## wpiea2024069-print-pdf

## Source details

**Canonical URL:** [wpiea2024069-print-pdf](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024069-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024069-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024069-print-pdf.pdf.json)

---

### Executive Summary — Introduction and research question
- Digital currencies denominated in legal tender and exchanged through smart phones described as "one of the most significant offspring of technological innovations in the financial sector."
- Paper studies "e-money" (nonbank deposit-taking e-money issuers, EMIs) and asks:
  - What is the impact of "e-money" development on monetary policy transmission?
  - What are the implications for designing other digital currencies such as central-bank digital currency (CBDC)?
- Central empirical ambiguity: EMIs can complement banks or substitute for banks, producing opposite effects on monetary policy transmission.

### Conceptual considerations and framework
- Complementarity channel:
  - EMIs (e.g., mobile network operators) can bring previously unbanked depositors into the banking system, increasing financial intermediation and strengthening monetary policy transmission.
- Substitution channel:
  - EMIs can move deposits away from banks to nonbank institutions, producing financial disintermediation and weakening monetary policy transmission.
- Four channels through which e-money affects monetary policy transmission:
  - Credit channel
  - Bank rates (lending/deposit rate) channel
  - Asset prices channel
  - Exchange rate channel
- Mechanisms that can enhance transmission (when EMIs complement banks) include channeling currency into banks, moving savings into deposits, increasing competition for wholesale depositors, and strengthening local currency demand.
- Mechanisms that can weaken transmission (when EMIs substitute banks) include ring-fenced nonloanable e-floats, savings shifting to less policy-sensitive digital assets, credit crunch concentrating lending to prime customers, and weaker exchange rate transmission.
- Key determinant: regulatory design (e.g., ability of EMIs to offer interest on balances or extend credit).

### Regulatory case studies (five Sub-Saharan African countries)
- Countries: Ghana, Kenya, Nigeria, Tanzania, Uganda.
- Common regulatory features documented:
  - Money collected by EMIs must be maintained in a trust fund in banks (all five countries require trust funds).
  - EMIs are generally not allowed to extend credit directly but can partner with banks to offer bank credit through smartphone features (all five countries allow interface with banking services).
  - Most EMIs can offer interest on customers’ balances; country heterogeneity exists:
    - Ghana: EMIs can earn interest; 2015 Guidelines require distribution of at least 80 percent of interest income to e-money holders.
    - Kenya: Interest distribution to e-money holders is not allowed; income typically donated to a public charitable organization.
    - Nigeria: PSBs can earn and distribute interest on e-money balances; PSBs are not allowed to make loans.
    - Tanzania: EMIs earn interest and commonly distribute interest since 2014.
    - Uganda: EMIs must keep 100 percent of e-money balances in liquid assets; interest earned on the trust fund should be distributed to customers quarterly.
- Collective interpretation: regulatory features observed point to EMIs complementing banks and enhancing monetary policy transmission via financial inclusion, channeling funds into banks, and boosting bank credit and competition.

### Data and empirical method
- Panel data coverage:
  - 21 countries at a monthly frequency.
  - 47 countries at an annual frequency.
  - Period: between 2001 and 2019 (monthly interest rate data January 2001 to December 2019; annual FAS data 2004 to 2019).
- E-money data source: Financial Access Survey (FAS), annual data covering about 190 countries from 2004 to 2019; e-money accounts per 1,000 adults and bank accounts per 1,000 adults used to construct intensity measures.
- Other data sources:
  - International Financial Statistics (IFS) for annual bank deposits, bank credits, government deposits, and bank credits to central government.
  - Monthly deposit, lending, and policy rates from IFS.
  - Macro variables from World Economic Outlook (WEO).
- Estimation approaches:
  - Two-way fixed effect (TWFE) estimator with a single treatment to estimate causal effects of e-money introduction on monetary policy transmission (equations (1)–(3)).
  - Fixed-effect (FE) estimator for bank deposits and credit (equation (4)) due to concerns about the parallel trends assumption for TWFE.
- Measurement and design details preserved exactly:
  - E-money intensity = registered e-money accounts per 1,000 adults divided by bank accounts per 1,000 adults.
  - Treatment variable: monthly change in policy rate ΔPolicyRatei,t; lending rate response measured from t-1 to t+6 (6-month lag).
  - Two e-money intensity dummies: D(emoneyL) = 1 if e-money introduced and e-money intensity < 0.5; D(emoneyH) = 1 if e-money introduced and e-money intensity > 0.5.
  - Standard errors clustered at the country level; country and time fixed effects included.
  - Samples split by initial level of financial inclusion measured by number of deposit accounts per 1,000 adults prior to e-money introduction.

### Key empirical findings — monetary policy transmission and rates
- E-money development accompanied by stronger monetary policy transmission, measured by responsiveness of interest rates to the policy rate:
  - In subsample with low initial financial inclusion, e-money development associated with an increase in the elasticity of the lending rate to the policy rate by 0.3.
  - Using the deposit rate, e-money development associated with an increase in elasticity of the deposit rate to the policy rate by about 0.4 (positive difference between β2L and β2H statistically significant).
- Regression point estimates (selected):
  - Changes in Policy Rate → Changes in Lending Rate: 0.86∗∗∗ (All), 0.89∗∗∗ (Low financial inclusion), 0.65∗ (High financial inclusion).
  - Interaction: High E-money Intensity × Changes in Policy Rate: 0.26 (All), 0.28∗∗ (Low), -0.05 (High).
  - Changes in Policy Rate → Changes in Deposit Rate: 0.94∗∗∗ (All), 0.93∗∗∗ (Low), 0.94∗∗ (High).
  - Interaction: High E-money Intensity × Changes in Policy Rate: 0.39∗∗∗ (All), 0.42∗∗∗ (Low), -0.44 (High).

