## wpiea2024191-print-pdf

## Source details

**Canonical URL:** [wpiea2024191-print-pdf](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024191-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024191-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024191-print-pdf.pdf.json)

---

### Executive Summary — Overview
- Remittance inflows have been growing fast in recent years, with remittance flows to Asia (as a share of GDP) dominating all other financial inflows within the region.
- Literature and policy debate have mainly focused on capital flows; remittances are treated in terms of development and social safety net benefits and whether policy responses reduce exchange rate volatility.
- Governments commonly use sterilized foreign exchange interventions (FXIs) to build reserves, as an effective non-conventional monetary policy tool when the policy rate suffers from weak transmission, and to manage short-term pressures from often large, permanent financial inflows, including remittances.
- Sterilized FXIs impose costs beyond direct sterilization costs or carry costs of holding international reserves.

### Data, scope, and methodology
- Sample: 157 emerging market and developing economies; years 2000–2022.
- Definition: Remittances characterized as small, periodical transfers between private individuals largely used for basic needs, health, education, debt repayments, and intertemporal consumption smoothing.
- Regions: Africa, Asia, Eastern Europe (EE), Middle East, Caucasus, and Central Asia (MECCA), Western Hemisphere (WH).
- Key data sources and variables:
  - Remittance inflow data: KNOMAD (annual).
  - FDI and portfolio inflows: IMF's International Financial Statistics (IFS).
  - Financial depth: ratio of bank deposits to GDP; private sector bank credit to GDP; banks' holdings of public debt to GDP (World Bank GFD).
  - Financial stability: regulatory capital ratio; net open FX position-to-capital ratio (FSI).
  - FXI data: Adler et al. (2021), updated in 2023.
  - Controls: inflation, monetary policy rates, central bank reserves, trade, GDP, GDP per capita, debt-to-GDP, capital openness, institutional quality, daily exchange rate, macroprudential measures, systemic banking crises.
- Empirical approach: panel estimation with random effects; baseline regresses financial depth measures on remittances, FDI, portfolio inflows (each with up to two lags), lagged controls and event dummies (Global Financial Crisis, Taper Tantrum, Covid).

### Stylized facts and key statistics
- Remittances reached about USD 0.8 trillion (nominal terms) in 2023.
- Since 2017, remittance inflows (to GDP) exceed FDI flows (to GDP).
- Sample descriptive statistics (selected, preserved as in source):
  - Sample: 157 economies; years 2000–2022.
  - Financial Inflows (to GDP) means:
    - Remittances (t-1): Mean 4.8; Stddev 7.1; P10 0.01; Median 2.1; P90 13.4 (Number 3,018).
    - Foreign Direct Investment (t-1): Mean 4.9; Stddev 15.3; P10 0.1; Median 2.8; P90 10.1 (Number 2,851).
    - Portfolio Flows (t-1): Mean 0.8; Stddev 3.8; P10 -0.5; Median 0.02; P90 3.1 (Number 2,543).
  - Banking sector depth and stability (selected means):
    - Deposits: Mean 42.83; Stddev 32.61; P10 12.23; Median 36.18; P90 82.2.
    - Bank credit to the private sector: Mean 34.92; Stddev 27.77; P10 7.6; Median 27.46; P90 68.8.
    - Public Debt Securities in Banks: Mean 42.83; Stddev 9.5; P10 13.73; Median 32.2; P90 76.3.
    - Regulatory Capital (percent): Mean 18.6; Stddev 6.1; P10 13.3; Median 17.3; P90 25.0.
    - Net Open FX Position Ratio: Mean 17.75; Stddev 50.3; P10 -1.8; Median 4.65; P90 50.8.
  - FX intervention statistics (FXI as percent of GDP; selected means):
    - Full sample: Number 4,046; Mean 1.2; Stddev 4.9; P10 -2.7; Median 0.6; P90 5.9.
    - By region: MECCA Mean 1.7; Stddev 6.7; P10 -4.2; Median 0.9; P90 8.4.
    - Remittance inflows size: Large recipients Mean 1.1; Stddev 4.1; P10 -2.4; Median 0.7; P90 5.2.
    - Exchange rate region: Peggers Mean 1.1; Stddev 5.4; P10 -3.4; Median 0.5; P90 6.2.

