## _wp0652

## Source details

**Canonical URL:** [_wp0652](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0652.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0652.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0652.pdf.json)

---

### I. Key findings on remittance scale and growth
- Total remittance receipts by developing countries reached 116 billion dollars in 2003, representing more than 1.5 percent of their total GDPs as a group.
- Remittance receipts grew steadily after 1990 with a slight decline in 1998.
- Share of remittance receipts (1990-2003 sample context):
  - Low-income countries: 32 percent of total remittance receipts of developing countries.
  - Middle-income countries: 68 percent.
- Average annual growth rates of remittance receipts since 1990:
  - Low income countries: 12.3 percent.
  - Middle income countries: 9.9 percent.
- Average annual growth rates of remittance receipts since 1999:
  - Low income countries: 15.9 percent.
  - Middle income countries: 10.8 percent.

### II. Importance and characteristics of remittances
- Remittances are an increasingly important channel for meeting external financing needs and are one of the largest sources of such financing.
- Remittances are generally less volatile and hence more dependable than private capital flows and foreign direct investment (FDI).
- As unilateral transfers, remittances do not create future liabilities such as debt servicing or profit transfers.

### III. Cyclicality, methodology, and main empirical result
- Cycles are defined as deviations of real variables from their respective trends following Lucas (1977) and Kydland and Prescott (1990).
- Analysis focuses on co-movements between deviations from trend of real remittances and deviations from trend of real GDP (i.e., cyclical components), rather than on long-run correlations from regression coefficients.
- Main empirical finding for the sample of 12 low-income (LI) and lower-middle income (LMI) countries:
  - For the group as a whole, remittances are countercyclical and lead the aggregate GDP cycle by one period.
  - At the individual-country level, remittances can be procyclical or acyclical for some countries; panel results may conceal important country-specific characteristics.

### IV. Mechanisms, caveats, and transmission risks
- Countercyclicality interpretation: migrant savings remitted to home countries tend to increase after periods of stagnation/crisis at home and decrease after periods of growth/boom for the group aggregate.
- Decision to remit is complex and driven by multiple factors beyond altruistic financing of current consumption; different drivers may imply procyclical, countercyclical, or acyclical behavior across countries.
- Remittances should also respond to the state of economic activity in host countries; synchronized cycles between home and host economies can limit the ability of migrants to support families during downturns at home.
- Remittance flows can transmit contractions in host economies to recipient countries through synchronized reductions in amounts remitted.
- Example of vulnerability: the 1990-91 Middle East conflict had disastrous consequences for economies receiving large amounts of remittances from Kuwait and other countries in the region.

### V. Selected 2001-level ratios highlighting importance of remittances
- GDP share of remittances as of 2001:
  - Low-income countries: 1.9 percent.
  - Lower-middle-income countries: 1.4 percent.
  - Upper-middle-income countries: 0.8 percent.
- Ratios of remittances to imports in 2001:
  - Low-income countries: 6.2 percent.
  - Lower-middle-income countries: 5.1 percent.
  - Upper-middle-income countries: 2.7 percent.

### Remittances: trend, volatility, and countercyclicality
- Remittances stood out among other sources of external financing by 2001 in UMI countries.
- Remittances are often observed to be generally less volatile than private capital flows that move procyclically with output in recipient countries.
- Evidence indicates a negative relationship between remittances and income for different countries and cross-country evidence that remittances would reduce the size of worst drops in GDP experienced during severe economic crises.
- Sharp increases in remittances after crises are documented for Indonesia (1997), Ecuador (1999) and Argentina (2001).

### Microeconomic motivations and behavioral complexity
- Migrant workers plausibly increase transfers to family members during domestic downturns to help compensate for lost family income due to unemployment or crisis-induced reasons.
- Remitting behavior is multifaceted and affected by many variables beyond altruism, including interest rate differentials and exchange rates, which can produce procyclicality, countercyclicality, or acyclicality depending on country-specific conditions.
- Given differing implications of countercyclicality versus procyclicality for stability and for using remittances as collateral, cyclical nature should be investigated at the country level rather than relying solely on group evidence.

