers and governments shift toward digitalization.
- Beyond cost reduction, fintech can promote financial inclusion: digital remittances can reach remote, low-income households, are faster than physical channels, and transaction information can complement traditional banking instruments for creditworthiness assessment.
- Options to foster fintech remittances include expanding mobile financial services (nascent in CA) and broadening alliances of telecommunications and fintech providers with traditional operators; both likely involve greater financial development and inclusion in CA. Public and private sectors should collaborate to enhance digital and financial literacy.
- Regulatory environment: room exists for regulators to enable digitalization while guarding against risks. Internet coverage is broadly adequate in CA, but financial literacy and digitalization—particularly mobile financial services penetration—lag. Expansion of digital remittances requires level-playing-field regulation for digital financial services and greater interoperability with payment systems. Trade-offs exist between proportionate KYC/AML/CFT requirements and easing digital services; technologies such as digital IDs can help reconcile digitalization and financial integrity.

### II. INTRODUCTION — Context and regional significance
- Workers’ remittances are a major financing source for many CA countries, reaching levels comparable with foreign direct investment or official development assistance.
- Remittances from CA migrants to the United States represent 82 percent of the total U.S.−CA corridor inflows; the U.S.−CA corridor had an inflow of US$24 billion in 2018 (7 percent of its GDP), making it the second largest corridor after U.S.−Mexico.
- Within the U.S.−CA corridor, Guatemala accounts for 35 percent of total remittances, followed by the Dominican Republic and El Salvador at 20 percent each.

Literature summary — documented effects of remittances:
- Stabilization role: remittances are often countercyclical with recipient country business cycles, complement external flows and support financial stability (Spatafora, 2005; Frankel, 2011; Bettin et al., 2014). In times of short-lived external shocks, remittances tend to dampen output volatility, relatively more so in LAC (Beaton et al., 2017).
- Growth effects: remittances have positive (though not always statistically significant) growth effects, largest in high-remittance-receiving subregions (Beaton et al., 2017). In LAC, countries with lower long-term average growth rate of real GDP per capita have received higher remittances as a share of GDP.
- Use of remittances: remittances are not associated with substantial increases in domestic investment but are rather directed to consumption (Barajas et al., 2009); substitution of remittance income for labor income and liquidity constraints may limit remittance-growth channels.
- Impact on inequality: remittances are found to decrease income inequality significantly by about 5 to 6 points in those CA countries that are large remittance recipients (Annex I).

Digitalization context:
- Fintech startups aim to make payments more secure and less expensive, leveraging high mobile phone ownership versus bank account ownership. Underbanked segments in Asia and Africa lead adoption; evidence suggests ample scope to enhance remittances’ digitalization in CA.
- Pre-COVID, traditional cash-based remitters had begun diversifying into digital transfers but adoption was slowed by inadequate market competition and behavioral inertia. COVID-19 accelerated digital uptake: Zelle payments value and volume in the U.S. both grew around 60 percent during the first half of 2020; Tigo Money’s digital transactions in LAC almost doubled during the second quarter of 2020.

Paper structure (sections summarized):
- Section III defines digital and fintech remittances.
- Section IV presents evidence of low remittances digitalization in U.S.−CA corridor and discusses underlying reasons (including regulatory impediments discussed in Section VII).
- Section V analyzes key drivers of remittance fees using data from one traditional MTO and two digital providers operating in about 30 U.S. corridors.
- Section VI assesses financial inclusion for remittance receivers in CA.
- Section VII uses country case studies to discuss supply, demand, and regulatory conditions supporting mobile-based financial remittances.
- Section VIII concludes with policy recommendations to foster digital remittances and mobile money financial inclusion.

### III. WHAT ARE FINTECH REMITTANCES?
Definitions used in the paper:
- Digital remittances: International money transfer services accessed and/or delivered through digital channels (internet, mobile phone). Digital remittance services usually employ agents and networks of third-party intermediaries to improve accessibility and lower delivery cost (Alliance for Financial Inclusion, 2016).
- Digital remittance operators: providers for which digitalized remittances are the main source of income (over 50 percent of revenues).
- Fintech remittance operators: firms incorporating technology advances that can potentially transform provision of financial services through new business models, applications, processes, and products (BFA, IMF and World Bank, 2018; Berkmen et al., 2019). Fintech firms have an efficiency edge over traditional firms, contesting extraordinary margins.

Remittance industry segments:
- Front-end: capturing and disbursing (interface with migrants and recipient households).
- Back-end: messaging (handling payment instructions) and settlement (domestic settlement via central bank reserves; cross-border settlement via correspondent banks).
- Traditional concentration drivers: (i) high fixed costs from physical agent locations (capturing, disbursing); (ii) inadequate interoperability (messaging); (iii) network externalities with higher transaction volumes (domestic settlements); (iv) better liquidity and FX risk management with higher volumes (cross-border settlements). AML/CFT and KYC compliance also add to fixed costs.

Fintech enhancements:
- Front-end improvements: mobile money and online wallets provide payment visibility and delivery tracking.
- Back-end innovations: DLT or blockchain token-based payments as decentralized alternatives to correspondent banking.
- Resulting transaction benefits: (i) faster, via real-time settlement and bidirectional messaging between banks; (ii) certain and transparent, via disclosure and validation of rich information prior to settlement (supported by Big Data and AI), facilitating AML/CFT and KYC compliance and credit scoring; (iii) cost-effective, through lower processing and liquidity costs (on-demand liquidity provision, high STP rates, reduced reliance on nostro accounts).

Global market structure:
- Traditional Money Transfer Operators (as of 2Q2019): Western Union and MoneyGram (200+ countries each, about 500,000 and 347,000 agents worldwide, respectively) continued to lead cross-border remittances.
- Both Western Union and MoneyGram have formed partnerships with blockchain/DLT platforms (e.g., Thunes, Ripple) to enable mobile wallet transfers and real-time messaging, clearing and settlement.
- Fintech money transfer operators (2010s entrants): TransferWise (P2P, developed-country focus), WorldRemit (African and working-class migrants), Remitly (US-based immigrants). Cost-effectiveness sources include: (i) matching transfers in opposite directions (TransferWise); (ii) price-differentiation by volume and speed (Remitly, fee-free for large transfers to India of US$ 1,000 or more); (iii) focus on mobile-to-mobile remittances (WorldRemit). These fintechs build global networks with financial institutions and mobile telecoms, enable accurate tracking, near-instant payments, and use AI/machine learning for identification and regulatory compliance.

### IV. U.S. TO CA REMITTANCES CORRIDOR: STYLIZED FACTS
Scope: CA comprises Belize, Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, and Panama.

