## _wp06130

## Source details

**Canonical URL:** [_wp06130](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06130.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06130.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06130.pdf.json)

---

### Introduction and context
- During 2000-04, workers’ remittances amounted to:
  - Egypt: 3.5 percent of GDP
  - Jordan: 20 percent of GDP
  - Morocco: 8 percent of GDP
  - Pakistan: 4 percent of GDP
  - Tunisia: 5 percent of GDP
- Trade deficits (2000-04 averages, percent of GDP):
  - Egypt: -7.5
  - Jordan: -23.8
  - Morocco: -10.2
  - Pakistan: -1.5
  - Tunisia: -10.1
- Remittances relative to exports of goods and services:
  - Egypt: more than 15 percent of exports of goods and services
  - Jordan: more than 40 percent of exports of goods and services
  - Morocco: about 25 percent of exports of goods and services
  - Pakistan: about 22 percent of exports of goods and services
  - Tunisia: about 11 percent of exports of goods and services
- Additional balance-of-payments context (2000-04 averages, selected items, percent of GDP):
  - Current Account Balance: Egypt 1.6; Jordan 3.7; Morocco 2.6; Pakistan 2.2; Tunisia -3.3
  - Exports of goods: Egypt 10.1; Jordan 29.3; Morocco 20.8; Pakistan 13.6; Tunisia 32.6
  - Imports of goods: Egypt -17.6; Jordan -53.1; Morocco -31.1; Pakistan -15.1; Tunisia -42.7
  - Services, net: Egypt 4.2; Jordan -1.3; Morocco 5.5; Pakistan -1.2; Tunisia 6.4
  - Current Transfers, net: Egypt 4.8; Jordan 27.2; Morocco 9.3; Pakistan 7.7; Tunisia 5.1
  - Remittances, credit: Egypt 3.5; Jordan 19.9; Morocco 8.2; Pakistan 3.6; Tunisia 4.8
  - Capital and Financial Account, net: Egypt -2.3; Jordan -5.8; Morocco -2.4; Pakistan -2.9; Tunisia 3.4
  - Direct Investment, net: Egypt 0.9; Jordan 4.3; Morocco 3.4; Pakistan 0.8; Tunisia 2.8
  - Reserve Assets, net: Egypt 0.2; Jordan -5.7; Morocco -3.9; Pakistan -2.2; Tunisia -1.3
  - Errors and Omissions, net: Egypt 0.7; Jordan 2.1; Morocco -0.2; Pakistan 0.7; Tunisia -0.1

### Motives to remit and implications for stability
- Altruism / solidarity motives
  - Defined as willingness of workers abroad to provide financial assistance in situations of “hardship.”
  - If important, remittances are expected to be relatively stable in developing countries because recipient families continue to depend on them.
  - Migration pattern changes may alter scope for altruism, but new emigration waves can offset this.
  - Altruistic flows may also reflect self-interested considerations (e.g., leaving someone to manage residential investments), reinforcing potential stability.
- Attachment to homeland
  - Remittances for building assets at home (real estate, small businesses) tend to be persistent.
  - Improved settlement opportunities abroad could reduce attachment, but such changes are likely slow.
- Portfolio diversification motives
  - Would make remittances sensitive to rates of return and potentially volatile, like other capital flows.
- Exchange rate effects
  - Ambiguous net effect: depreciation can both reduce needed transfer amounts (cheaper local goods) and increase purchasing power of expatriates (wealth effect), possibly increasing remittances.
  - Exchange rate also captures unit-of-account differences; therefore it is retained in the empirical specification to control for such changes.
- National policies toward diaspora
  - Institutional features (for example, ministries for nationals abroad) can streamline transactions and contribute to stability.

### Empirical framework
- Long-run relation estimated:
  - Remittances = α Agricultural GDP + β Exchange rate + γ Trend + ε
- Variables and measurement:
  - Remittances: logarithm of yearly dollar value of remittance inflows (for Tunisia remittances in local currency are used)
  - Agricultural GDP: logarithm of real agricultural GDP in local currency (proxy for “hardship”; a fall interpreted as an increase in hardship)
  - Exchange rate: logarithm of exchange rate (US dollar per national currency)
  - Trend: linear trend to capture other motives to remit
  - ε: stationary error term
- Data coverage:
  - Egypt: 1977-2002
  - Jordan: 1975-2002
  - Morocco: 1975-2002
  - Pakistan: 1976-2002
  - Tunisia: 1976-2002
- Estimation approach:
  - Cointegration techniques applied to logarithms of original variables to estimate long-run relations without differencing and allowing for endogenous regressors.
  - Appendix reports stationarity, cointegration rank tests, and VECM diagnostics for each country.

### Key empirical findings
- Agricultural GDP coefficient (α)
  - The coefficient on “Agricultural GDP” (α) is negative and statistically significant for all five countries (Egypt, Jordan, Morocco, Pakistan, Tunisia).
  - Interpretation: In the long run, when agricultural GDP falls (interpreted as increased hardship), remittances tend to increase, consistent with altruistic motives.
  - Exclusion tests indicate “Agricultural GDP” significantly enters the long-run relation for each country.
  - Magnitude remarks: α is notably larger for Egypt and Pakistan than for Jordan, Morocco, and Tunisia; this could reflect omitted variables or instability due to sample size limitations.
- Trend and other motives
  - The linear trend capturing other motives to remit was significant in the estimates, implying other motives besides altruism also matter.
  - The relative importance of altruistic motives versus other motives could not be precisely quantified within this framework.
- Exchange rate coefficient (β)
  - The sign and magnitude of the coefficient on the Exchange rate are more difficult to interpret because of offsetting theoretical effects; exclusion tests for exchange rate vary by country.