### Key empirical findings — Financial depth
- Remittances and deposits:
  - Remittances tend to have a positive impact on banking system depth by increasing bank deposits in several regions, especially in the short run.
  - In aggregate, a 1 percent increase in remittances leads to a 0.5 percent increase in the bank deposits-to-GDP ratio (short term).
  - Aggregate effect on deposits is statistically insignificant overall due to strong regional heterogeneity.
  - Regional heterogeneity:
    - Positive and significant impact on bank deposits in EE and WH; only EE after two periods in some specifications.
    - WH shows a direct regional effect that disappears and becomes negative over time (coefficient -3.6, reversing previous period’s positive effects).
  - Exchange rate regime heterogeneity:
    - Peggers show a large positive and significant coefficient (+0.8) with remittances after one lag.
- Asset allocation and credit:
  - Banks adopt safe-to-safe transformation when remittances stay in the banking system; remittances associated with larger holdings of sovereign debt.
    - EE after two lags: a one percent increase in remittances increases public debt holdings by 5.8 percentage points in the second period.
  - Credit channel:
    - In EE, banks fund some private sector credit with higher deposits from remittances; in Asia, only after two periods.
    - In most other regions, remittances substitute for bank credit (consumption/investment financed through remittances rather than credit).
    - Negative and significant coefficient of remittance inflows on credit for small remittance recipients and in WH, supporting a crowding-out effect.
  - FDI and portfolio inflows:
    - FDI inflows reduce bank deposits (negative coefficient statistically significant in most specifications); WH with an FDI coefficient of -0.8 with one lag.
    - Portfolio inflows have small effects; if significant, effects are smaller than remittances.

### Key empirical findings — Financial stability
- General patterns:
  - Remittance inflows support higher solvency ratios with lags when statistically significant; mixed effects on regional net FX mismatches.
  - Remittances have mixed effects by region:
    - EE: a one percentage point increase in remittances leads to an initial decline in capital buffers that is overcompensated over time, with a net effect of 0.2 percentage points in actual regulatory capital ratios after two periods.
      - EE banks experience a net effect of 2.8 percentage points on net open FX position ratios — roughly 10 times higher impact relative to capital ratio effect (from a relatively low average level).
    - Asia: higher remittances increase bank credit and capital ratios with a two-lag delay.
    - WH: FDI tends to increase FX exposures with higher net open positions.
- Exchange rate regime effects:
  - Peg regimes: remittance inflows appear to reduce bank capital ratios over time (reduced FX risk lowers need for buffers).
  - Floating regimes: banks try to expand internal buffers (not always statistically significant).
- FDI and portfolio flows on stability:
  - FDI flows: mixed effects; may increase capital ratios with a lag for full sample; regional coefficients generally close to 0 or negative and nonsignificant.
  - Portfolio inflows: tend to reduce capital ratios with lags, especially in Asia; in EE, portfolio inflows reduce FX open positions.

### FX interventions (FXIs) — Usage and impacts
- FXIs included as central bank purchases of FX; model specifications include FXIi,t-1 and interaction terms (Rem*FXI, FDI*FXI, PI*FXI).
- Main findings:
  - FXIs, when statistically significant, have an overall negative impact on deposits, loans to private sector, and holdings of public debt securities (negative coefficients ɸ_FXI and ɸ*_FXI).
  - Mechanism: FXIs reduce exchange rate uncertainty and preserve the value of inflows → recipients face lower budget constraints → consume more and invest in higher-return assets → reduce temporary deposit holdings.
- Flow-specific marginal effects:
  - FDI: FXIs mitigate negative impact of FDI on deposits (interactive coefficient ɸ_FDI,FXI = 0.01 percentage points, positive and significant) — reduces urgency of profit repatriation and cost of holding balances locally.
  - Portfolio inflows: FXIs tend to reduce deposits and encourage channeling into higher-return opportunities.
  - Remittances: FXIs make remittance-driven deposits more temporary (coefficient ɸ2_Rem < 0, albeit statistically nonsignificant).
  - Bank credit: FXIs statistically affect bank credit when remittance interaction included; excess liquidity from large remittances prompts some additional private credit (effect small).
  - Government debt holdings: Banks reduce holdings of government debt securities with FX interventions regardless of inflow size.
- Regional heterogeneity in FXI effects:
  - WH: FXIs reduce positive impact of remittances on deposits built in t-1; two-year lag FXIs accentuate deposit withdrawal; FXIs seem to encourage holdings of government securities when remittances are high.
  - EE: Size of remittances matters—bank deposits drop but by a slightly smaller amount with FXIs when remittances are large; FXIs do not significantly impact government bond market unless inflows are large.
  - Asia: FXIs and interactions with flows statistically insignificant.
  - Low remittance recipient economies: FXIs' negative effect on deposits is moderated or can become positive if unexpectedly receiving large remittance inflows.
  - Controlling for exchange rate regime: FXIs discourage holding deposits in banking system.

### Mechanisms, interpretation, and moral hazard concerns
- Safety-net nature of remittances limits applicability of typical IPF tools (macroprudential and capital flow management) that work better for capital flows.
- Sterilized FXIs smooth FX market disorder and build reserves but have side effects:
  - Reduce perceived FX risk, lowering precautionary demand for deposits.
  - Create incentives for higher risk-taking by depositors and banks due to subsidized FX risk.
  - Undermine financial sector development by reducing deposit accumulation and bank buffers.
- FXIs act as an insurance tool that, by reducing the cost of foreign currency exposures, push agents to keep lower balances in banks and banks to hold lower buffers, encouraging moral hazard behavior.
- Behavioral and balance-sheet effects add to other FXI-related direct costs already identified (sterilization costs, carry costs of holding international reserves).