### Key empirical findings — bank competition, spreads, and rates
- Deposit rate and spread results (selected exact coefficients):
  - High E-money Intensity coefficient on deposit rate: 0.81∗∗∗ (All), 0.78∗∗ (Low), 0.39 (High).
  - High E-money Intensity coefficient on lending-minus-deposit spread: -0.84∗∗∗ (All), -0.71∗ (Low), -0.48 (High).
  - Policy Rate coefficient on deposit rate: 0.56∗∗∗ (All), 0.64∗∗∗ (Low), 0.47∗∗∗ (High).
- Interpretation: decline in lending-to-deposit spreads with e-money development implies increased competition for household savings and efficiency gains in financial intermediation; effects more pronounced where initial financial inclusion is low.

### Key empirical findings — bank deposits and credit (size of banking sector)
- FE regression results linking growth of e-money intensity to bank credit and deposits (selected exact coefficients and significance):
  - Growth of e-money intensity → Growth of bank credit: 0.04∗∗∗ (All, column 1), 0.03∗∗∗ (All with country FE, column 2), 0.05∗∗∗ (Low, column 3), 0.04∗∗ (Low with FE, column 4), 0.01 (High, column 5), 0.02 (High with FE, column 6).
  - Growth of e-money intensity → Growth of credit to private sector: 0.04∗∗ (All), 0.04∗∗ (All with FE), 0.05∗∗ (Low), 0.04∗ (Low with FE), 0.01 (High), 0.02 (High with FE).
  - Growth of e-money intensity → Growth of total deposits: 0.03∗∗ (All), 0.03∗∗∗ (All with FE), 0.05∗∗∗ (Low), 0.05∗∗∗ (Low with FE), -0.01 (High), 0.00 (High with FE).
  - Growth of e-money intensity → Growth of private-sector deposits: 0.03∗ (All), 0.03∗∗ (All with FE), 0.05∗∗ (Low), 0.05∗∗∗ (Low with FE), -0.01 (High), 0.00 (High with FE).
- No robust evidence that e-money development increases credit to the government; coefficients on credit to government generally imprecise and not significant.

### Key empirical findings — financial inclusion
- Introduction of e-money associated with increases in share of banked population:
  - Cross-sectional TWFE: on average, e-money development associated with a 4 percent increase in the share of banked population (β positive and statistically significant in full sample and subsample with low initial financial inclusion).
  - Dynamics around inception:
    - Post0 (inception year): coefficient not consistently significant.
    - Post1 and Post2 and PostBeyond: coefficients become statistically significant and larger, indicating gradual effects over several years.
  - Selected coefficients (Post effects):
    - Post: 4.04∗∗ (All, col 1), 4.44∗∗ (All with controls, col 2).
    - Post1: 3.30∗ (All), 4.98∗∗ (All with controls).
    - Post2: 5.61∗∗ (All), 7.68∗∗ (All with controls).
    - PostBeyond: 7.37∗∗ (All), 7.28∗∗ (All with controls).
- Conclusion: e-money development promotes financial inclusion, with effects materializing over one to several years post inception.

### Policy implications and recommendations
- Design and regulation of CBDC and e-money should:
  - Ensure accessibility without the need for a bank account to improve financial inclusion.
  - Encourage regulatory design that fosters complementarity between e-money and the banking sector (for example, channeling e-money balances to banks to make them available for loans).
  - Enhance credit registries to facilitate more efficient lending by banks.
  - Encourage collaboration between EMIs and banks to extend credit to the private sector, fostering financial deepening while safeguarding financial stability.
- Stability considerations:
  - Banks are typically subject to regulation and supervision while MNOs or other EMIs are typically subject to payment system oversight only; safeguarding financial sector stability remains important.

### Concluding remarks — summary of main findings
- All countries, irrespective of initial conditions, experience stronger monetary policy transmission with e-money development; elasticity of lending and deposit rates with respect to the policy rate increases with e-money intensity, especially where initial financial inclusion is limited.
- E-money and the banking sector appear to develop in tandem (complementarity), with stronger correlation between growth in e-money and growth in deposits and credit in countries with limited initial financial inclusion.
- E-money development promotes financial inclusion (on average a 4 percent increase in the share of banked population), with gradual dynamics observed over one to several years post inception.
- Bank lending-deposit spreads tend to decline with e-money development, consistent with increased competition among banks.

*IMF Working Paper — E-Money and Monetary Policy Transmission (wpiea2024069-print-pdf)*

### Executive Summary ......................................................................................................