A. Operators
- RSP in LAC increased from an average of 14 providers per country in 2010 to 20 in 2016 (Orozco et al., 2016); distribution networks in top 11 LAC remittance-receiving countries increased tenfold since 2009 and doubled since 2012 to an estimated 500 thousand payment locations in 2016.
- Financial institutions and traditional MTOs continue to dominate LAC remittance markets (Berkmen et al., 2019). Traditional MTOs have been gaining market share against banks following regulatory changes favorable to retail agent expansion.
- Mobile money remittances in 2019: about 0.15 percent of all transfers for LAC (versus 14.3 percent in SSA and 1.2 percent in EAP), and much smaller than LAC’s share in world remittances of 15 percent.
- Digital remittances in LAC are primarily offered by TransferWise (P2P) and WorldRemit (partnerships with MNOs, MTOs, banks). LAC fintech remittance startups were less than 1 percent of total startups in 2018.

B. Modes of Transfer
- Pre-COVID survey data for CA show dominance of physical locations and cash at origination and heterogeneity in delivery methods.
- Origination: 80 percent of migrants send remittances through agent-based interactions and pay with cash; Western Union, MoneyGram, and Ria dominate agent-based transactions. Financial sector has a limited role in originating U.S. remittances; digital-originated transactions (e.g., Xoom) remain nascent.
- Delivery heterogeneity: almost half of remittances are received via bank accounts or cashed out at bank branches; the other half via traditional chain stores and MTOs allied with local retail stores or transfer companies. Country-specific factors matter (e.g., El Salvador’s large banking geographic outreach increases financial-sector payout share). Xoom and Remitly offer cash pick-up in chain stores (Elektra in Honduras and Guatemala; Super Selectos in El Salvador), in microfinance institutions (MiCoope in Guatemala; Fedecredito in El Salvador; Intibucana in Honduras), or cash home-delivery (Dominican Republic).
- Overall digitalization has been slow: traditional remitters diversified into digital but adoption impaired by inadequate competition and behavioral inertia; COVID-19 shifted preferences toward digital remittance and payments.

C. Remittances Costs
- Cost comparability: available data allow comparisons across world regions; costs depend on speed, mode of transfer, amount transmitted, and corridor size.
- Focus: same-day or less US$ transfers, prominent transfer type.
- 2019 averages: the average fee for a US$200 transfer to both LAC and CA was the second lowest worldwide behind South Asia. Within LAC, average fees were highest for transfers to Paraguay (twice as much as low-fee transfers to Colombia, Ecuador and Mexico).
- Average fees to CA are lower than for LAC overall (but still higher than fees directed to Mexico), reflecting dominance of less-costly cash-to-cash transfer modes in CA.
- Despite relatively low averages, remittance fees in all LAC countries remained above the SDG goal of 3 percent (or US$6 for a US$200 transfer).
- Trend: average fee for a US$200 transfer to both LAC and CA had been gradually falling over the past decade alongside rising number of RSPs (traditional and fintech) and tighter competition, mirroring global trends as digital operators entered the market.

*Source: wpiea2021175-print-pdf - REFERENCES _____________________________________________________________________________________________33*

### Box 1 for mobile money providers in

### Box 1. Mobile Money Providers and Remittances Costs

### Key findings on global and regional remittance costs
- The cost of sending remittances in 2019 stood at US$30 billion globally, or 7 percent of total remittances.
- Over the past decade, the average fee fell by 21 percent in LAC and 17 percent in CA, somewhat below the 26 percent world decline, given their lower initial costs.
- Within LAC, the fall in remittance fees was driven by Bolivia, Colombia, and Paraguay.
- By mode:
  - The decline in the card mode of transfer was widespread in LAC and particularly marked over 2011−13 (more than 30 percent) and 2017−2019 (20 percent).
  - Fees charged for cash remittances—starting from already low levels in 2010—remained relatively unchanged in LAC and CA during the same period.

### Effects of mobile money RSPs on competition and costs (2016−19)
- Mobile money RSPs would seemingly increase the degree of competition in those corridors where they operate (GSMA, 2016). The cost distribution for corridors with mobile money RSP shows a leftward shift of the same distribution for corridors without mobile money RSP (box Figure 1).
- For a US$200 delivery, the differential cost between corridors with and without mobile money services providers was the most pronounced for bank transfers, and the least marked for cash remittances—the riskier mode of transfer given the difficulty to verify identity and comply with KYC and AML/CFT regulations.
- In corridors with mobile money operators, the cost of traditional RSP providers proved elastic to that of mobile money RSP. It fell, on average, by 30 cents to every 10-cent reduction in mobile money RSP costs over 2016−19. Such decline was most pronounced for bank transfers but was also apparent for the cash and card modes.
- Even with declines, there remained a sizable cost gap in 2019 between traditional delivery modes and mobile remittances for a US$200 delivery.

### Volume-differentiated fees and recent trends (2016−19)
- The share of corridors with volume-differentiated fees has decreased over the past three years.
- While price differentiation remains between the two most common transfer amounts of US$200 and US$500:
  - The share of corridors applying volume-differentiated fees over 2016−19 declined from 39 to 32 percent for cash corridors.
  - The share declined from 37 to 21 percent for non-cash corridors.

### Sample and cost-distribution notes
- The sample includes 354 corridors, of which 56 with mobile money providers and 298 without.
- Regional/type split (corridors without/with mobile money providers):
  - East Asia and Pacific 74 (56/18)
  - Europe and Central Asia 50 (50/0)
  - Latin America and Caribbean 41 (41/0)
  - Middle East and North Africa 37 (35/2)
  - South Asia 68 (55/13)
  - Sub-Saharan Africa 84 (61/23)

### Fintech remittances and channel-cost nuances
- Fintech remittances in the region do not necessarily cost less:
  - A survey-based study (Martin et al., 2019) covering selected U.S. corridors to LAC and CA countries found that, initiating a remittance in the U.S. towards the Dominican Republic and El Salvador using cash at a brick-and-mortar location is cheaper, on average, than originating it online (or with a mobile app) and funding through a bank account.
  - The opposite holds for transfers to Mexico and Colombia, although the cost advantage of digital channels over brick-and-mortar locations can be quite small.

### Illustrative operator sample: Guatemala (2019)
- The sample includes 159 thousand payout points in 2019 and comprises seven fintech operators, seven non-fintech operators, and six hybrid operators.
- The average cost of sending remittances to Guatemala was 4.7 percent in 2019 overall, and 4.3, 4.6, and [data truncated in source].