### Robustness, specification tests, and controls
- Methods and tests:
  - Models estimated using panel regression techniques with a random-effects component; Hausmann test confirmed no correlation between explanatory variables and error term.
  - Instrumenting inflows with output growth attempted but instrument not strong enough for inference.
  - Breusch-Pagan test confirmed homoscedasticity of residuals for regressions (1)-(6); elsewhere a contrasting note indicates rejection of homoscedasticity at the individual level in other tests.
  - Random effects model appeared more robust than fixed effects; results robust across pooled OLS, random effects, and fixed effects estimations.
  - Time fixed effects inclusion did not change policy results.
- Notable control variable findings:
  - Log of GDP positively associated with financial development, especially in EE.
  - Inflation negatively impacts financial development, especially in EE and pegger countries.
  - Higher FX volatility penalizes deposits.
  - Global Financial Crisis: negative impact on financial depth.
  - COVID-19: improved level of financial development as captured by deposits (lockdowns → inability to spend → funds stayed in financial system).
  - Monetary policy rate used to abstract from unsterilized FX interventions; effect varies by region.
  - Systemic banking crises associated with higher actual regulatory capital ratio (post-crisis risk aversion and tighter regulation).
  - Debt-to-GDP increases regulatory ratio (precautionary behavior).
  - Macroprudential policy tightening impact largely zero and statistically insignificant (limited variation before 2008).

### Conclusions and policy implications
- Core conclusions:
  - Remittance inflows differ from FDI and portfolio investment although all inject FX and liquidity; drivers and impacts differ.
  - Remittances tend to increase deposits and sovereign debt holdings, with important regional and exchange regime heterogeneity.
  - FXIs have costs for the banking sector: they reduce deposits in the banking sector and encourage risk-taking away from the banking sector.
  - When remittance inflows are quite large, FXIs can have a positive impact in specific circumstances.
- Policy recommendations:
  - Use FXIs sparingly and primarily to address episodes of large remittance inflows with disorderly conditions (in line with IMF 2023c principles).
  - Develop private-sector risk-management instruments to transfer market risk from the public sector to the private sector.
  - Complement short-term policies with longer-term structural reforms to increase absorptive capacity and avoid macrofinancial imbalances.
  - Policymakers should weigh FXI benefits (exchange rate stability, reserve accumulation, protection of tradable sectors) against medium-term costs to banking depth and financial stability.
  - Integrated policy frameworks (monetary policy, FX intervention, capital flow management, macroprudential policies) remain relevant, but many tools used for capital inflows are less suitable for regular, low-value remittance inflows.

*International Monetary Fund — Executive Summary (wpiea2024191-print-pdf)*

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

### Executive Summary

### Overview
- Remittance inflows have been growing fast in recent years, with remittance flows to Asia (as a share of GDP) dominating all other financial inflows within the region.
- The literature and policy debate have largely focused on capital flows and their impact; remittance flows have been treated mainly in terms of development and social safety net benefits and whether policy responses reduce exchange rate volatility.
- Governments commonly use sterilized foreign exchange interventions (FXIs) to build reserves, as an effective non-conventional monetary policy tool when the policy rate suffers from weak transmission, and to manage short-term pressures from often large, permanent financial inflows, including remittances.
- These sterilized FXIs impose costs beyond direct sterilization costs or carry costs of holding international reserves.

### Data and Scope
- The analysis covers a sample of 157 emerging market and developing economies.
- Remittances are characterized as small, periodical transfers between private individuals largely used to cover basic needs, health, education, and debt repayments, and for intertemporal consumption smoothing (spending).

### Key empirical findings
- FXIs are quick, temporary fixes that:
  - Hinder the development of the recipient country’s financial sector.
  - Often do not contribute to financial stability over the medium term.
- While remittances can help build bank deposits (supporting financial depth), sterilized FXIs, when statistically significant, tend to partially offset these positive effects, especially when remittance inflows are not that large.
- FXIs remove market risk uncertainty and ameliorate FX volatility, which can:
  - Ease recipients’ credit constraints.
  - Reduce the need for maintaining deposits to ensure intertemporal consumption smoothing.
  - Encourage spending, the main use associated with remittances according to countries’ household surveys.
- The negative effect of FXIs on financial depth may attenuate when receiving large flows of remittances, with extra resources being kept in bank deposits for later use.
- On financial stability, FXIs function as an insurance tool that, by reducing the cost of foreign currency exposures, push agents to keep lower balances in banks and banks to hold lower buffers.
- By subsidizing FX-related risks, FXIs protect remittance recipients and tradable sectors but encourage higher risk-taking (moral hazard) behavior.
- These behavioral and balance-sheet effects add to other FXI-related direct costs already identified in the literature (e.g., sterilization costs, carry costs of holding international reserves).

### Mechanisms and interpretation
- The safety-net nature of remittances (private, periodic, consumption-oriented) limits applicability of typical IPF policy tools (macroprudential and capital flow management measures) that work better for capital flows.
- Sterilized FXIs therefore act to smooth FX market disorder and reserve accumulation but have side effects:
  - They reduce perceived FX risk, lowering precautionary demand for deposits.
  - They create incentives for higher risk-taking by both depositors and banks due to subsidized FX risk.
  - They can undermine the development of the financial sector by reducing deposit accumulation and bank buffers.