### Executive Summary

### Introduction and research question
- The development of digital currencies is described as "one of the most significant offspring of technological innovations in the financial sector."
- The paper studies "e-money" — digital currencies denominated in legal tender and exchanged through features of smart phones — and asks: What is the impact of "e-money" development on monetary policy transmission? What are the implications for designing other digital currencies such as central-bank digital currency (CBDC)?
- The central empirical ambiguity: nonbank deposit-taking e-money issuers (EMIs) can either complement or substitute banks, producing opposite effects on monetary policy transmission.

### Conceptual considerations
- Complementarity channel: EMIs such as mobile network operators (MNOs) can complement banks by bringing previously unbanked depositors into the banking system, potentially leading to higher financial intermediation and stronger monetary policy transmission.
- Substitution channel: EMIs can substitute for banks by moving bank deposits away from banks to nonbank financial institutions, potentially causing financial disintermediation and weaker monetary policy transmission.
- The net effect of e-money development on monetary policy transmission is therefore an empirical question.
- Typical features of e-money regulations in Sub-Saharan Africa point to regulators’ preference for EMIs to complement rather than substitute banks.

### Data and empirical method
- Panel data coverage: 
  - 21 countries at a monthly frequency, and 
  - 47 countries at an annual frequency,
  - for the period between 2001 and 2019.
- Estimation approach: two-way fixed effect estimator with a single treatment to estimate causal effects of e-money development on monetary policy transmission.

### Key empirical findings
- E-money development has accompanied:
  - (i) stronger monetary policy transmission, measured by the responsiveness of interest rates to the policy rate;
  - (ii) growth in bank deposits and credit;
  - (iii) competition among banks and efficiency gains in financial intermediation, measured by deposit-to-lending rate spreads.
- Evidence of these effects is more pronounced in countries "where e-money development takes off in a context of limited financial inclusion."

### Implications emphasized
- E-money development can strengthen monetary policy transmission, particularly in countries with limited financial inclusion.
- The regulatory design that encourages EMIs to complement banks appears aligned with stronger intermediation and policy transmission outcomes.

*Source: IMF Working Paper — Executive Summary*

### Introduction

### wpiea2024069-print-pdf - Introduction

### Introduction
- Digital currencies—denominated in legal tender and exchanged through features of smart phones—have gained widespread adoption in many low-income and emerging market economies.
- Key policy questions: What is the impact of “e-money” development on monetary policy transmission? What are the implications for designing other digital currencies such as central-bank digital currency (CBDC)?
- EMIs (nonbank deposit-taking e-money issuers) can either complement or substitute banks:
  - Complementarity example: mobile network operators (MNOs) can bring previously unbanked depositors into the banking system, potentially increasing financial intermediation and strengthening monetary policy transmission.
  - Substitutability example: EMIs could move deposits away from banks to nonbank financial institutions, potentially causing financial disintermediation and weakening monetary policy transmission.
- E-money regulatory features matter (e.g., whether EMIs can offer interest on savings or extend credit). The paper reviews e-money regulations across countries and presents case studies of five Sub-Saharan African countries: Ghana, Kenya, Nigeria, Tanzania, and Uganda.
- Common regulatory features found in these case studies:
  - Money collected by EMIs must be maintained in a trust fund in banks.
  - EMIs are not allowed to extend credit but are allowed to partner with banks to offer bank credit through smartphone features.
  - Most EMIs can offer interest on their customers’ balances.
- Under these regulatory features, EMIs are likely to complement rather than substitute banks, pointing to stronger monetary policy transmission.

### Main empirical questions and methods
- Empirical examination uses panel data covering:
  - 21 countries at a monthly frequency.
  - 47 countries at an annual frequency.
  - Period between 2001 and 2019 (monthly interest rate data from January 2001 to December 2019; annual FAS data from 2004 to 2019).
- The paper examines whether the development of e-money:
  - (i) increases or reduces banks’ ability to create credit;
  - (ii) increases or reduces the responsiveness of lending and deposit rates to the policy rate;
  - (iii) enhances or weakens competition among banks and thereby the efficiency of financial intermediation.

### Main findings
- All countries, irrespective of initial conditions, experience stronger monetary policy transmission with the development of e-money. This transmission, measured by the elasticity of lending and deposit rates with respect to the policy rate, becomes higher with the introduction of e-money.
  - Evidence is particularly pronounced in countries where e-money development takes off in a context of limited financial inclusion (e.g., low-income countries).
- E-money and the banking sector seem to develop in tandem, suggesting complementarity:
  - Growth in e-money and growth in deposits and credit are more correlated in countries with limited initial levels of financial inclusion.
  - There is evidence that e-money development promotes financial inclusion.
- Bank lending-to-deposit rate spreads tend to decline with the development of e-money, suggesting increased competition in the banking sector and efficiency gains in financial intermediation.