*Source: IMF staff summary of Box 1 content from the provided PDF chapter/section.*

### 5.2 percent for non-fintech, fintech,  and

### wpiea2021175-print-pdf - 5.2 percent for non-fintech, fintech,  and

### V. Empirical analysis of remittance fees’ drivers
- Remittance fees to CA countries remain above the SDG goal of 3 percent, or US$6, for a transfer of US$200.
- Sample and scope:
  - Focus on three operators: one traditional MTO, Western Union, and two digital operators, Xoom and Remitly.
  - Sample covers 31 corridors (of which 15 LAC corridors) originating from the US.
  - Two modes of transactions considered: card-to-cash and bank transfer-to-cash, for both a US$200 and a US$500 remittance.
- Key regression findings (see Annex IV for detailed tables):
  - Fees are differentiated by remittance amount and mode.
  - Evidence that remittance companies differentiate fees in response to:
    - Competitors’ fees (elasticity generally higher for card-to-cash than for bank transfer-to-cash).
    - The size of the corridor (some evidence that higher remittance corridors have lower fees, indicating economies of scale).
    - The level of financial development and overall competition in host and recipient countries.
  - Specific quantitative finding: an improvement in the financial system development index by 10 percent could allow for a reduction in remittance fees from 2 to 3 percent across the three firms.
- Other empirical notes:
  - For CA countries where the FX margin plays a role (e.g., Guatemala and Dominican Republic), digital providers hold a comparative advantage in origination fees, while cash-intensive traditional MTOs are most competitive in FX rate at payout.
  - In dollarized (El Salvador) or crawling peg (Honduras) countries, traditional providers offer more competitive fees than digital operators.
  - Non-fintech operator pricing example: Wells Fargo offered the lowest average cost (3.1 percent through Bank account transfer) and Western Union the highest (7.2 percent), although Western Union’s cash option was cheaper (4.6 percent).

### VI. Remittances and financial inclusion in CA
- Literature context:
  - Remittances positively associated with credit- and deposit-to-GDP ratios (Aggarwal et al., 2011).
  - Higher remittances broaden financial inclusion and economic development in middle-income countries (Tu et al., 2019).
  - Financial access and deepening support growth and lower income inequality (Sahay et al., 2015a; Čihák and Sahay, 2020).
  - Digital payments yield growth effects via lower transaction costs and higher liquidity and creditworthiness (Sahay et al., 2020).
- Empirical strategy for CA (four largest remittance-receiving countries: El Salvador, Guatemala, Honduras, Dominican Republic):
  - Two probit regressions estimated:
    - Remittance-receivers_i = α + βX_i + μ_i (household-level, 2016).
    - Account_i = α + βX_i + μ_i (individual-level, Findex 2017).
  - Remittance-receivers_i is a dummy = 1 if household receives remittances.
  - Account_i is a dummy = 1 if an individual has an account (financial institution, post office, or elsewhere).
  - Vector X_i includes household socioeconomic characteristics; in account equation expanded to include remoteness to financial institutions and mobile phone use.
- Empirical findings on who receives remittances:
  - Higher chances for rural households where head is a young and educated woman.
  - Woman-headed households are more likely to receive remittances by 13, 8 and 7 percentage points in Guatemala, El Salvador and Dominican Republic, respectively.
  - Rural households more likely to receive remittances by 4 and 8 percentage points in Guatemala and El Salvador, respectively.
  - Households’ head with primary education are less likely to receive remittances relative to those with middle and university schooling.
  - Chances of receiving are higher for households’ heads who are unemployed in all countries.
  - Remittance-receiving households tend to be smaller in size in El Salvador, Dominican Republic and Honduras.
  - Households with more diversified income sources have higher chances to receive remittances.
  - Having remittance-receiving neighbors increases chances in Dominican Republic, El Salvador, and Guatemala.
- Attributes that increase likelihood of receiving remittances also reduce odds of financial inclusion:
  - Odds of being financially included higher for male, older, urban individuals.
  - Less educated and younger individuals more likely to be excluded.
  - Individuals with primary education are less likely to have an account by 39 and 32 percentage points in Guatemala and the Dominican Republic, respectively.
  - Being unemployed reduces chances of having an account by 16 percentage points in Guatemala and El Salvador, and by 12 and 18 percentage points in Dominican Republic and Honduras, respectively.
  - Individuals who own mobile phones are more likely to have access to an account in all four countries.

### VII.A Remittances digitalization in CA: supply, demand, and regulatory conditions
- Technology availability:
  - Smartphone use widespread among Hispanic migrants; principal means of internet access for Hispanics in the United States.
  - Mobile and 3G network coverage in CA match LAC coverage for most CA countries.
- Consumer digital uptake is moderate due to behavioral inertia, informal employment at origination, and inadequate financial literacy at receiving end.
  - At origination, only 20 percent of migrants use online banking platforms; odds of using digital channel higher when senders are paid into a bank account.
  - On receiving end, over two thirds of the population in Guatemala, Honduras, and Nicaragua, and over a third in El Salvador and Dominican Republic, is yet to experience the mobile internet.
- Regulatory environment (GSMA MMRI):
  - CA scores poorly on authorization, transaction limits, and KYC regulations; fares well on infrastructure, agent network, and consumer protection dimensions.
  - Country-specific regulatory weaknesses:
    - Dominican Republic and Nicaragua score poorly on authorization requirements, including proportionality of capital requirements.
    - El Salvador and Nicaragua score poorly on infrastructure and investment environment (sector-specific taxation, ID verification infrastructure, interoperability infrastructure, utilization of interest income).
    - All CA countries score high on agent network ease for cash-in/cash-out.
    - Guatemala scores poorly on consumer protection (complaint procedures, providers’ disclosure, safeguards for customer funds).
    - Dominican Republic, El Salvador, Guatemala, and Honduras score poorly on transaction limits (proportionality of entry-level and transaction ceilings).
    - Dominican Republic, Guatemala, Honduras, and Nicaragua score poorly on proportionality of KYC and AML/CFT requirements by user risk profile.