### Policy implications and recommendations
- In line with IMF (2023c) key principles for the use of FXIs:
  - FXIs should be used sparingly and only as a tool to deal with particularly large remittance inflows that cause disorderly conditions.
  - The development of private sector risk management tools should support longer-term structural reforms to increase the absorptive capacity of additional FX inflows.

_International Monetary Fund — Executive Summary (wpiea2024191-print-pdf)_

### Introduction

### Introduction

### Policy context and challenges
- Remittance inflows have shown double-digit growth trends and, in several countries, account for over 20-30 percent of GDP.
- Remittances function as a safety net that funds consumption among vulnerable segments and provide a net inflow of foreign exchange.
- Costs and macroeconomic pressures associated with remittances:
  - Upward pressure on the exchange rate and on domestic prices.
  - Injection of liquidity that needs to be managed.
  - Potential to undermine exporters' competitiveness and create Dutch disease by altering the consistency between the external position and fundamentals.
  - Excessive international reserves could be costly for central banks' income models.
- Policy toolkit limitations:
  - Because of remittances’ social role and income-account nature, many standard tools for managing capital inflows are less suitable.
  - Foreign exchange interventions (FXIs) have been widely used, often sterilized to some degree, as the primary policy response.

### This paper: focus, scope, and methodology
- Objective:
  - Examine the effects of remittance flows on the banking sector and evaluate the performance of FXIs as a policy response to remittance inflows using cost-benefit criteria applied to the financial sector.
- Definitions in the cost-benefit framework:
  - Costs: how FXIs hinder financial sector development via bank deposits, credit to the private sector, and sovereign debt holdings.
  - Benefits: potential stability gains—higher regulatory capital ratios and lower net open FX positions.
- Data and sample:
  - Financial flows analyzed for 157 emerging and developing economies during the years 2000-2022.
  - Five regions: Africa, Asia, Eastern Europe (EE), Middle East, Caucasus, and Central Asia (MECCA), and Western Hemisphere (WH).
- Key variables and sources:
  - Remittance inflow data: KNOMAD (annual).
  - FDI and portfolio inflows: IMF's International Financial Statistics (IFS).
  - Financial depth measures (from World Bank GFD): ratio of bank deposits to GDP; private sector bank credit to GDP; banks' holdings of public debt to GDP.
  - Financial stability measures (from FSI): regulatory capital ratio (regulatory capital-to-risk weighted assets); net open FX position-to-capital ratio.
  - FXI data: Adler et al. (2021), updated in 2023.
  - Controls: inflation, monetary policy rates, central bank reserves, trade, GDP, GDP per capita, debt-to-GDP, capital openness (Chinn & Ito (2008)), institutional quality (WGI), daily exchange rate (Bloomberg), macroprudential measures (Alam et al. (2019)), systemic banking crises (Laeven & Valencia (2018)).
- Empirical approach:
  - Panel estimation with random effects.
  - Baseline specification regresses financial depth measures on remittances, FDI, and portfolio inflows (each with up to two lags), plus lagged controls and event dummies (Global Financial Crisis, Taper Tantrum, Covid).
  - Explicit equation uses lagged remittance ratios Remi,t−k, FDIi,t−k, PIi,t−k and controls X_i,t−1.

### Stylized facts from the data
- Remittances:
  - Remittance inflows have been growing fast, stable, and long-lasting over the past 20 years, dwarfing other financial inflows in aggregate.
  - Since 2017, remittance inflows (to GDP) exceed FDI flows (to GDP).
  - Remittances reached about USD 0.8 trillion (nominal terms) in 2023.
- Regional patterns:
  - Asia showed the largest increase in remittance inflows, followed by WH and EE (KNOMAD).
  - MECCA accumulated the largest FX balances via FXIs between 2000 and 2022—average FX balance via FXIs is 1.7 percent of GDP.
  - EE economies also accumulated FX reserves via FXIs (weaker than MECCA). WH and Africa rank lower in FXI accumulation.
- Summary descriptive statistics (selected, preserved as in source):
  - Sample: 157 economies; years 2000–2022.
  - Financial Inflows (to GDP) means:
    - Remittances (t-1): Mean 4.8; Stddev 7.1; P10 0.01; Median 2.1; P90 13.4 (Number 3,018).
    - Foreign Direct Investment (t-1): Mean 4.9; Stddev 15.3; P10 0.1; Median 2.8; P90 10.1 (Number 2,851).
    - Portfolio Flows (t-1): Mean 0.8; Stddev 3.8; P10 -0.5; Median 0.02; P90 3.1 (Number 2,543).
  - Banking sector depth and stability (selected means):
    - Deposits: Mean 42.83; Stddev 32.61; P10 12.23; Median 36.18; P90 82.2.
    - Bank credit to the private sector: Mean 34.92; Stddev 27.77; P10 7.6; Median 27.46; P90 68.8.
    - Public Debt Securities in Banks: Mean 42.83; Stddev 9.5; P10 13.73; Median 32.2; P90 76.3.
    - Regulatory Capital (percent): Mean 18.6; Stddev 6.1; P10 13.3; Median 17.3; P90 25.0.
    - Net Open FX Position Ratio: Mean 17.75; Stddev 50.3; P10 -1.8; Median 4.65; P90 50.8.
  - FX intervention statistics (FXI as percent of GDP; selected means):
    - Full sample: Number 4,046; Mean 1.2; Stddev 4.9; P10 -2.7; Median 0.6; P90 5.9.
    - By region: MECCA Mean 1.7; Stddev 6.7; P10 -4.2; Median 0.9; P90 8.4.
    - Remittance inflows size: Large recipients Mean 1.1; Stddev 4.1; P10 -2.4; Median 0.7; P90 5.2.
    - Exchange rate region: Peggers Mean 1.1; Stddev 5.4; P10 -3.4; Median 0.5; P90 6.2.