### Literature Review (summary)
- Existing literature largely studies single-country or regional cases; this paper fills a gap by providing quantitative analysis across up to 47 countries.
- Selected prior results documented in the literature:
  - Mobile money mitigates asymmetric information in bank lending and fosters risk-sharing to unbanked low-income households (Aron, 2018).
  - E-money shapes saving and consumption behavior (Coulibaly, 2020; Ky and others, 2017).
  - M-Pesa facilitated risk-sharing and consumption smoothing in Kenya (Jack and Suri, 2014) and increased household probability of having bank access (Mbiti and Weil, 2014).
  - E-money has enhanced financial inclusion in Sub-Saharan African countries (Anmad, Green and Jiang, 2020).
  - E-money users are more likely to obtain bank credit and face mitigated firm financing problems (Gosavi, 2018).
  - FinTech and cashless payments predict higher likelihood of loan approval and more favorable lending terms in India (Gosh and others, 2022).
  - Interoperability among mobile money operators can have trade-offs for coverage and inclusion (Brunnermeier and others, 2023).
- Conceptual work on CBDC highlights complex implications for monetary policy, payment services, and bank intermediation (Das and others, 2023; Cœuré and Loh, 2018; Kahn and others, 2022; Malloy and others, 2022; Chiu and others, 2023; Burlon and others, forthcoming).

### Conceptual framework: four channels through which e-money affects monetary policy transmission
- Credit channel
- Bank rates (bank lending/deposit rate) channel
- Asset prices channel
- Exchange rate channel

Mechanisms that can enhance transmission (when EMIs complement banks):
- E-money can channel currency in circulation outside the banking system into banks (e.g., agents depositing banknotes/coins exchanged for airtime), increasing the pool of loanable deposits and strengthening the credit channel.
- Digitization and financial inclusion move savings from traditional, less policy-sensitive instruments (currency, gold, real estate) into bank deposits, bolstering the asset price channel.
- Increased competition to attract wholesale depositors (e.g., MNOs) can lower banks’ excess reserves and strengthen transmission to lending rates (bank-rate channel).
- Safe and convenient e-money can increase the appeal of domestic currency, reduce demand for foreign currency deposits, strengthen local currency, and reduce currency risk premium (exchange rate channel).

Mechanisms that can weaken transmission (when EMIs substitute banks):
- If e-money deposits are not loanable (e.g., ring-fenced in e-float), they can reduce the supply of loans and weaken the credit channel.
- Savings could shift toward digital assets like cryptocurrencies that may be less responsive to domestic monetary policy, reducing the asset price channel.
- Non-loanable e-money balances could trigger a credit crunch, concentrating lending to prime customers whose demand is less sensitive to policy rate changes, weakening the bank-rate channel.
- Weaker monetary policy effectiveness could also weaken transmission to the exchange rate.

Key determinant: whether EMIs complement or substitute banks—this depends on regulatory design (e.g., ability to offer interest, extend credit).

### Case studies: regulatory features in five African countries
- Countries studied: Ghana, Kenya, Nigeria, Tanzania, Uganda.
- Regulatory aspects reviewed:
  - Are EMIs required to maintain a pool of liquid funds equivalent to the aggregate balance of clients’ e-wallets?
  - Are EMIs allowed to lend e-money balances?
  - Are EMIs required or allowed to pay interest on e-money balances?

Ghana
- EMIs required to keep counterpart of e-money issued in trust funds held in commercial banks; trust funds should be in cash balances or other qualified liquid assets.
- EMIs not allowed to make direct lending but can work with commercial banks to provide banking services, including lending.
- EMIs can earn interest on e-money balances.
- 2015 Guidelines set the minimum share of distributed interest income at 80 percent—EMIs must distribute at least 80 percent of interest income to e-money holders.
- 2015 policy change encouraged EMIs to work with commercial banks, stimulating rapid growth in accounts and transactions (Bank of Ghana, 2022).
- Additional regulations released under the Payment Systems and Services Act in 2019.

Kenya
- E-money began in 2007 with M-Pesa (Safaricom Ltd); National Payment System Regulations enacted in 2014 as legal framework.
- EMIs must establish trust funds in banks; trust funds treated equally with other bank accounts regarding withdrawals, reserve requirements, and other regulations.
- Interest distribution to e-money holders is not allowed: any income generated from placements of trust funds must be donated to a public charitable organization (practice: M-Pesa Foundation).
- EMIs prohibited from lending directly but can partner with commercial banks to provide lending (example: M-Shwari offered by Commercial Bank of Africa and Safaricom, where loans/savings are on the bank’s balance sheet).

Nigeria
- In 2022, MTN received the first license to operate mobile money services.
- Circular to Payment Service Banks (PSBs) in 2020: EMIs in Nigeria can apply to be a PSB to carry out deposits and withdrawals and issue debit cards.
- PSBs are not allowed to make loans, including credit card, and currently do not have options to work together with banks to make loans.
- PSBs can earn interest on e-money balances and can distribute interest income to e-money holders.
- PSBs can invest in central bank securities, government T-bills, and other short-term government securities.

Tanzania
- Under the National Payment Systems Act, EMIs must maintain e-money balances in trust funds in commercial banks.
- EMIs not allowed to make loans directly but can cooperate with regulated financial institutions to provide loan services (example: M-Pawa, cooperation between Vodacom and Commercial Bank of Africa).
- E-money operators earn interest from e-money balances; they are allowed (but not mandated) to distribute interest to e-money holders. In practice, EMIs distribute interest since 2014.
- Examples: Tigo Pesa and Airtel distribute interest in proportion to customer balances and transaction volumes.

Uganda
- National Payment Systems Regulations require e-money providers to maintain funds in trust funds held in commercial banks.
- EMIs are not allowed to make loans but can partner with banks for savings and credit products (example: MoKash launched in August 2016 by MTN Uganda and Commercial Bank of Africa).
- EMIs must keep 100 percent of e-money balances in liquid assets such as cash balances, treasury bills, and government bonds.
- Interest earned on the trust fund should be distributed to customers at the end of every quarter.