### VII.B Country case studies — Guatemala and Honduras
- Guatemala: stylized facts
  - 2019 remittances inflows reached US$10½ billion (over 13 percent of GDP; 90 percent from the U.S.).
  - Remittances account for about 45 percent of recipient households’ income (one in ten households).
  - Digital origination is nascent (0.3 percent of total); cash pick-up dominates (89 percent of total remittances).
  - Banks play a large role in payout (91 percent of total payout, of which 80 in cash); banking network coverage: 28 bank branches per 1,000 km2.
  - Tigo Money: as of June 2020, only 4.1 percent of Tigo’s customer base had active mobile financial services; receiving cross-border payments through Tigo’s e-wallet represents about 2 percent of total remittance inflows.
  - Digital origination cost dynamics:
    - Entry of fintech providers reduced average cost of digital remittances by almost 30 percent over the last decade.
    - New digital-only operators partnered with local banks, cooperatives, and retail stores to expand payout networks, reducing average and dispersion of transfer costs across modes.
  - Demand constraints: financial literacy and mobile penetration lag; only 44 percent of population has an account with a financial institution or mobile money provider.
  - Regulatory environment: mobile financial services regulated as services channeled through supervised financial institutions; lack of dedicated MFS regulatory framework hinders MFS potential; mobile money providers lack direct access to national payments and settlement system.
- Honduras: stylized facts
  - 2019 remittances: US$5.4 billion (22 percent of GDP; about 85 percent from the United States).
  - Supply conditions:
    - Remittances mainly via traditional MTOs, followed by banks; virtually all transactions originated in cash, including those via banks.
    - Tigo Money licensed in 2019 (DINELSA); Tengo created in 2013 (electronic wallets).
    - In 2019, 17.6 percent of Tigo Money transactions and 8.8 percent of Tengo transactions were to receive remittances, but together accounted for an estimated 1 percent of total remittances.
  - Costs:
    - 2019 average cost of sending remittances from the U.S. to Honduras was about 3¾ percent for a $200 transfer (World Bank), slightly above the SDG target.
    - Senders reported average costs of 2 to 3 percent during the last two years (Central Bank of Honduras’ remittances survey).
    - Digitally-originated remittances declined about 45 percent over the last decade; card-funded transactions remain least cost-effective.
  - Demand: most recipients financially excluded; only 17 percent of remittance recipients had a checking account; some two-thirds had a savings account or investment fund; 92 percent had a mobile phone; nearly two-thirds used the internet.
  - Regulatory environment:
    - 2016 mobile money regulation is a welcome step but framework not well defined; high requirements for agents and authorized transactional centers may exclude small informal merchants.
    - High minimum capital requirement: 30 million Lempiras (about US$1.2 million) could be a barrier for new entrants.
    - Transactional limits set at 30,000 Lempiras a month (around US$1,250) for maximum stock and flow.
    - Virtually no interoperability between mobile money and banks’ payment systems; only a few bilateral agreements exist.

### VII.C COVID-19 pandemic reshaping digital remittances and payments
- Pandemic effects:
  - Western Union and MoneyGram recorded around 100 percent expansion of digital channels during Q2 2020, accounting for around 22 percent of customer-to-customer revenues (versus 14 percent before the pandemic).
  - Governments’ social protection responses promoted adoption of digital payments:
    - Honduras launched Bono Único: vouchers of 4,000 Lempira to more than 70,000 households delivered electronically (mobile phone text message).
    - Guatemala provided Bono Familia: cash transfers of up to 2,600 quetzals through electronic transfers and tokens to vulnerable households.

### VIII. Concluding remarks: perspectives for payments and remittances digitalization in CA
- Main conclusions:
  - Traditional operators in CA provide remittance services at competitive global prices, but average fees remain above the 3 percent SDG.
  - Greater competition could further decrease fees, improve transparency, and reduce transaction times.
  - Digital remittances can reach remote, low-income households timely and securely and help strengthen financial inclusion, particularly if accompanied by enhanced interoperability.
  - CA has been slow in embracing digitalization despite mobile penetration and regulatory progress; COVID-19 has boosted demand for digital payment services.
- Policy implications and recommendations:
  - Expand use of mobile financial services and promote adequate regulation to make digital financial services more cost-effective.
  - Maintain use of digital means for social welfare payments beyond the pandemic.
  - Regulators and public authorities should ease transfers while safeguarding customer protection, financial stability and integrity.
  - Facilitate entry of new operators and broaden collaborations between telecommunication and fintech companies, banks, and traditional MTOs to expand digital financial services and inclusion.
  - Regulatory principles to implement risk-sensitive and proportionate frameworks:
    - Continuous risk assessment: monitor product risks targeted at underserved groups and gradually ease KYC and CDD as warranted.
    - Risk mitigation via product design: mobile financial products for entry-level customers should limit features or adopt progressive CDD; transactional limits could be used depending on CDD scope and customers’ risk profiles; additional features (credit scoring, lending, insurance) could be unlocked after low-risk track record under simplified KYC/CDD.
    - Appropriate consumer protection: establish initial capital requirements; separate customers’ funds from providers’ capital in trusts; encourage interoperability to promote competition, reduce fixed costs and unlock economies of scale.
  - Country-specific regulatory advice:
    - For Guatemala: regulate e-money issuance to provide a level playing field; widen scope and reach of operations (e.g., interoperability); promote risk-based CDD.
    - For Honduras: refine existing regulations to broaden use of mobile financial services, improve interoperability, and adopt proportionate risk-based customer-due diligence.

*Source: wpiea2021175-print-pdf*

### REFERENCES

### wpiea2021175-print-pdf - REFERENCES

### Annex I: Impact of Remittances on Households' Income — scope and data
- Sample covers four CA countries: Dominican Republic, Guatemala, El Salvador, and Honduras.
- The sample includes CA countries with remittances over 5 percent of GDP. Nicaragua is excluded due to lack of data.
- Household survey data used to compare remittance-receiving and non-remittance-receiving households and to generate income distributions and Gini coefficients with and without remittances.

### Key findings on remittances and household income
- Per capita income:
  - On average across countries, per capita income in remittance-receiving households is over 50 percent higher than in non-remittances receiving households.
  - Country range: between 24 percent higher in Guatemala and 70 percent higher in Honduras.
- Poverty:
  - For remittance-receiving households, the poorest quintiles benefit relatively more from remittances than wealthier households.
- Inequality:
  - Evidence suggests remittances reduce income inequality significantly by about 5 to 6 Gini points in all four countries, consistent with Vacaflores (2018), Akobeng (2016), and Adams and Page (2005).