### Key empirical findings
- Impact of remittances on financial depth and stability:
  - Remittances tend to have a positive impact on banking system depth by increasing bank deposits in a number of regions, especially in the short run.
  - In aggregate, a 1 percent increase in remittances leads to a 0.5 percent increase in the bank deposits-to-GDP ratio (short term).
  - Remittances can be associated with higher holdings of public debt securities by banks (a safe-to-safe strategy) funded through higher deposit holdings.
  - FDI inflows to remittance-receiving economies tend to lead to lower bank deposits, consistent with banks playing a pure intermediation role for FDI.
  - Portfolio inflows have a positive, albeit statistically insignificant, effect on deposits in this sample.
  - Remittances appear to support higher solvency (regulatory capital) with a lag when statistically significant.
  - Remittances have a mixed effect on net FX mismatches that is region-dependent:
    - Tend to be associated with lower mismatches in Asia.
    - Tend to be associated with higher mismatches in EE.
    - Domestic net open position regulations and their enforcement can help explain regional differences.
- Effects of FX interventions (FXIs):
  - FXIs tend to reduce the benefits of remittance inflows for the banking sector.
  - FXIs operate as a subsidy benefiting remittance recipients and the tradable sector at the expense of the central bank and the depth and financial stability of the banking sector.
  - By reducing market-risk uncertainty, FXIs support recipients’ objectives to finance consumption, reinforcing the consumption-smoothing role of remittances.
  - For the full sample and especially EE, larger remittance inflows encourage keeping additional resources as deposits for future spending, but this does not offset the negative effect of FXIs on deposits.
  - Evidence points to FXIs imposing medium-term costs on the banking sector in addition to direct costs borne by central banks.
  - Costs associated with FXIs include direct fiscal costs, potential inconsistencies between monetary and FX policies, induced risk-taking, reduced development of private FX and hedging markets, and shallower financial buffers in the banking sector.

### Policy implications and recommendations
- Use FXIs sparingly:
  - FXIs should be used sparingly and primarily to address episodes of large remittance inflows with disorderly conditions.
- Develop private-sector risk-management instruments:
  - Renewed efforts are needed to develop risk management instruments that transfer market risk from the public sector to the private sector.
- Complement short-term policies with structural reforms:
  - Longer-term structural reforms are required to create space to absorb inflows without creating macrofinancial imbalances.
- Consider the trade-offs:
  - Policymakers should weigh FXI benefits (exchange rate stability, reserve accumulation, financial-sector buffers) against medium-term costs to banking depth and financial stability.
- Follow integrated frameworks:
  - General principles from integrated policy frameworks (monetary policy, FX intervention, capital flow management, macroprudential policies) remain relevant, but many tools used for capital inflows are less suitable for regular, low-value remittance inflows.

*INTERNATIONAL MONETARY FUND*

### 0.2 percent after two periods (Table 5a, first column). The results are in line with other studies.

### wpiea2024191-print-pdf - 0.2 percent after two periods (Table 5a, first column). The results are in line with other studies.