Summary of case-study regulatory commonalities (Text Table 1 synthesized into points)
- Start years and regulation/oversight exist for each country (details in source).
- Trust fund: required in Ghana, Kenya, Nigeria, Tanzania, Uganda (all yes).
- Interface with banking services: Ghana yes; Kenya yes; Nigeria yes; Tanzania yes; Uganda Yes.
- Interest earned distribution: Ghana yes; Kenya no; Nigeria no; Tanzania yes; Uganda yes.

Collective interpretation from case studies:
- These regulatory features suggest e-money development would enhance monetary policy transmission by:
  - Enhancing financial inclusion;
  - Channeling currency into the banking sector;
  - Bolstering banking sector credit creation;
  - Increasing competition in the banking sector.

### Data
- E-money data source: Financial Access Survey (FAS), annual data covering about 190 countries from 2004 to 2019. FAS provides the number of e-money registered accounts per one-thousand adults and the number of deposit accounts with commercial banks per one-thousand adults.
- Annual data on bank deposits, bank credits, government deposits, and bank credits to central government from International Financial Statistics (IFS).
- Monthly data on deposit rate, lending rate, and policy rate from IFS.
- Macro variables from the World Economic Outlook (WEO).
- Monthly interest rate data span January 2001 to December 2019.
- Baseline sample: 21 countries in the monthly data analysis and 47 countries in the annual data analysis over a 15-year period from 2004 to 2019.

*International Monetary Fund — IMF Working Papers: E-Money and Monetary Policy Transmission (Introduction section).*

### 2019. The list of countries in each panel is shown in Table 1. We exclude the post-2020

### E-Money and Monetary Policy Transmission (2019)

### Empirical method and measurement
- E-money development measured by “e-money intensity”: ratio of the number of registered e-money accounts per 1,000 adults to the number of bank accounts per 1,000 adults. Alternative measures were explored and did not change empirical findings materially.
- Strength of monetary policy transmission measured by:
  - responsiveness of bank lending and deposit rates to changes in the monetary policy rate (MPR);
  - the spread between lending and deposit rates;
  - size of bank deposits and credits.
- Estimators used:
  - Two-way fixed effect (TWFE) estimator with a single treatment (e-money introduction) for equations (1), (2), and (3).
  - Fixed-effect (FE) estimator for equation (4) (bank deposits and credit) due to concerns about the parallel trends assumption for TWFE.
- Key design choices preserved exactly:
  - Monthly change in policy rate 훥PolicyRatei,t as treatment variable; change in bank lending rate measured from t-1 to t+6 (a 6-month lag allowed for lending rate response).
  - Two e-money intensity dummies: D(emoneyL) equals 1 if e-money introduced and e-money intensity < 0.5; D(emoneyH) equals 1 if e-money introduced and e-money intensity > 0.5. Cutoff 0.5 chosen so that there is roughly an equal number of countries in two groups.
  - Standard errors clustered at the country level; country and time fixed effects included.
- Samples are split for robustness by initial level of financial inclusion (FI) measured by the number of deposit accounts per 1,000 adults prior to e-money introduction.

### Elasticity of bank lending and deposit rates with respect to the policy rate
- Baseline specification (equation (1)) includes interaction terms of monetary policy rate changes with D(emoneyL) and D(emoneyH); variables of interest are 훽2퐿 and 훽2퐻, with elasticity for low-penetration countries = 훽1+훽2퐿 and for high-penetration countries = 훽1+훽2퐻.
- Main empirical findings:
  - In the subsample of countries with low initial financial inclusion, e-money development is associated with an increase in the elasticity of the lending rate to the policy rate by 0.3.
  - Using the deposit rate (Table 4), a positive difference between 훽2퐿 and 훽2퐻 is statistically significant in the full sample and in the subsample with low initial financial inclusion; on average e-money development is associated with an increase in the elasticity of the deposit rate to the policy rate by about 0.4.

### Lending-to-deposit rate spread and deposit rate (banking sector competition)
- Specification for spread: equation (2); for deposit rate: equation (3). Controls include monetary policy rates and country/time fixed effects.
- Point estimates:
  - In countries with high e-money intensity the deposit rate is 0.81 percentage point higher.
  - In countries with high e-money intensity the lending-to-deposit spread is 0.84 percentage point lower.
  - These results are statistically significant in the full sample.
- Heterogeneity by initial financial inclusion:
  - Effects on deposit rate and spread are more pronounced in countries with low initial financial inclusion.
  - Results statistically insignificant for countries with higher initial financial inclusion.
- Interpretation: negative coefficient on spread and positive coefficient on deposit rate consistent with e-money increasing competition for household savings and, where applicable, competition for lending opportunities.