### Country-level household characteristics and key statistics (from household surveys)
- Guatemala (IADB Household Survey for Guatemala, 2016; Observations 458 remittance, 4,934 no remittances, 5,392 total)
  - Mean age of Household head (years): Receive remittances 51.198; Receive no remittances 48.765; Total 48.982
  - Area (Urban=1): Receive remittances 0.464; Receive no remittances 0.553; Total 0.545
  - Mean monthly per capita income (Inc. Remittances) in quetzals: Receive remittances 1,576.011; Receive no remittances 1,244.254; Total 1,274.513
  - Remittances as percent of total per capita income (Inc. Remittances): Receive remittances 0.279; Receive no remittances 0.000; Total 0.025

- El Salvador (IADB Household Survey for El Salvador, 2016; Observations 4,425 remittance, 16,184 no remittances, 20,609 total)
  - Mean age of Household head (years): Receive remittances 56.98; Receive no remittances 50.36; Total 51.68
  - Area (Urban=1): Receive remittances 0.53; Receive no remittances 0.66; Total 0.64
  - Mean monthly per capita income (Inc. Remittances) in US$: Receive remittances 250.97; Receive no remittances 163.09; Total 179.47
  - Remittances as percent of total per capita income (Inc. Remittances): Receive remittances 0.31; Receive no remittances 0.00; Total 0.06

- Dominican Republic (IADB Household Survey for Dominican Republic, 2016; Observations 1,080 remittance, 6,927 no remittances, 8,007 total)
  - Mean age of Household head (years): Receive remittances 50.65; Receive no remittances 48.78; Total 49.05
  - Area (Urban=1): Receive remittances 0.72; Receive no remittances 0.66; Total 0.67
  - Mean monthly per capita income (Inc. Remittances) in Dominican Peso: Receive remittances 16,025.67; Receive no remittances 11,637.96; Total 12,276.90
  - Remittances as percent of total per capita income (Inc. Remittances): Receive remittances 0.18; Receive no remittances 0.00; Total 0.03

- Honduras (IADB Household Survey for Honduras, 2016; Observations 1,065 remittance, 5,146 no remittances, 6,211 total)
  - Mean age of Household head (years): Receive remittances 54.99; Receive no remittances 48.97; Total 50.04
  - Area (Urban=1): Receive remittances 0.62; Receive no remittances 0.56; Total 0.57
  - Mean monthly per capita income (Inc. Remittances) in Lempira: Receive remittances 4,944.01; Receive no remittances 2,963.71; Total 3,317.80
  - Remittances as percent of total per capita income (Inc. Remittances): Receive remittances 0.23; Receive no remittances 0.00; Total 0.04

### Additional empirical notes and visual evidence
- Figures and distributions:
  - Figure AI.1 shows that the poorest quantiles benefit relatively more from remittances than wealthier households (source: author calculations based on Household Surveys).
  - Panel charts present per capita income distributions (US$ or local currency) comparing remittances vs. no remittances for HND, GTM, SLV, DR.
- Household composition patterns:
  - Across countries, remittance-receiving households tend to have older household heads on average and differing urban shares versus non-receiving households (country-specific patterns noted in tables).

*Source: wpiea2021175-print-pdf - REFERENCES (Annex I and referenced household survey tables).*

### Annex II. Probit Models: Data and Regression Results

### Annex II. Probit Models: Data and Regression Results

### A. Data — Household Survey Variables
- Source: IADB Household Surveys for Dominican Republic, Guatemala, El Salvador, and Honduras (2016) and Costa Rica (2015).
- Dependent variable:
  - Remittance Receivers: A dummy which takes value of 1 if the household receive remittances and 0 otherwise.
- Household/head characteristics (dummies and continuous):
  - Head Female: A dummy which takes value of 1 if the household is female and 0 otherwise.
  - Head Age: Age in years.
  - Head Primary: A dummy which takes value of 1 if the household head completed primary education and 0 otherwise.
  - Head Poor Class: A dummy which takes value of 1 if the household head is poor and 0 otherwise.
  - Neighbors: This variable shows if the neighbors also receive remittances.
  - Head Unemployed: A dummy which takes value of 1 if household head is unemployed and zero otherwise.
  - Household Size: Provides information regarding number of people in a household.
  - Number of income sources: Dummy variable for number of income sources. Takes value of 1 if household has one source of income, 0 otherwise.
- Interaction terms included:
  - Female*Age; Female*Primary Educ; Female*poor class; Age*Primary educ; Age* Poor class; Primary educ*Poor class.
  - Higher-order interactions: Female*Age Primary educ; Female*Age*Poor class; Age* Primary educ Poor class; Female*Age*Primary educ*poor class.

### A. Data — Findex Variables (Financial Inclusion)
- Source: 2017 Findex database, World Bank. 1000 respondents are randomly selected for each country.
- Empirical patterns reported in the survey data:
  - (i) The share of male individuals having an account at a financial institution is larger than for females in all five CA countries.
  - (ii) Individuals with primary education are less-banked than those with secondary and tertiary education in all five CA countries.
  - (iii) The share of those having an account at a financial institution is higher for individuals aged below 50, except for Costa Rica.
  - (iv) The proportion of those having an account at a financial institution is lower among individuals with low income.
- Variables (dummies and continuous):
  - Account (Dependent variable): A dummy which takes value of 1 if the individual is having account and 0 otherwise.
  - Female: A dummy which takes value of 1 if the individual is female and 0 otherwise.
  - Age: Age in years.
  - Primary: A dummy which takes value of 1 if the individual completed primary education and 0 otherwise.
  - Poor Class: A dummy which takes value of 1 if the individual is poor and 0 otherwise.
  - Unemployed: A dummy which takes value of 1 if the individual is unemployed and 0 otherwise.
  - Interaction terms: Female*Age; Female*Primary Educ; Female*poor class; Age*Primary educ; Age* Poor class; Primary educ*Poor class; Female*Age Primary educ; Female*age*Poor class; Age* Primary educ Poor class; Female*age*Primary educ*poor class.