### Remittances, bank deposits, and bank asset allocation
- Prior literature:
  - Demirgüç-Kunt et al. (2006): panel of 99 low-income remittance recipients (1975-2003) — remittances increase deposits to GDP by 0.5-0.6 percent when remittances increase by one percent.
  - Aggarwal et al. (2016) for Mexico: statistically significant causal link between remittances and banking sector development using micro-level data and instruments.
- Aggregate findings in this study:
  - Aggregate effect on deposits is statistically insignificant overall.
  - Strong regional heterogeneity:
    - Positive and significant impact on bank deposits in EE and WH; only EE after two periods (Table 5a, second to fourth columns).
    - Direct regional effect disappears and becomes negative in WH over time (coefficient -3.6, reversing previous period’s positive effects), indicating temporal nature of deposits.
  - Exchange rate regime heterogeneity:
    - Peggers show a large positive and significant coefficient (+0.8) with remittances after one lag (Table 5b).
- Asset-side behavior:
  - Banks adopt a safe-to-safe transformation strategy when remittance inflows stay in the banking system (third column specifications in Table 5a).
  - Economies receiving higher remittance inflows are associated with larger holdings of sovereign debt.
    - EE region after two lags: a one percent increase in remittances increases public debt holdings by 5.8 percentage points in the second period.
  - Credit channel:
    - In EE, banks appear to fund some private sector credit with higher deposits from remittances; in Asia, only after two periods.
    - In most other regions, remittances substitute for bank credit (consumption/investment financed through remittances rather than credit).
    - Negative and significant coefficient of remittance inflows on credit for small remittance recipients (Table 5b) and in WH economies, supporting a crowding-out effect.
  - FDI and portfolio inflows:
    - FDI inflows reduce bank deposits (negative coefficient statistically significant in most specifications), notably WH with an FDI coefficient of -0.8 with one lag (possible profit repatriation).
    - Portfolio inflows generally show small effects; if statistically significant, effects are smaller than remittances.

### Impact of remittances and other flows on financial stability (specification and measures)
- Empirical specification (2): dependent variable FS (financial stability) with FSi,t ∈ {actual regulatory capital ratio, net open foreign exchange position to bank capital}.
- Controls Z = {exports-to-GDP, imports-to-GDP, private credit-to-GDP, macroprudential policy, monetary policy rate, systemic banking crisis dummy, debt-to-GDP, GDP per capita (in PPP U.S.$), central bank reserves-to-GDP and regulatory capital}.
- Key coefficients: {γ1*Rem, γ2*Rem} capture remittances’ effects on financial stability.

### Financial stability results
- General pattern:
  - Remittance inflows support higher solvency ratios with lags when statistically significant; mixed effects on regional net FX mismatches.
- Region-specific effects:
  - EE:
    - A one percentage point increase in remittances leads to an initial decline in capital buffers that is overcompensated over time, with a net effect of 0.2 percentage points in actual regulatory capital ratios after two periods.
    - EE banks experience a relatively significant net effect of 2.8 percentage points on their net open FX position ratios (Table 6a) — roughly 10 times higher impact relative to capital ratio effect (albeit from a relatively low average level).
    - Higher remittance inflows intermediated through banking lead to higher bank credit to private sector relative to deposits (including FX-denominated), exacerbating EE banks’ exposure to FX risk.
  - Asia:
    - Higher remittances increase bank credit and capital ratios with a two-lag delay.
  - WH:
    - FDI tends to increase FX exposures with higher net open positions.
- Exchange rate regime effects:
  - Peg regimes: remittance inflows appear to reduce bank capital ratios over time (lack of FX risk reduces need for buffers).
  - Floating regimes: banks try to expand internal buffers (signs not always statistically significant).
- FDI and portfolio flows on stability:
  - FDI flows: mixed effects; higher FDI may increase capital ratios with a lag for full sample; regional coefficients generally close to 0 or negative and nonsignificant (Table 6a).
  - Portfolio inflows: tend to reduce capital ratios with lags, especially in Asia; in EE, portfolio inflows reduce FX open positions, indicating temporality of these inflows.
- Interaction of exchange rate regime and remittance dependence (Table 6b):
  - Peggers: remittances reduce bank capital ratios over time.
  - Floating or low-remittance economies: FDI positively impacts capital ratios; portfolio inflows associated with lower capital ratios.
  - In economies receiving large remittance inflows, additional FDI or portfolio inflows translate into higher FX exposures in the banking sector.

### FX interventions (FXIs): usage and impacts on financial depth and stability
- Policy focus: central bank purchases of FX (FXIs) and interaction with remittance, FDI, and portfolio inflows.
- Model extensions: specifications (3) and (4) include FXIi,t-1 and interaction terms (Rem*FXI, FDI*FXI, PI*FXI) estimated with panel random effects.
- Main findings:
  - FXIs, when statistically significant, have an overall negative impact on deposits, loans to private sector, and holdings of public debt securities (negative coefficients ɸ_FXI and ɸ*_FXI in Table 7a).
  - Mechanism: FXIs reduce exchange rate uncertainty and preserve value of inflows → recipients face lower budget constraints → consume more and invest in higher-return assets → reduce temporary deposit holdings.
- Distinct marginal effects by flow type:
  - FDI:
    - FXIs mitigate the negative impact of FDI on deposits (interactive coefficient ɸ_FDI,FXI = 0.01 percentage points, positive and significant in Table 7a) — reduces urgency of profit repatriation and cost of holding balances locally.
  - Portfolio inflows:
    - FXIs tend to reduce deposits; provide FX stability that increases incentives to channel portfolio investments into higher-return opportunities.
  - Remittances:
    - FXIs make remittance-driven deposits more temporary (coefficient ɸ2_Rem < 0, albeit statistically nonsignificant).
  - Bank credit:
    - FXIs statistically affect bank credit when remittance interaction included; excess liquidity from large remittances is costly for banks and prompts some additional private credit (effect small).
  - Government debt holdings:
    - Banks reduce holdings of government debt securities with FX interventions regardless of inflow size.
- Regional heterogeneity (Table 7b, Table 8):
  - WH:
    - FXIs reduce positive impact of remittances on deposits built in t-1; two-year lag FXIs further accentuate deposit withdrawal (specification (2b)).
    - FXIs seem to encourage holdings of government securities; higher remittance inflows → higher holdings (Table 8).
  - EE:
    - Size of remittances matters: bank deposits drop but by a slightly smaller amount with FXIs when remittances are large (specification (3a)).
    - FXIs do not significantly impact government bond market unless remittance or portfolio inflows are large enough.
  - Asia:
    - FXIs and interactions with all financial flows remain statistically insignificant (Table 7c).
  - Low remittance recipient economies:
    - FXIs' negative effect on remittances is moderated or can become positive if unexpectedly receiving large remittance inflows (Table 7d) — indicating a positive role for FXIs in cases of punctual large inflows.
  - Controlling for exchange rate regime: FXIs discourage holding deposits in banking system.