### Bank deposits and credit (size of banking sector)
- FE regression (equation (4)): 훥log(yit) = 훼i + 훽1 Δlog(intensity)it + εit, where 훥log(yit) is bank deposit growth or credit growth (log differences) and Δlog(intensity)it is growth of e-money intensity (log differences).
- Results (Tables 6–11):
  - E-money development is significantly associated with growth in both credit and deposits.
  - E-money development leads to a significant expansion in credit to the private sector (Tables 7 and 8); no evidence found for credit to the government.
  - E-money development leads to an increase in private-sector deposits (Tables 10 and 11); no evidence found for government deposits.
- Mechanisms noted: e-money can channel currency in circulation into the banking system either indirectly through MMOs or directly through banks, increasing private-sector deposits and loanable funds for credit.

### Financial inclusion
- Cross-sectional TWFE regression (equation (5)): FIit = 훼i + 훼t + 훽 D(emoney)it + controls + εit, where FI is share of banked population and controls include lagged log of number of bank branches and ATMs.
- Main findings:
  - 훽 is positive and statistically significant in the full sample and in the subsample of countries with low initial financial inclusion.
  - On average, e-money development is associated with a 4 percent increase in the share of banked population.
- Dynamics (equation (6)):
  - Coefficient not statistically significant in the year of e-money inception (Post0).
  - Coefficient becomes statistically significant and larger from one-year post inception (Post1 and Post2 and PostBeyond), indicating a gradual effect over several years.

### Policy implications
- Design and regulation of CBDC and e-money should ensure:
  - Accessibility without the need for a bank account to improve financial inclusion.
  - Regulation that encourages complementarity between e-money and banking sector growth, for example by channeling e-money balances to banks to make them available for loans.
  - Enhanced credit registries to facilitate more efficient lending by banks.
  - Encouragement of collaboration between EMIs and banks to extend credit to the private sector (financial deepening), fostering broader financial access while safeguarding financial sector stability.
- Stability consideration: only banks are typically subject to regulation and supervision while MNOs or other EMIs are typically subject to payment system oversight only; safeguarding financial sector stability is therefore important.

### Concluding remarks — summary of main findings
- All countries, irrespective of initial conditions, experience stronger monetary policy transmission with the development of e-money; transmission (elasticity of lending and deposit rates with respect to the policy rate) increases with e-money intensity, especially pronounced where initial financial inclusion is limited.
- E-money and the banking sector appear to develop in tandem, consistent with complementarity rather than pure substitution; correlation between growth in e-money and growth in deposits and credit is stronger in countries with limited initial financial inclusion.
- E-money development promotes financial inclusion (on average a 4 percent increase in the share of banked population) with effects materializing over one to several years post inception.
- Bank lending-deposit spreads tend to decline with e-money development, suggesting increased competition among banks as e-money intensity rises.

*IMF Working Paper — E-Money and Monetary Policy Transmission*

### References

### References

### Mobile money, financial inclusion, and firm/household impacts
- Ahmad, A. H., C. Green, and F. Jiang., 2020, “Mobile money, financial inclusion and development: A review with reference to African experience,” Journal of Economic Surveys, Vol. 34, No. 4, pp. 753-92.  
- Aron, J., 2018, “Mobile money and the economy: A review of the evidence,” The World Bank Research Observer, Vol. 33, No. 2, pp. 135-88.  
- Jack, W., and T. Suri, 2011, “Mobile money: The economics of M-PESA,” NBER Working Paper No. 16721 (Cambridge, Massachusetts: National Bureau of Economic Research).  
- Jack, W., and T. Suri, 2014, “Risk sharing and transactions costs: Evidence from Kenya’s mobile money revolution.” American Economic Review, Vol. 104, No. 1, pp. 183-223.  
- Ky, S., C. Rugemintwari, and A. Sauviat, 2018, “Does mobile money affect saving behaviour? Evidence from a developing country,” Journal of African Economies, Vol. 27, No. 3, pp. 285-320.  
- Gosavi, A., 2018, “Can mobile money help firms mitigate the problem of access to finance in Eastern sub-Saharan Africa?” Journal of African Business, Vol. 19, No. 3, pp. 343-60.  
- Mbiti, I., and D. N. Weil, 2015, “Mobile banking: The impact of M-Pesa in Kenya,” (citation entry ends in source text).

### Regional studies and regulatory evolution
- Bank of Ghana, 2022, The Evolution of Bank of Ghana Policies on the Ghanaian Payment System (Acura: Bank of Ghana).  
- Coulibaly, S. S., 2020, “Financial Inclusion through mobile money: an examination of the decision to use mobile money accounts in WAEMU countries.” AERC Research Paper 371 (Nairobi: African Economic Research Consortium).  
- Macmillan, R., A. Paelo, and T. Paremoer, 2016, “The “evolution” of regulation in Uganda’s mobile money sector,” The African Journal of Information and Communication, Vol. 2016, No. 17, pp. 89-110.  
- Aron, J., 2018, “Mobile money and the economy: A review of the evidence,” The World Bank Research Observer, Vol. 33, No. 2, pp. 135-88. (also relevant to regional evidence)