### Data, sample, and detrending methodology
- Annual data used cover 1976–2003.
- Sample: 12 countries (six LI and six LMI):
  - LI: Bangladesh (BGD), India (IND), Côte d’Ivoire (CIV), Lesotho (LSO), Pakistan (PAK), Senegal (SEN).
  - LMI: Algeria (DZA), Dominican Republic (DOM), Jamaica (JAM), Jordan (JOR), Morocco (MAR), Turkey (TUR).
- Combined share of the 12 countries in total remittance receipts of all LI and LMI countries:
  - 86.2 percent in 1976 and 39.8 percent in 2003 (27.5 percent for the six LI countries and 12.3 percent for the remaining six).
- Real remittances series converted from nominal U.S. dollar terms to real terms using the U.S. GDP deflator (2000=100). Real GDP measured in constant US dollars at 2000 prices.
- Trends were removed by fitting polynomials of degree k to each series:
  - GDP trend polynomial form: y_t = α0 + α1 t + α2 t^2 + ... + α_k t^k
  - Remittances trend polynomial form: r_t = β0 + β1 t + β2 t^2 + ... + β_k t^k
- Trend polynomials chosen so estimated coefficients were statistically significant and detrended series passed stationarity checks (ADF).
- Aggregate (group) series constructed by weighting country outputs and remittances by average shares in total real GDP and total remittances of all LI and LMI countries.

### Estimated trends (group of 12)
- Estimated trend for aggregate real GDP (weighted total) — third-order polynomial (T_t y):
  - Constant: 4.916883 (2341.189)
  - t: 0.006593 (9.610)
  - t^2: 0.000189 (3.172)
  - t^3: -4.52E-06 (-3.115)
  - R^2: 0.998 [Adj.R^2 0.998]
  - F-statistic: 4626.6
- Estimated trend for aggregate real remittances (weighted) — fifth-order polynomial (T_t r):
  - Constant: 2.287111 (140.928)
  - t: 0.125088 (9.681)
  - t^2: -0.022463 (-7.234)
  - t^3: 0.001679 (5.644)
  - t^4: -5.43E-05 (-4.435)
  - t^5: 6.33E-07 (3.511)
  - R^2: 0.942 [Adj.R^2 0.929]
  - F-statistic: 72.1

### Identification of cyclical comovements
- After detrending, cyclical components (stationary with zero mean) were used to compute contemporaneous and asynchronous cross correlations between real output and real remittances.
- Asynchronous correlations considered remittances at t-1, t, and t+1 relative to output at t to detect leads/lags and phase shifts.

### Main results (summary)
- Group-level behavior (whole sample, 1976–2003):
  - Remittance receipts by the group move countercyclically with aggregate output.
  - Specifically, real remittances lag real GDP with a phase difference of one year: remittances tend to increase following a cyclical drop in home-country output, peaking within one year after a trough in output.

### Cross-correlation results (selected coefficients from Table 2)
- Whole sample:
  - Rrem(t-1): 0.3032
  - Rrem(t): -0.2696
  - Rrem(t+1): -0.3639*  (Countercyclical: Remittances lag output)
- Country-specific (Rrem at t-1, t, t+1; nature of co-movement) — selected entries:
  - DZA (LMI/MENA):
    - Rrem(t-1): -0.1346
    - Rrem(t): -0.2447
    - Rrem(t+1): -0.0739
    - Nature: Acyclical
  - BGD (LI/SA):
    - Rrem(t-1): 0.0536
    - Rrem(t): -0.4145*
    - Rrem(t+1): -0.1329
    - Nature: Countercyclical and synchronous
  - CIV (LI/SSA):
    - Rrem(t-1): 0.2482
    - Rrem(t): 0.0767
    - Rrem(t+1): -0.0885
    - Nature: Acyclical
  - DOM (LMI/LAC):
    - Rrem(t-1): 0.2497
    - Rrem(t): 0.2289
    - Rrem(t+1): 0.1109
    - Nature: Acyclical
  - IND (LI/SA):
    - Rrem(t-1): 0.3747
    - Rrem(t): -0.0143
    - Rrem(t+1): -0.3798*
    - Nature: Countercyclical: Remittances lag output
  - JAM (LMI/LAC):
    - Rrem(t-1): -0.0630
    - Rrem(t): -0.1812
    - Rrem(t+1): -0.0846
    - Nature: Acyclical
  - JOR (LMI/MENA):
    - Rrem(t-1): 0.3689
    - Rrem(t): 0.8704*
    - Rrem(t+1): 0.6472
    - Nature: Procyclical and synchronous
  - LSO (LI/SSA):
    - Rrem(t-1): 0.2193
    - Rrem(t): -0.0105
    - Rrem(t+1): -0.0434
    - Nature: Acyclical
  - MAR (LMI/MENA):
    - Rrem(t-1): -0.1217
    - Rrem(t): 0.2167
    - Rrem(t+1): 0.3832*
    - Nature: Procyclical: Remittances lag output
  - PAK (LI/SA):
    - Rrem(t-1): -0.0574
    - Rrem(t): -0.1539
    - Rrem(t+1): [table continues beyond provided excerpt]
- Note: In Table 2, the largest absolute-value correlation for each country is shown in bold and statistically significant coefficients at the 95 percent level are marked with an asterisk.