### B. Regression Results — Probit Model for Remittances and Household Characteristics (Table AII.3)
- Sample sizes:
  - N 5392 (GTM), 20609 (SLV), 8007 (DR), 6211 (HND).
- Reported coefficients (standard errors in parentheses). Significance codes: * p < 0.1, ** p < 0.05, *** p  < 0.0.
- Selected coefficients by country (coefficient [standard error] and significance):
  - Head Female:
    - GTM: 0.13*** (0.04)
    - SLV: 0.08** (0.04)
    - DR: 0.07** (0.04)
    - HND: 0.14* (0.07)
  - Head Rural:
    - GTM: 0.04*** (0.01)
    - SLV: 0.08*** (0.01)
    - DR: -0.01 (0.01)
    - HND: 0.01 (0.02)
  - Head Age:
    - GTM: 0.00 (0.00)
    - SLV: -0.00*** (0.00)
    - DR: -0.00*** (0.00)
    - HND: -0.00 (0.00)
  - Head Prim-educ:
    - GTM: -0.05** (0.02)
    - SLV: -0.32*** (0.03)
    - DR: -0.16*** (0.05)
    - HND: -0.21*** (0.05)
  - Head Poor Class:
    - GTM: -0.09 (0.10)
    - SLV: -0.37*** (0.12)
    - DR: 0.10 (0.12)
    - HND: -0.85** (0.41)
  - Female*Age:
    - GTM: -0.00** (0.00)
    - SLV: 0.00 (0.00)
    - DR: -0.00 (0.00)
    - HND: -0.00 (0.00)
  - Female*Prim-educ:
    - GTM: 0.05 (0.04)
    - SLV: 0.17*** (0.06)
    - DR: -0.04 (0.08)
    - HND: 0.05 (0.11)
  - Female*poor class:
    - GTM: -0.01 (0.06)
    - SLV: -0.20 (0.12)
    - DR: -0.23** (0.12)
    - HND: 0.23 (0.27)
  - Age*Prim-educ:
    - GTM: 0.00* (0.00)
    - SLV: 0.01*** (0.00)
    - DR: 0.00** (0.00)
    - HND: 0.00*** (0.00)
  - Age*Poor Class:
    - GTM: 0.00 (0.00)
    - SLV: 0.00 (0.00)
    - DR: -0.01* (0.00)
    - HND: 0.01* (0.01)
  - Prim-educ*Poor Class:
    - GTM: 0.01 (0.10)
    - SLV: 0.34*** (0.13)
    - DR: -0.03 (0.15)
    - HND: 0.62 (0.40)
  - Female*Age*Prim-educ:
    - GTM: -0.00 (0.00)
    - SLV: -0.00*** (0.00)
    - DR: 0.00 (0.00)
    - HND: 0.00 (0.00)
  - Female*age*Poor Class:
    - GTM: -0.00 (0.00)
    - SLV: 0.00 (0.00)
    - DR: 0.00* (0.00)
    - HND: -0.00 (0.00)
  - Age*Prim-educ*Poor Class:
    - GTM: -0.00 (0.00)
    - SLV: -0.01** (0.00)
    - DR: 0.00 (0.00)
    - HND: -0.01* (0.01)
  - Female*Rural*Age*Prim-educ:
    - GTM: -0.00 (0.00)
    - SLV: 0.00 (0.00)
    - DR: 0.00 (0.00)
    - HND: -0.00 (0.00)
  - Neighbors:
    - GTM: 0.01*** (0.00)
    - SLV: 0.00*** (0.00)
    - DR: 0.00*** (0.00)
    - HND: -0.00* (0.00)
  - Head Unemployment:
    - GTM: 0.06*** (0.01)
    - SLV: 0.18*** (0.01)
    - DR: 0.18*** (0.01)
    - HND: 0.14*** (0.02)
  - HH Size:
    - GTM: -0.00 (0.00)
    - SLV: -0.00** (0.00)
    - DR: -0.01*** (0.00)
    - HND: -0.01*** (0.00)
  - Income Sources:
    - GTM: 0.06*** (0.01)
    - SLV: 0.08*** (0.00)
    - DR: 0.09*** (0.01)
    - HND: 0.07*** (0.01)

### B. Regression Results — Probit Model for Financial Inclusion and Household Characteristics, Findex Data (Table AII.4)
- Sample sizes:
  - N 1000 (GTM), 1000 (SLV), 1000 (DR), 1000 (HND).
- Reported coefficients (standard errors in parentheses). Significance codes: * p  < 0.1, ** p < 0.05, *** p < 0.01.
- Selected coefficients by country (coefficient [standard error] and significance):
  - Female:
    - GTM: -0.00 (0.07)
    - SLV: -0.02 (0.07)
    - DR: -0.14* (0.08)
    - HND: 0.16** (0.07)
  - Remote:
    - GTM: -0.26*** (0.04)
    - SLV: -0.30*** (0.05)
    - DR: -0.34*** (0.05)
    - HND: -0.26*** (0.03)
  - Age:
    - GTM: 0.01 (0.00)
    - SLV: 0.01*** (0.00)
    - DR: 0.00 (0.00)
    - HND: 0.01*** (0.00)
  - Primary-Educ:
    - GTM: -0.39*** (0.14)
    - SLV: -0.05 (0.12)
    - DR: -0.32*** (0.12)
    - HND: 0.00 (0.13)
  - Poor Class:
    - GTM: -0.09 (0.16)
    - SLV: -0.24 (0.20)
    - DR: -0.14 (0.17)
    - HND: -0.25 (0.18)
  - Female*Age:
    - GTM: -0.00 (0.00)
    - SLV: -0.00 (0.00)
    - DR: 0.00 (0.00)
    - HND: -0.00*** (0.00)
  - Female*Prim-educ:
    - GTM: 0.22*** (0.07)
    - SLV: 0.02 (0.06)
    - DR: 0.11* (0.06)
    - HND: -0.01 (0.06)
  - Female*Poor Class:
    - GTM: -0.20** (0.08)
    - SLV: 0.13 (0.09)
    - DR: 0.00 (0.08)
    - HND: -0.01 (0.08)
  - Age*Prim-educ:
    - GTM: -0.00 (0.00)
    - SLV: -0.00* (0.00)
    - DR: -0.00 (0.00)
    - HND: -0.00* (0.00)
  - Age*Poor Class:
    - GTM: 0.00 (0.00)
    - SLV: -0.00 (0.00)
    - DR: -0.00 (0.00)
    - HND: 0.00 (0.00)
  - Prim-educ*Poor class:
    - GTM: 0.21** (0.10)
    - SLV: 0.05 (0.10)
    - DR: 0.08 (0.08)
    - HND: 0.13 (0.09)
  - Unemployment:
    - GTM: -0.16*** (0.03)
    - SLV: -0.16*** (0.03)
    - DR: -0.12*** (0.03)
    - HND: -0.18*** (0.03)
  - Owns Mobile phone:
    - GTM: 0.14*** (0.03)
    - SLV: 0.20*** (0.03)
    - DR: 0.19*** (0.04)
    - HND: 0.21*** (0.04)