### Central bank digital currency (CBDC), digital money, and monetary policy
- Cœuré, B., and J. Loh, 2018, Central bank digital currencies (Basel: Bank for International Settlements).  
- Burlon, L., Muñoz, M. A., and Smets, F. Forthcoming, “The Optimal Quantity of CBDC in a Bank-Based Economy.” American Economic Journal: Macroeconomics.  
- Malloy, M. F. Martinez, M. Styczynski, A. Thorp, 2022, “Retail CBDC and US Monetary Policy Implementation: A Stylized Balance Sheet Analysis,” Finance and Economics Discussion Series No. 2022-032 (Washington, D.C.: Federal Reserve Board).  
- Das, M., Mancini Griffoli, T., Nakamura, F., Otten, J., Soderberg, G., Sole, J., and Tan, B., 2023, “Implications of Central Bank Digital Currencies for Monetary Policy Transmission.” IMF Fintech Note 23/010 (Washington, D.C.: International Monetary Fund).  
- Kahn, C., M. Singh, and J. Alwazir, 2022, “Digital Money and Central Bank Operations,” IMF Working Paper No. 2022/085 (Washington, D.C.: International Monetary Fund).  
- Brunnermeier, M. K., Limodio, N., and Spadavecchia, L., 2023, “Mobile Money, Interoperability, and Financial Inclusion.” Social Science Research Network (SSRN).  
- Chiu, J., Davoodalhosseini, S. M., Jiang, J., and Zhu, Y., 2023. “Bank Market Power and Central Bank Digital Currency: Theory and Quantitative Assessment.” Journal of Political Economy, Vol. 131, No 5, pp. .  
- Burlon, L., Muñoz, M. A., and Smets, F. Forthcoming, “The Optimal Quantity of CBDC in a Bank-Based Economy.” American Economic Journal: Macroeconomics. (listed under CBDC theory)

### Crypto, fintech cycles, and virtual currencies
- Benigno, G., and Rosa, C., 2023, “The Bitcoin–Macro Disconnect.” Federal Reserve Bank of New York Staff Reports, No. 1052 (New York; Federal Reserve Bank of New York).  
- Che, N., Copestake, A., Furceri, D., and Terracciano, T., 2023, “The Crypto Cycle and US Monetary Policy,” IMF Working Paper 23/163 (Washington, D.C.: International Monetary Fund).  
- He, D., K. Habermeier, R. Leckow, V. Haksar, Y. Almeida, M. Kashima, N. Kyriakos-Saad, H. Oura, T. S. Sedik, N. Stetsenko, and C. Verdugo-Yepes, 2016, “Virtual Currencies and Beyond: Initial Considerations,” IMF Staff Discussion Note (Washington, D.C.: International Monetary Fund).  
- Ghosh, P., Vallee, B., and Zeng, Y., Forthcoming. “FinTech Lending and Cashless Payments.” Journal of Finance.  
- Benigno, G., and Rosa, C., 2023, “The Bitcoin–Macro Disconnect.” Federal Reserve Bank of New York Staff Reports, No. 1052 (New York; Federal Reserve Bank of New York). (crypto–macro link)

### Regulatory, technical, and empirical methods
- Dias, D. and M. Kerse, 2021, “Regulatory Approaches to the Interest Earned on E-Money Float Accounts,” Technical Note (Washington, D.C.: Consultative Group to Assist the Poor).  
- De Chaisemartin, Clement, and Xavier D’haultfœuille. "Two-way fixed effects and differences-in-differences estimators with several treatments." Journal of Econometrics 236.2 (2023): 105480.  
- Malloy, M. F. Martinez, M. Styczynski, A. Thorp, 2022, “Retail CBDC and US Monetary Policy Implementation: A Stylized Balance Sheet Analysis,” Finance and Economics Discussion Series No. 2022-032 (Washington, D.C.: Federal Reserve Board). (methodological balance-sheet analysis)

*wpiea2024069-print-pdf - References*

### Chapter 7 in African Successes, Volume III: Modernization and development (Chicago:

### Chapter 7 — E-Money and Monetary Policy Transmission

### E-money and the elasticity of bank lending rates
- Changes in Policy Rate → Changes in Lending Rate: 0.86∗∗∗ (All), 0.89∗∗∗ (Low financial inclusion), 0.65∗ (High financial inclusion).
- High E-money Intensity main effect on lending rate: -0.03 (All), 0.03 (Low), -0.18 (High).
- Low E-money Intensity main effect on lending rate: 0.11 (All), 0.25 (Low), -0.1 (High).
- Interaction: High E-money Intensity × Changes in Policy Rate: 0.26 (All), 0.28∗∗ (Low), -0.05 (High).
- Interaction: Low E-money Intensity × Changes in Policy Rate: -0.03 (All), -0.02 (Low), 0.13 (High).
- Sample sizes and fit: Observations 2,903 (All), 1,102 (Low), 1,441 (High); R-squared 0.17 (All), 0.36 (Low), 0.22 (High).

### E-money and the elasticity of deposit rates
- Changes in Policy Rate → Changes in Deposit Rate: 0.94∗∗∗ (All), 0.93∗∗∗ (Low), 0.94∗∗ (High).
- High E-money Intensity main effect on deposit rate: 0.31 (All), 0.47 (Low), 0.08 (High).
- Low E-money Intensity main effect on deposit rate: 0.30∗ (All), 0.43∗ (Low), -0.02 (High).
- Interaction: High E-money Intensity × Changes in Policy Rate: 0.39∗∗∗ (All), 0.42∗∗∗ (Low), -0.44 (High).
- Interaction: Low E-money Intensity × Changes in Policy Rate: 0.21 (All), 0.24 (Low), -0.15 (High).
- Sample sizes and fit: Observations 2,977 (All), 1,102 (Low), 1,523 (High); R-squared 0.2 (All), 0.3 (Low), 0.26 (High).