### Robustness checks and control variables
- Estimation checks:
  - Random effects estimation shown; similar results obtained under OLS pooled estimation (not reported in annex but available from authors).
  - Re-estimated models with fixed effects (results referenced).
- Control variable findings:
  - Level of economic development (log of GDP coefficient) positively associated with financial development, especially in EE; positive coefficient preserved for Asia but significance disappears.
  - Size of economy and size of remittance inflows improve financial depth (size of remittance inflows result not shown but available from authors).
  - Inflation has a negative impact on financial development, especially in EE and pegger countries.
  - Higher FX volatility penalizes deposits by increasing financial market uncertainty.
  - Crisis episodes:
    - Global Financial Crisis: negative impact on financial depth.
    - COVID-19: improved level of financial development as captured by deposits (national lockdowns → inability to spend → funds stayed in financial system); similar effects for private credit and public debt securities.
  - Monetary policy rate included to abstract from non-sterilized FX interventions (effect varies by region).
  - Capital and trade openness: positive but statistically insignificant.

*Source: wpiea2024191-print-pdf.*

### Annex III) confirm the findings of the baseline analysis. The inclusion of time fixed effects also did not

### Annex III

### Methods and Specification Tests
- Models estimated using panel regression techniques with a random-effects component.
- Instrumenting inflows with output growth using an IV estimator was attempted but “the instrument was not strong enough to make any inference.” IV estimation results are available on request.
- The Hausmann specification test “confirmed no correlation between the explanatory variables and the error term.”
- When comparing models, “the random effects model appeared to be more robust than the fixed effects model.”
- The Breusch-Pagan test “confirmed homoscedasticity of the residual of both models, financial depth and stability, i.e., regressions (1)-(6).”
- Time fixed effects inclusion “did not change the policy results of the paper.”
- Baseline models for financial stability extend model (2) to include FXI and interaction terms (financial flows * FXIs); equations (5) and (6) in the source specify the functional form used.

### Foreign Exchange Intervention (FXI) and Remittances — Stability Effects
- Research question: “Do positive FX interventions reduce the vulnerabilities in the banking sector by increasing regulatory capital ratios and reducing exposures to foreign currency?” The paper notes a potential moral hazard: FXI “subsidization” that lowers FX volatility may “further increasing FX risk exposures within the banking sector and reducing buffers in the system (moral hazard behavior).”
- FXIs characterized as “discretionary, non-conventional monetary policy tool[s] used by authorities to smooth exchange rate volatility.” They are costly and may provide a “false sense of comfort.”

### Empirical Results — Aggregate Sample
- Sample: “157 emerging market and developing economies from 2000 to 2022.”
- Table 9a (specifications (3-3’)) findings for emerging markets and developing countries:
  - FXIs (i) “reduce current regulatory capital ratios, reducing the positive effect of remittances in strengthening banks’ buffers,” and
  - (ii) “exacerbate banks’ net FX open positions,” although the latter result is “statistically insignificant.”
  - Results hold when controlling for the level of central bank reserves (lagged by one period).
- Mechanisms described:
  - FXIs “reduce deposits,” which “reduce the required holdings of Common Equity Tier 1 capital and thus the institution’s regulatory requirements under the Basel framework.”
  - By investing in high-quality public debt, banks are “not required to hold higher buffers,” reinforcing the result.

### Regional Findings
- Table 9b (regional analysis):
  - FXIs “exacerbate banking sector stability risks while having little impact on FX exposures in banks.”
  - In the EE and WH regions, FXIs “reduce banks’ capital ratios with coefficients being negative and statistically significant.”
  - “With large enough remittance inflows, however, the negative impact can be offset.”
  - FXIs appear to have “positive but non-significant effect on banks’ net open positions.”
  - Interaction terms: “higher remittances to EE (controlling for FXIs) appear to increase banks' net open FX positions by about 0.1 percentage point (specification 3b).”
  - Summary for EE region: FXIs reduce banks’ capital ratios (specification 3a); large remittance inflows mitigate capital buffer reductions but “FX mismatches would appear to be exacerbated (specification 3b).”
  - WH region results are “non-statistically significant in the most part.”
- Table 9c: results confirm that “when statistically significant, FXIs result in lower banks’ buffers (specifications 1’b-2’b in Table 9c).”