### Annex III. Mobile Money Remittances: Regional Coverage (Table AIII.1)
- List of countries by region (selected entries as presented):
  - Afghanistan — South Asia
  - Albania — Europe and Central Asia
  - Argentina — Latin America and the Caribbean
  - Armenia — Europe and Central Asia
  - Bangladesh — South Asia
  - Benin — Sub-Saharan Africa
  - Bolivia — Latin America and the Caribbean
  - Botswana — Sub-Saharan Africa
  - Brazil — Latin America and the Caribbean
  - Burkina Faso — Sub-Saharan Africa
  - Burundi — Sub-Saharan Africa
  - Cambodia — East Asia and Pacific
  - Cameroon — Sub-Saharan Africa
  - Central African Republic — Sub-Saharan Africa
  - Chad — Sub-Saharan Africa
  - Colombia — Latin America and the Caribbean
  - Congo — Sub-Saharan Africa
  - Congo, Democratic Republic — Sub-Saharan Africa
  - Côte d'Ivoire — Sub-Saharan Africa
  - Dominican Republic — Latin America and the Caribbean
  - Egypt — Middle East and North Africa
  - El Salvador — Latin America and the Caribbean
  - eSwatini — Sub-Saharan Africa
  - Ethiopia — Sub-Saharan Africa
  - Fiji — East Asia and Pacific
  - Gabon — Sub-Saharan Africa
  - Gambia — Sub-Saharan Africa
  - Georgia — Europe and Central Asia
  - Ghana — Sub-Saharan Africa
  - Guatemala — Latin America and the Caribbean
  - Guinea — Sub-Saharan Africa
  - Guinea-Bissau — Sub-Saharan Africa
  - Guyana — Latin America and the Caribbean
  - Haiti — Latin America and the Caribbean
  - Honduras — Latin America and the Caribbean
  - India — South Asia
  - Indonesia — East Asia and Pacific
  - Iran — Middle East and North Africa
  - Iraq — Middle East and North Africa
  - Jamaica — Latin America and the Caribbean
  - Jordan — Middle East and North Africa
  - Kazakhstan — Europe and Central Asia
  - Kenya — Sub-Saharan Africa
  - Kyrgyzstan — Europe and Central Asia
  - Lesotho — Sub-Saharan Africa
  - Liberia — Sub-Saharan Africa
  - Madagascar — Sub-Saharan Africa
  - Malawi — Sub-Saharan Africa
  - Malaysia — East Asia and Pacific
  - Maldives — South Asia
  - Mali — Sub-Saharan Africa
  - Mauritania — Sub-Saharan Africa
  - Mauritius — Sub-Saharan Africa
  - Mexico — Latin America and the Caribbean
  - Mongolia — East Asia and Pacific
  - Morocco — Middle East and North Africa
  - Mozambique — Sub-Saharan Africa
  - Myanmar — East Asia and Pacific
  - Namibia — Sub-Saharan Africa
  - Nepal — South Asia
  - Nicaragua — Latin America and the Caribbean
  - Niger — Sub-Saharan Africa
  - Nigeria — Sub-Saharan Africa
  - Pakistan — South Asia
  - Papua New Guinea — East Asia and Pacific
  - Paraguay — Latin America and the Caribbean
  - Peru — Latin America and the Caribbean
  - Philippines — East Asia and Pacific
  - Qatar — Middle East and North Africa
  - Russian Federation — Europe and Central Asia
  - Rwanda — Sub-Saharan Africa
  - Samoa — East Asia and Pacific
  - Senegal — Sub-Saharan Africa
  - Seychelles — Sub-Saharan Africa
  - Sierra Leone — Sub-Saharan Africa
  - Singapore — East Asia and Pacific
  - Solomon Islands — East Asia and Pacific
  - Somalia — Sub-Saharan Africa
  - South Africa — Sub-Saharan Africa
  - South Sudan — Sub-Saharan Africa
  - Sri Lanka — South Asia
  - Sudan — Sub-Saharan Africa
  - Tajikistan — Europe and Central Asia
  - Tanzania — Sub-Saharan Africa
  - Thailand — East Asia and Pacific
  - Togo — Sub-Saharan Africa
  - Tonga — East Asia and Pacific
  - Tunisia — Middle East and North Africa
  - Turkey — Europe and Central Asia
  - Uganda — Sub-Saharan Africa
  - United Arab Emirates — Middle East and North Africa
  - Vanuatu — East Asia and Pacific
  - Vietnam — East Asia and Pacific
  - Zambia — Sub-Saharan Africa
  - Zimbabwe — Sub-Saharan Africa

*Source: wpiea2021175-print-pdf - Annex II. Probit Models: Data and Regression Results*

### Annex IV. Drivers of Remittance Fees’: A Firm Perspective

### Annex IV. Drivers of Remittance Fees’: A Firm Perspective

### A. Empirical Approach
- Reduced-form pricing equation (equation (4)):
  - ln(ffs)i = α0 + α1 ln(amount)i + α2 ln(fee competitor)i + α3 ln(remittances)i + α4 Xi + εi
  - Definitions preserved from source:
    - ln(ffs)i: natural log of remittances fees charged by the MTO for transfers to country i
    - ln(amount)i: natural log of the amount of remittances transferred
    - ln(fee competitor)i: natural log of the average fee for competitors
    - ln(remittances)i: natural log of the total volume of remittances sent from the U.S. to country i
    - Xi: indicator of financial development in country i
- Mode-specific specification (equation (5)) allowing elasticities to differ by transaction mode (card-to-cash and transfer-to-cash), focusing on sender use of the agent’s mobile app or website and cash payout on the receiving end.
  - Mode-interacted elasticities:
    - ∂ ln(ffs)i / ∂ ln(amount)i = CarCardi * β2 + TransToCashi * α1  (equation (6))
    - ∂ ln(ffs)i / ∂ ln(fee competitor)i = CarCardi * β3 + TransToCashi * α2  (equation (7))
- Priors for elasticities (verbatim):
  - (i) fee’s elasticity to the amount of remittances transferred, α1 + β2, expected to be positive if larger transactions were to be more costly because of their lower frequency, higher AML/CFT regulatory costs, or inadequate competition in that market segment;
  - (ii) fee’s elasticity to the competitors’ fee, α2 + β3, expected to be positive and higher the more firms align their fees to peers (the greater the competition amongst firms);
  - (iii) fee’s elasticity to the total volume of remittances, α3, expected to be negative if firms in larger remittances corridors were to invest more to reduce their marginal costs;
  - (iv) sign of α4 expected to be negative if more developed financial systems in the recipient countries result in less costly remittances, ceteris paribus the regulatory framework;
  - (v) by mode, fees’ elasticities for bank transfers (both to the amount of remittances transferred and to the competitors’ fees) are expected to be lower than for card transfers, because bank accounts transactions are strictly monitored, which entails longer processing on average and less variation in fees.
- Data note (verbatim highlights):
  - Focus on U.S. senders and two specific channels; cash remittances on the receiving end given its dominance.
  - Consideration of digital remittances at the payout, particularly for LAC, would restrict the sample considerably, given their low incidence vis-à-vis the cash options.