### E-money, deposit rates, spreads, and bank competition
- Deposit Rate regressions:
  - High E-money Intensity coefficient on deposit rate: 0.81∗∗∗ (All), 0.78∗∗ (Low), 0.39 (High).
  - Low E-money Intensity coefficient: -0.09 (All), -0.23 (Low), -0.14 (High).
  - Policy Rate coefficient on deposit rate: 0.56∗∗∗ (All), 0.64∗∗∗ (Low), 0.47∗∗∗ (High).
  - Observations 2,987 (All), 1,081 (Low), 1,584 (High); R-squared 0.93, 0.93, 0.93 respectively.
- Spread (Lending minus Deposit) regressions:
  - High E-money Intensity coefficient on spread: -0.84∗∗∗ (All), -0.71∗ (Low), -0.48 (High).
  - Low E-money Intensity coefficient on spread: -0.04 (All), 0.22 (Low), 0.06 (High).
  - Policy Rate coefficient on spread: -0.50∗∗∗ (All), -0.54∗∗∗ (Low), -0.44∗∗∗ (High).
  - Observations 2,987 (All), 1,081 (Low), 1,584 (High); R-squared 0.92, 0.92, 0.94 respectively.

### E-money and bank credit
- Growth of Bank Credit (country-year, 2004–2019):
  - Growth of e-money intensity → Growth of bank credit: 0.04∗∗∗ (All, column 1), 0.03∗∗∗ (All with country FE, column 2), 0.05∗∗∗ (Low, column 3), 0.04∗∗ (Low with FE, column 4), 0.01 (High, column 5), 0.02 (High with FE, column 6).
  - Observations 247 in columns (1)–(2); 123 in columns (3)–(4); 98 in columns (5)–(6).
  - R-squared: 0.040, 0.049, 0.074, 0.069, 0.004, 0.013 respectively.

### E-money and credit to government
- Growth of e-money intensity → Growth of credit to government:
  - Coefficients: -0.02 (All), -0.03 (All with FE), -0.04 (Low), -0.05 (Low with FE), -0.01 (High), -0.01 (High with FE).
  - Observations 247 (columns 1–2), 123 (3–4), 98 (5–6); R-squared values near 0.000–0.004.

### E-money and credit to the private sector
- Growth of e-money intensity → Growth of credit to private sector:
  - 0.04∗∗ (All), 0.04∗∗ (All with FE), 0.05∗∗ (Low), 0.04∗ (Low with FE), 0.01 (High), 0.02 (High with FE).
  - Observations and country counts: same structure as other country-year regressions; R-squared 0.036, 0.040, 0.057, 0.047, 0.008, 0.022.

### E-money and deposits
- Growth of e-money intensity → Growth of total deposits:
  - 0.03∗∗ (All), 0.03∗∗∗ (All with FE), 0.05∗∗∗ (Low), 0.05∗∗∗ (Low with FE), -0.01 (High), 0.00 (High with FE).
  - R-squared: 0.026, 0.043, 0.068, 0.095, 0.004, 0.000.

### E-money and government deposits
- Growth of e-money intensity → Growth of government deposits:
  - Coefficients: 0.48 (All), 1.35 (All with FE), 0.98 (Low), 2.40 (Low with FE), -0.02 (High), 0.00 (High with FE).
  - Constants and noise: results imprecise; R-squared values near 0.000–0.002.

### E-money and private-sector deposits
- Growth of e-money intensity → Growth of private-sector deposits:
  - 0.03∗ (All), 0.03∗∗ (All with FE), 0.05∗∗ (Low), 0.05∗∗∗ (Low with FE), -0.01 (High), 0.00 (High with FE).
  - R-squared: 0.021, 0.034, 0.054, 0.076, 0.003, 0.000.

### E-money emergence and financial inclusion (share of depositors)
- Introduction of e-money (Post dummy effects on share of depositors, dependent variable = number of depositors in 1,000 adults divided by 10):
  - Post: 4.04∗∗ (All, col 1), 4.44∗∗ (All with controls, col 2), 8.94 (High inclusion sample, col 5 but not significant).
  - Post0 (inception year): 1.66 (All), 3.55∗∗ (All with controls), 1.55 (High).
  - Post1 (one year after inception): 3.30∗ (All), 4.98∗∗ (All with controls), 2.93 (High).
  - Post2 (two years after inception): 5.61∗∗ (All), 7.68∗∗ (All with controls), 6.91 (High).
  - PostBeyond (at least three years after introduction): 7.37∗∗ (All), 7.28∗∗ (All with controls), 21.23 (High).
  - Observations and coverage: Observations 339 (columns 1–2), 221 (cols 3–4), 73 (cols 5–6); No. of countries 32, 20, 9 respectively.
  - R-squared: 0.44–0.71 depending on sample and specification.
- Controls included lagged number of ATMs and lagged number of bank branches; country and time fixed effects implemented in all reported specifications.

*Chapter 7, African Successes, Volume III: Modernization and development — Tables and regression results on e-money intensity, monetary transmission, bank competition, credit, deposits, and financial inclusion.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024069-print-pdf.pdf_