### Financial Depth and Flow-Specific Findings
- Aggregate-level effects of inflows:
  - Remittance inflows “tend to increase the level of deposits, and sovereign debt holdings.”
  - At the aggregate, remittances “do not appear to have a statistically significant effect on deposits.”
  - Strong regional component:
    - Remittances have a “positive and significant effect on bank deposits in EE and WH.”
    - Strong FX regime effect: “peggers showing a large positive and significant effect, but with a lag.”
  - Impact on public debt holdings is “lagged in the EE region.”
  - When confronted with large remittance inflows or a floating exchange regime, “banks appear to opt for a safe bank deposit-to-government securities investment strategy.”
  - FDI inflows “negatively impact bank deposits for the whole sample,” possibly due to profit repatriation.
  - Portfolio inflows “don’t appear to have a significant effect except on sovereign debt holdings, with a drop.”
- Financial stability outcomes:
  - For the full sample, remittances “do not appear to impact significantly banks’ regulatory capital ratios or banks’ net open FX positions,” especially when seen as temporary.
  - Regionally, and with lags, “remittances help increase capital buffers in EE and Asia.”
  - FDI appears to have a “positive impact on banks’ buffers in economies with floating exchange rate regimes and on FX exposures in economies receiving large remittance inflows.”
  - Portfolio inflows show mixed results, “except on government debt holdings, which decline.”

### Robustness Checks
- Robustness confirmed across estimation methods: pooled ordinary least squares, random effects (Table 1) and fixed effects (Table 2).
- Tables 1-3 (Annex IV) report controls used to analyze financial stability effects of FXIs using actual regulatory capital ratio as dependent variable and applying random effects (Table 1) and fixed and time effects estimation (Tables 2 and 3).
- Estimated regressions using pooled OLS show “similar patterns.”
- Controls and notable control results:
  - Coefficient on systemic banking crises is “positive and statistically significant” on the actual regulatory capital ratio, attributed to “risk-averse behavior of banks and tighter regulatory measures taken by financial supervisors following a banking crisis episode.”
  - Higher debt-to-GDP ratio “increases the regulatory ratio,” suggesting precautionary behavior in response to higher fiscal risks.
  - GDP per capita “supports a higher actual regulatory ratio,” implying higher capital buffers in more developed economies.
  - Monetary policy rate used to abstract from unsterilized FXIs “has a zero effect.”
  - Macroprudential policy tightening impact is “zero and statistically insignificant,” explained by “limited variation in the data for macroprudential policy before 2008.”
  - The effect of the lagged regulatory capital ratio “varies by the size of remittance inflows and is region-specific.”
- Note: a contrasting statement notes that “The Breusch-Pagan test also rejected homoscedasticity at the individual level,” indicating heteroscedasticity in a different test context.

### Conclusions and Policy Implications
- Context: “A growing body of literature has highlighted the trend of remittance inflows taking over from other financial inflows.” Literature gap on remittances’ effects on monetary and financial sectors is noted.
- Cost‑benefit framework: the paper points to the Integrated Policy Framework (IPF) approach, advocating “multiple policy tools based on shocks and country characteristics” including “monetary policy, foreign exchange intervention, capital controls, and domestic macroprudential measures.”
- Key conclusions:
  - Remittance inflows differ from FDI and portfolio investment though all inject foreign exchange and liquidity; drivers and impacts differ.
  - Remittances tend to increase deposits and sovereign debt holdings, with important regional and exchange regime heterogeneity.
  - FXIs have costs for the banking sector: FXIs “reduce the level of deposits in the banking sector and encourage risk-taking away from the banking sector.”
  - However, “when remittance inflows are quite large, FXIs can have a positive impact.”
- Policy recommendations:
  - “The remittance strategy should be geared towards the less use of FXIs while risk management tools are developed for the private sector to take over and hedge against FX volatility.”
  - In line with IMF principles for FX intervention (IMF 2023c), “there is a case for FXIs, when remittance inflows are large enough to cause disorderly conditions in the economy.”
  - Structural policies that create conditions to absorb continued remittance inflows are “the longer-term and sustainable response to these flows.”

*Source: Excerpt from Annex III and related sections of the IMF Working Paper (wpiea2024191-print-pdf).*

### Annex II. Financial Inflows by Region, 2000-2020

### Annex II. Financial Inflows by Region, 2000-2020

### Figure 1. Financial Inflows by Region
- All Countries, ....
- In Eastern Europe, FDI dominates all inflows in these past years....
- In WHD, remittances just caught up FDI ....
- In Asia, remittances continue to dominate ....
- Remittances have been declining in ME economies...      ...and in Africa in favor of FDI in the recent years

*Source: Authors and IMF database.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024191-print-pdf.pdf_