### Implications from results
- Group-level countercyclicality (remittances lagging output by one year) suggests remittances can act as a partial buffer to home-country output declines by rising after downturns.
- Country-level heterogeneity is pronounced:
  - Some countries exhibit countercyclical remittances that lag output (e.g., IND, whole sample), others exhibit procyclical and synchronous behavior (e.g., JOR), and many appear acyclical.
- Country-specific analysis is crucial because aggregate/group evidence can mask important differences in remittance-output dynamics across countries.

### Summary of core findings on remittance cyclicality
- Aggregate result: Remittances received by 12 countries in the sample are countercyclical with their aggregate output and respond to drops in output with a lag of one year.
- Heterogeneity: Individual-country results vary — some countries exhibit countercyclical remittances, some procyclical, and some acyclical behavior.
- Timing differences:
  - Bangladesh: remittances increase in the same year as an output drop (synchronous countercyclical response).
  - India: remittances increase with a lag of one year after output drops (lagged countercyclical response).
  - Jordan: remittances move procyclically and synchronously with Jordanian real GDP.
  - Morocco: remittances move procyclically but lag output by one year.

### Country-specific patterns and interpretations
- Countercyclical (evidence varies in strength/statistical significance):
  - Algeria, Jamaica, Lesotho, Pakistan, Turkey — appear countercyclical but degree not strong enough for confident classification using annual-data correlations.
- Procyclical (or apparently procyclical but significance varies):
  - Jordan and Morocco — migrants increase transfers during good times at home (implying stronger investment motive/higher risk aversion). Jordan shows strong and quick procyclical response.
  - Dominican Republic, Ivory Coast, Senegal — seemingly procyclical relationships fail statistical significance tests and thus are classified as acyclical in this analysis.
- Acyclical:
  - Many countries' output and remittance cycles are uncorrelated. Lesotho is highlighted: despite almost 70 percent (1990-2003 average) ratio of remittances to GDP, remittances are acyclical, reflecting volatility driven by employment changes of Basotho miners in South Africa rather than remitter behavior.

### Mechanisms and plausible explanations
- Consumption smoothing motive: Migrants from Bangladesh and India increase transfers during home-country downturns consistent with consumption smoothing.
- Investment motive / weakened confidence motive: Migrants from Jordan and Morocco increase transfers during home-country expansions; during downturns they may retain savings in host country due to declining returns or reduced confidence in home-country financial systems.
- Reverse causality possibility: For countries heavily dependent on remittances (e.g., Jordan with an average share close to 20 percent over 1990-2003), output drops may be caused by drops in remittances themselves (example: 1990-91 conflict in the Middle East reduced remittances to Jordan via job losses in Kuwait and Saudi Arabia).

### Timing and frequency issues (annual vs quarterly data)
- Annual-data limitations: Results may improve with higher frequency data.
- Turkey case:
  - Annual data indicates no synchronous/asynchronous correlation between real GDP and remittances.
  - Quarterly analysis (Turkish output and remittances from Turkish workers in Germany) shows:
    - Over 1987:1-2003:3 remittances from Germany are procyclical with Turkish output and follow the output cycle by a lag of one quarter.
    - Over 1987:1-1994:2 sub-period, remittances were countercyclical, following Turkish output with a lag of one or two quarters in the opposite direction.
  - Table 3 cross-correlations (i = -4,...,4) excerpted results:
    - 1987:1-2003:3 row (i = -4 to 4): -0.0516, 0.0161, 0.1940, 0.2838, 0.3566*, 0.3683*, 0.2766, 0.1494, 0.0830
    - 1987:1-1994:2 row (i = -4 to 4): 0.1761, 0.2345, 0.2700, 0.2055, -0.2990, -0.4128*, -0.4168*, -0.3599, -0.2439
    - Interpretation: largest contemporaneous and lag correlations for 1987:1-2003:3 are 0.3683 and 0.3566; for 1987:1-1994:2 procyclical signs reverse to negative in the later i positions indicating countercyclicality in the sub-period.