### B. Fees Charged by Selected Operators for Remittances Originating in the U.S.
- Table A.IV.1 Average Fees Charged by Three MTOs, 2019 (US$200 and US$500, same day or less) — selected entries (preserved exactly as in source):
  - LAC countries (Xoom Remitly Western Union)
    - Argentina 13.5 7.1 20.0
    - Bolivia 11.5 7.0 10.0
    - Brazil 6.2 4.5 5.5
    - Colombia 6.2 4.0 5.6
    - Costa Rica 16.0 7.6 5.5
    - Dominican Republic 9.2 7.0 8.5
    - Ecuador 13.0 8.5 9.8
    - El Salvador 13.7 8.5 9.8
    - Guatemala 7.0 6.6 5.5
    - Honduras 13.7 5.2 10.0
    - Jamaica 9.2 4.9 5.5
    - Mexico 7.0 2.0 5.5
    - Nicaragua 13.7 8.5 18.8
    - Peru 9.0 8.5 8.5
  - Non-LAC countries (Xoom Remitly Western Union)
    - Bangladesh 5.7 4.0 5.5
    - China 12.5 4.0 28.5
    - India 7.0 4.0 5.5
    - Indonesia 6.6 7.5 5.5
    - Kenya 6.6 0.0 5.5
    - Nigeria 6.6 4.0 5.5
    - Pakistan 5.7 2.0 5.5
    - Philippines 5.5 10.2 5.0
    - Romania 6.4 5.0 5.5
    - Russia 26.8
    - South Africa 8.0 5.7 22.3
    - Sri Lanka 13.0 6.0 5.5
    - Thailand 6.6 3.5 5.5
    - Turkey 4.2 8.5
    - Ukraine 6.6 16.3
    - Vietnam 5.2 6.6 8.5
  - Total 9.0 5.6 9.8
- Data sources: Xoom, Remitly, and Western Union websites; and authors' calculations.

### C. Regression Results — Key Findings and Statistical Evidence
- Aggregate characterization:
  - Results presented for three country groupings: emerging market, LAC, and non-LAC countries.
- LAC region key findings (verbatim summary points preserved):
  - Fee’s elasticity to the amount of remittances transferred, α1 + β2.
    - Evidence of fee differentiation (or lack of) is weak both for cash-to-cash and transfer-to-cash.
  - Fee’s elasticity to the competitors’ fee, α2 + β3.
    - The elasticity with respect to competitor fees is found to be significant and large, suggesting that firms operating in LAC country destinations are highly sensitive to price movements vis-à-vis their competitors.
    - The estimated elasticities do not differ widely across transaction modes.
  - Fee’s elasticity to the total volume of remittances, α3.
    - The demand elasticity coefficient was statistically significant only for one operator.
    - Overall, fees do not differ widely across country destinations, which suggests a weak role for economies of scale in the provision of remittances.
    - A low demand elasticity may be indication of markets’ segmentation across firms.
  - Fees and development of financial system, α4.
    - (i) Banking sector concentration. A highly-concentrated banking sector in LAC countries is associated with higher remittances fees for only one of the three companies, suggesting that that particular firm may have difficulties accessing the markets leading to an extra premium for the remittances fee.
    - (ii) Financial development. Tables A.IV.4–A.IV.7 report results for a general index and subcomponents of the financial development index. Greater financial development and deeper financial institutions are associated with lower remittances fees. There is some evidence that improved financial access is associated with lower remittances fees. However, the coefficients for the financial institutions development index do not seem robust across firms.
- Specific statistical patterns from firm regressions (tables A.IV.2–A.IV.7):
  - Ln(fee competitor)*Card to cash (D) coefficients are frequently positive and often significant at conventional levels across firms and subsamples (examples preserved from tables):
    - 0.660*** (p-value (0.000)), 0.392** (0.044), 0.792*** (0.010) in pooled specifications.
    - 0.594*** (0.001), 0.307* (0.099), 0.816*** (0.009) in emerging markets subsample.
    - 0.761*** (0.009), 0.421 (0.126), 0.779* (0.059) in LAC subsample with Financial Development Index.
  - Ln(remittances) coefficients are often negative and occasionally significant:
    - Examples: -0.044** (0.049), -0.046 (0.124), -0.158*** (0.000) in pooled specifications; -0.062** (0.030) and -0.071** (0.040) in subsamples; -0.055** (0.042), -0.097*** (0.007) when using Financial Development Index.
  - Financial development measures show negative associations with fees in several specifications:
    - Financial Development Index (IMF): -0.201** (0.015), -0.250*** (0.008) in pooled/selected regressions.
    - Financial Institutions Index (IMF): positive and negative coefficients across firms (e.g., 0.421** (0.024) for one firm; -0.998*** (0.000) for another), indicating lack of robustness across firms.
    - Financial Institutions Depth Index (IMF): 0.147* (0.063) for one firm, -0.315*** (0.007) for another.
    - Financial Institutions Access Index (IMF): 0.218** (0.017) for one firm, -0.351** (0.018) for another.
  - Card-to-cash dummy coefficients and interactions with ln(amount) show heterogeneity by firm and subsample:
    - Examples: Card to cash (D) = -1.463 (0.101), 0.719 (0.496), 0.056 (0.965) in pooled; -1.740** (0.042) for XOOM in emerging markets; -2.283** (0.035) in one pooled specification.
    - Ln(amount)*Card to cash (D) examples: 0.259* (0.061), 0.047 (0.777), -0.056 (0.802) pooled; 0.327** (0.015) emerging markets; 0.407** (0.013) pooled for a specific firm.
  - Model fit statistics (examples preserved):
    - R-squared values vary across specifications and firms, e.g., 0.565, 0.353, 0.255 in pooled panels; 0.586, 0.367, 0.229 in emerging markets panels; 0.596, 0.390, 0.281 when including Financial Development Index.
- Caveats emphasized in source (verbatim points):
  - (i) inclusion of a larger number of MTOs, financial institutions, and fintech operators, would allow to distill more systematic differences in price setting in the industry;
  - (ii) consideration of time-varying data would allow to check firms’ pricing strategy in response to new entrants or price changes by competitors;
  - (iii) the above data limitations restrain the sample size for subregions, precluding an analysis for CA;
  - (iv) exclusion of appropriate control variables such as regulation, the extent of firms’ payment network, and the efficiency of the payment systems in the receiving countries, may lead to omitted variable bias.

*Source: Annex IV. Drivers of Remittance Fees’: A Firm Perspective (from wpiea2021175-print-pdf).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021175-print-pdf.pdf_