### Empirical caveats and interpretation guidance
- Cross-country averages can be misleading because they conceal substantial heterogeneity in individual-country remittance behavior.
- High remittance-to-GDP ratios do not automatically imply procyclicality (Lesotho example).
- Structural breaks and major crises (e.g., Turkey 1994 crisis, Turkey post-2000 crises, Jordan 1990-91 conflict) can change remittance cyclicality patterns, potentially switching countercyclicality to procyclicality.
- Exogenous dynamics in host-country labor demand and volatility in migrant employment can drive remittance volatility independent of home-country business cycles.

### Policy implications and recommendations
- Individual-country characteristics matter for policy design related to remittance flows.
- Caution is required when using potential remittance receipts as collateral for external funding because procyclicality or countercyclicality have opposite implications for crisis-coping capacity.
- Policymakers should account for:
  - Whether remittances are countercyclical or procyclical for their country.
  - The timing (synchronous vs lagged) of remittance responses to output changes.
  - Host-country employment dynamics and potential for exogenous shocks to remittance flows.
- Higher-frequency data (quarterly) is recommended to more accurately identify remittance-output comovements and detect structural changes over time.

*Source: _wp0652 - References (PDF) and IMF Working Paper (excerpted content provided).*

### References .......................................................................................... 18

### References

### I. Key findings on remittance scale and growth
- Total remittance receipts by developing countries reached 116 billion dollars in 2003, representing more than 1.5 percent of their total GDPs as a group.
- Remittance receipts grew steadily after 1990 with a slight decline in 1998.
- Share of remittance receipts (1990-2003 sample context):
  - Low-income countries: 32 percent of total remittance receipts of developing countries.
  - Middle-income countries: 68 percent.
- Average annual growth rates of remittance receipts since 1990:
  - Low income countries: 12.3 percent.
  - Middle income countries: 9.9 percent.
- Average annual growth rates of remittance receipts since 1999:
  - Low income countries: 15.9 percent.
  - Middle income countries: 10.8 percent.

### II. Importance and characteristics of remittances
- Remittances have become an increasingly important channel for meeting external financing needs of developing countries and are one of the largest sources of such financing (Ratha, 2003; Spatafora, 2005).
- Remittances are generally less volatile and hence more dependable than private capital flows and foreign direct investment (FDI) (Ratha, 2003; Buch and Kuckulenz, 2004).
- As unilateral transfers, remittances do not create future liabilities such as debt servicing or profit transfers.

### III. Cyclicality, methodology, and main empirical result
- Cycles are defined as deviations of real variables from their respective trends following Lucas (1977) and Kydland and Prescott (1990).
- Analysis focuses on co-movements between deviations from trend of real remittances and deviations from trend of real GDP (i.e., cyclical components), rather than on long-run correlations from regression coefficients.
- Main empirical finding for the sample of 12 low-income (LI) and lower-middle income (LMI) countries:
  - For the group as a whole, remittances are countercyclical and lead the aggregate GDP cycle by one period.
  - At the individual-country level, remittances can be procyclical or acyclical for some countries; panel results may conceal important country-specific characteristics.

### IV. Mechanisms, caveats, and transmission risks
- Countercyclicality interpretation: migrant savings remitted to home countries tend to increase after periods of stagnation/crisis at home and decrease after periods of growth/boom for the group aggregate.
- Decision to remit is complex and driven by multiple factors beyond altruistic financing of current consumption (see Russell, 1986); different drivers may imply procyclical, countercyclical, or acyclical behavior across countries.
- Remittances should also respond to the state of economic activity in host countries; synchronized cycles between home and host economies can limit the ability of migrants to support families during downturns at home (Sayan, 2004).
- Remittance flows can transmit contractions in host economies to recipient countries through synchronized reductions in amounts remitted (Sayan and Tekin-Koru, 2005).
- Example of vulnerability: the 1990-91 Middle East conflict had disastrous consequences for economies receiving large amounts of remittances from Kuwait and other countries in the region (Wahba, 1991).

### V. Selected 2001-level ratios highlighting importance of remittances
- GDP share of remittances as of 2001:
  - Low-income countries: 1.9 percent.
  - Lower-middle-income countries: 1.4 percent.
  - Upper-middle-income countries: 0.8 percent.
- Ratios of remittances to imports in 2001:
  - Low-income countries: 6.2 percent.
  - Lower-middle-income countries: 5.1 percent.
  - Upper-middle-income countries: 2.7 percent.

*Source: _wp0652 - References (PDF)*

### 213.5 and 666.1 percent for LI, 43.7 and 44.9 percent for LMI, and 21.7 and 20.2 percent for

### _wp0652 - 213.5 and 666.1 percent for LI, 43.7 and 44.9 percent for LMI, and 21.7 and 20.2 percent for

### Remittances: trend, volatility, and countercyclicality
- Remittances grew to stand out among other sources of external financing by 2001 in UMI countries (Ratha, 2003).
- Remittances are often observed to be generally less volatile than private capital flows that move procyclically with output in recipient countries.
- Evidence indicates a negative relationship between remittances and income for different countries and cross-country evidence that remittances would reduce the size of worst drops in GDP experienced during severe economic crises.
- Sharp increases in remittances after crises are documented for Indonesia (1997), Ecuador (1999) and Argentina (2001) (Spatafora, 2005).

### Microeconomic motivations and behavioral complexity
- Migrant workers plausibly increase transfers to family members during domestic downturns to help compensate for lost family income due to unemployment or crisis-induced reasons.
- Remitting behavior is multifaceted and affected by many variables beyond altruism, including interest rate differentials and exchange rates, which can produce procyclicality, countercyclicality, or acyclicality depending on country-specific conditions.
- Given differing implications of countercyclicality versus procyclicality for stability and for using remittances as collateral, cyclical nature should be investigated at the country level rather than relying solely on group evidence.

### Data, sample, and detrending methodology
- Annual data used cover 1976–2003.
- Sample: 12 countries (six LI and six LMI):
  - LI: Bangladesh (BGD), India (IND), Côte d’Ivoire (CIV), Lesotho (LSO), Pakistan (PAK), Senegal (SEN).
  - LMI: Algeria (DZA), Dominican Republic (DOM), Jamaica (JAM), Jordan (JOR), Morocco (MAR), Turkey (TUR).
- In the sample, the combined share of the 12 countries in total remittance receipts of all LI and LMI countries was 86.2 percent in 1976 and 39.8 percent in 2003 (27.5 percent for the six LI countries and 12.3 percent for the remaining six).
- Real remittances series were converted from nominal U.S. dollar terms to real terms using the U.S. GDP deflator (2000=100). Real GDP measured in constant US dollars at 2000 prices (World Development Indicators).
- Trends were removed by fitting polynomials of degree k to each series:
  - GDP trend polynomial form: y_t = α0 + α1 t + α2 t^2 + ... + α_k t^k
  - Remittances trend polynomial form: r_t = β0 + β1 t + β2 t^2 + ... + β_k t^k
- The trend polynomials were chosen so that estimated coefficients were statistically significant and detrended series passed stationarity checks (ADF).
- Aggregate (group) series constructed by weighting country outputs and remittances by average shares in total real GDP and total remittances of all LI and LMI countries.

### Estimated trends (group of 12)
- Estimated trend for aggregate real GDP (weighted total) — third-order polynomial (T_t y):
  - Constant: 4.916883 (2341.189)
  - t: 0.006593 (9.610)
  - t^2: 0.000189 (3.172)
  - t^3: -4.52E-06 (-3.115)
  - R^2: 0.998 [Adj.R^2 0.998]
  - F-statistic: 4626.6
- Estimated trend for aggregate real remittances (weighted) — fifth-order polynomial (T_t r):
  - Constant: 2.287111 (140.928)
  - t: 0.125088 (9.681)
  - t^2: -0.022463 (-7.234)
  - t^3: 0.001679 (5.644)
  - t^4: -5.43E-05 (-4.435)
  - t^5: 6.33E-07 (3.511)
  - R^2: 0.942 [Adj.R^2 0.929]
  - F-statistic: 72.1

### Identification of cyclical comovements
- After detrending, cyclical components (stationary with zero mean) were used to compute contemporaneous and asynchronous cross correlations between real output and real remittances.
- Asynchronous correlations considered remittances at t-1, t, and t+1 relative to output at t to detect leads/lags and phase shifts.

### Main results (summary)
- 1) Group-level behavior (whole sample, 1976–2003):
  - Remittance receipts by the group move countercyclically with aggregate output.
  - Specifically, real remittances lag real GDP with a phase difference of one year: remittances tend to increase following a cyclical drop in home-country output, peaking within one year after a trough in output.

### Cross-correlation results (selected coefficients from Table 2)
- Whole sample:
  - Rrem(t-1): 0.3032
  - Rrem(t): -0.2696
  - Rrem(t+1): -0.3639*  (Countercyc.: Remittances lag output)
- Country-specific entries (Rrem at t-1, t, t+1; nature of co-movement):
  - DZA (LMI/MENA):
    - Rrem(t-1): -0.1346
    - Rrem(t): -0.2447
    - Rrem(t+1): -0.0739
    - Nature: Acyclical
  - BGD (LI/SA):
    - Rrem(t-1): 0.0536
    - Rrem(t): -0.4145*
    - Rrem(t+1): -0.1329
    - Nature: Countercyclical and synchronous
  - CIV (LI/SSA):
    - Rrem(t-1): 0.2482
    - Rrem(t): 0.0767
    - Rrem(t+1): -0.0885
    - Nature: Acyclical
  - DOM (LMI/LAC):
    - Rrem(t-1): 0.2497
    - Rrem(t): 0.2289
    - Rrem(t+1): 0.1109
    - Nature: Acyclical
  - IND (LI/SA):
    - Rrem(t-1): 0.3747
    - Rrem(t): -0.0143
    - Rrem(t+1): -0.3798*
    - Nature: Countercyclical: Remittances lag output
  - JAM (LMI/LAC):
    - Rrem(t-1): -0.0630
    - Rrem(t): -0.1812
    - Rrem(t+1): -0.0846
    - Nature: Acyclical
  - JOR (LMI/MENA):
    - Rrem(t-1): 0.3689
    - Rrem(t): 0.8704*
    - Rrem(t+1): 0.6472
    - Nature: Procyclical and synchronous
  - LSO (LI/SSA):
    - Rrem(t-1): 0.2193
    - Rrem(t): -0.0105
    - Rrem(t+1): -0.0434
    - Nature: Acyclical
  - MAR (LMI/MENA):
    - Rrem(t-1): -0.1217
    - Rrem(t): 0.2167
    - Rrem(t+1): 0.3832*
    - Nature: Procyclical: Remittances lag output
  - PAK (LI/SA):
    - Rrem(t-1): -0.0574
    - Rrem(t): -0.1539
    - Rrem(t+1): [table continues beyond provided excerpt]

- Note: In Table 2, the largest absolute-value correlation for each country is shown in bold and statistically significant coefficients at the 95 percent level are marked with an asterisk.

### Implications from results
- Group-level countercyclicality (remittances lagging output by one year) suggests remittances can act as a partial buffer to home-country output declines by rising after downturns.
- Country-level heterogeneity is pronounced:
  - Some countries exhibit countercyclical remittances that lag output (e.g., IND, whole sample), others exhibit procyclical and synchronous behavior (e.g., JOR), and many appear acyclical.
- Country-specific analysis is crucial because aggregate/group evidence can mask important differences in remittance-output dynamics across countries.

*Source: IMF Working Paper (excerpted content provided).*

### 0.1499                           Acyclical

### _wp0652 - 0.1499                           Acyclical

### Summary of core findings on remittance cyclicality
- Aggregate result: Remittances received by 12 countries in the sample are countercyclical with their aggregate output and respond to drops in output with a lag of one year.
- Heterogeneity: Individual-country results vary — some countries exhibit countercyclical remittances, some procyclical, and some acyclical behavior.
- Timing differences:
  - Bangladesh: remittances increase in the same year as an output drop (synchronous countercyclical response).
  - India: remittances increase with a lag of one year after output drops (lagged countercyclical response).
  - Jordan: remittances move procyclically and synchronously with Jordanian real GDP.
  - Morocco: remittances move procyclically but lag output by one year.

### Country-specific patterns and interpretations
- Countercyclical (evidence varies in strength/statistical significance):
  - Algeria, Jamaica, Lesotho, Pakistan, Turkey — appear countercyclical but degree not strong enough for confident classification using annual-data correlations.
- Procyclical (or apparently procyclical but significance varies):
  - Jordan and Morocco — migrants increase transfers during good times at home (implying stronger investment motive/higher risk aversion). Jordan shows strong and quick procyclical response.
  - Dominican Republic, Ivory Coast, Senegal — seemingly procyclical relationships fail statistical significance tests and thus are classified as acyclical in this analysis.
- Acyclical:
  - Many countries' output and remittance cycles are uncorrelated. Lesotho is highlighted: despite almost 70 percent (1990-2003 average) ratio of remittances to GDP, remittances are acyclical, reflecting volatility driven by employment changes of Basotho miners in South Africa rather than remitter behavior.

### Mechanisms and plausible explanations
- Consumption smoothing motive: Migrants from Bangladesh and India increase transfers during home-country downturns consistent with consumption smoothing.
- Investment motive / weakened confidence motive: Migrants from Jordan and Morocco increase transfers during home-country expansions; during downturns they may retain savings in host country due to declining returns or reduced confidence in home-country financial systems.
- Reverse causality possibility: For countries heavily dependent on remittances (e.g., Jordan with an average share close to 20 percent over 1990-2003), output drops may be caused by drops in remittances themselves (example: 1990-91 conflict in the Middle East reduced remittances to Jordan via job losses in Kuwait and Saudi Arabia).

### Timing and frequency issues (annual vs quarterly data)
- Annual-data limitations: Results may improve with higher frequency data.
- Turkey case:
  - Annual data here indicates no synchronous/asynchronous correlation between real GDP and remittances.
  - Quarterly analysis (Turkish output and remittances from Turkish workers in Germany) shows:
    - Over 1987:1-2003:3 remittances from Germany are procyclical with Turkish output and follow the output cycle by a lag of one quarter.
    - Over 1987:1-1994:2 sub-period, remittances were countercyclical, following Turkish output with a lag of one or two quarters in the opposite direction.
  - Table 3 cross-correlations (i = -4,...,4) excerpted results:
    - 1987:1-2003:3 row (i = -4 to 4): -0.0516, 0.0161, 0.1940, 0.2838, 0.3566*, 0.3683*, 0.2766, 0.1494, 0.0830
    - 1987:1-1994:2 row (i = -4 to 4): 0.1761, 0.2345, 0.2700, 0.2055, -0.2990, -0.4128*, -0.4168*, -0.3599, -0.2439
    - Interpretation: largest contemporaneous and lag correlations for 1987:1-2003:3 are 0.3683 and 0.3566; for 1987:1-1994:2 procyclical signs reverse to negative in the later i positions indicating countercyclicality in the sub-period.

### Empirical caveats and interpretation guidance
- Cross-country averages can be misleading because they conceal substantial heterogeneity in individual-country remittance behavior.
- High remittance-to-GDP ratios do not automatically imply procyclicality (Lesotho example).
- Structural breaks and major crises (e.g., Turkey 1994 crisis, Turkey post-2000 crises, Jordan 1990-91 conflict) can change remittance cyclicality patterns, potentially switching countercyclicality to procyclicality.
- Exogenous dynamics in host-country labor demand and volatility in migrant employment can drive remittance volatility independent of home-country business cycles.

### Policy implications and recommendations
- Individual-country characteristics matter for policy design related to remittance flows.
- Caution is required when using potential remittance receipts as collateral for external funding because procyclicality or countercyclicality have opposite implications for crisis-coping capacity.
- Policymakers should account for:
  - Whether remittances are countercyclical or procyclical for their country.
  - The timing (synchronous vs lagged) of remittance responses to output changes.
  - Host-country employment dynamics and potential for exogenous shocks to remittance flows.
- Higher-frequency data (quarterly) is recommended to more accurately identify remittance-output comovements and detect structural changes over time.

*Source: Author’s calculations based on the World Bank data; content extracted from _wp0652 - 0.1499 Acyclical.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0652.pdf_