### Conclusions and policy implications
- Altruism has been an important determinant of remittances to Egypt, Jordan, Morocco, Pakistan, and Tunisia over the sample period.
- Because altruistic/solidarity motives partly drive remittances, a sudden drop or reversal of these flows is considered unlikely in the foreseeable future; many recipients are likely to continue depending on them.
- Policy-relevant implications:
  - Remittances are likely to continue mitigating external vulnerabilities for the countries studied.
  - Remittances influence liquidity conditions and monetary policy considerations, especially where remittances are large relative to exports or where exchange rate regimes are managed.
- Limitations and suggestions for further research:
  - Using a linear trend to capture all non-altruistic motives is a simplifying assumption; explicit inclusion of other determinants (wages abroad, nonagricultural GDP, interest differentials) may refine understanding.
  - Further work needed to quantify the relative importance of altruistic motives versus other motives.

### Econometric diagnostics and country summaries (selected results)
- Stationarity (Elliott-Rothenberg-Stock DF-GLS test)
  - For Egypt, Jordan, Morocco, Pakistan, and Tunisia, variables in levels were generally found to have a unit root and their first differences were stationary (integrated of order one, I(1)), with country- and variable-specific test statistics reported in the appendix.
- Cointegration tests (Johansen Trace and Maximum Eigenvalue) — selected outcomes:
  - Egypt (Tables A2–A3): Trace "None" Prob. 0.079; Max-Eigen "None" Prob. 0.036.
  - Jordan (Tables A6–A7): Trace "None" Prob. 0.002; Max-Eigen "None" Prob. 0.006.
  - Morocco (Tables A10–A11): Trace "None" Prob. 0.008; Max-Eigen "None" Prob. 0.040.
  - Pakistan (Tables A14–A15): Trace "None" Prob. 0.11; Max-Eigen "None" Prob. 0.046.
  - Tunisia (Tables A18–A19, local currency): Trace "None" Prob. 0.028; Max-Eigen "None" Prob. 0.125.
- VECM diagnostics — selected statistics and interpretations:
  - Egypt (Table A4): Multivariate LM test values by lag include 6.03, 4.88, 1.98, 3.14, 2.45, 6.28 with p-values [0.20][0.30][0.74][0.53][0.65][0.18]; Univariate exclusion tests for Remittances and Agricultural GDP: 16.46 [0.00] and 7.79 [0.00].
  - Jordan (Table A8): Multivariate LM test values by lag include 4.31, 4.80, 3.86, 1.43, 1.51, 1.43 with p-values [0.37][0.31][0.42][0.84][0.82][0.84]; Univariate exclusion tests: 9.80 [0.00] and 6.8 [0.01].
  - Morocco (Table A12): Multivariate LM test statistics (lags 1–6): 13.62, 10.77, 7.50, 10.69, 9.01, 1.02 with p-values [ 0.14][0.29][ 0.58][ 0.30] [0.44][ 0.999]; Univariate weak exogeneity statistics (Remittances, Agricultural GDP, Exchange rate): 4.81 [0.03], 6.44 [0.01], 1.31 [0.25].
  - Pakistan (Table A16): Multivariate LM test statistics (lags 1–6): 3.15, 6.30, 11.24, 5.43, 10.61, 10.88 with p-values 0.96 0.71 0.26 0.80 0.30 0.28; Univariate exclusion test statistics (Remittances, Agricultural GDP, Exchange rates): 10.16 [0.00], 17.44 [0.00], 1.20 [0.27].
  - Tunisia (Table A20): Multivariate LM test statistics (lags 1–6): 2.97, 2.73, 1.70, 0.38, 0.92, 2.75 with p-values [ 0.56][ 0.60][ 0.79][ 0.98][ 0.92][ 0.60]; Univariate exclusion test (Remittances, Agricultural GDP): 5.07 [0.02], 3.27 [0.07].
- Country-specific diagnostic notes:
  - Egypt: Exclusion tests indicate all variables in cointegration relation are significant at 5 percent; model could explain remittance flows.
  - Jordan: Johansen tests indicate cointegration at 5 percent; exclusion tests show variables significant at 1 percent.
  - Morocco: Evidence of cointegration; VECM residuals not autocorrelated at most lags; variables significant in long-run relation.
  - Pakistan: Mixed cointegration evidence (Trace none Prob. 0.11; Max-Eigen none Prob. 0.046); exchange rate may not enter long-run relation; models without exchange rate raised weak-exogeneity concerns.
  - Tunisia: Dollar-denominated analysis produced uncertainty; local-currency analysis suggested cointegration (Trace none Prob. 0.028) and model adequacy.

*Source: IMF working paper chapter "2. Long-Run Determinants of Remittances" (document _wp06130)*

### References..............................................................................................................

### References

### Appendix overview
- Appendix..................................................................................................................................15
- A. Egypt .......................................................................................................................15
- B. Jordan ......................................................................................................................18
- C. Morocco ..................................................................................................................20
- D. Pakistan ...................................................................................................................23
- E. Tunisia .....................................................................................................................25

### Tables
- 1.  Balance of Payments, 2000-04 Averages .............................................................................4

*Source: _wp06130 - References..............................................................................................................*

### 2. Long-Run Determinants of Remittances ...............................................................................9

### 2. Long-Run Determinants of Remittances

### Introduction and context
- During 2000-04, workers’ remittances amounted to:
  - Egypt: 3.5 percent of GDP
  - Jordan: 20 percent of GDP
  - Morocco: 8 percent of GDP
  - Pakistan: 4 percent of GDP
  - Tunisia: 5 percent of GDP
- Trade deficits (2000-04 averages, percent of GDP):
  - Egypt: -7.5
  - Jordan: -23.8
  - Morocco: -10.2
  - Pakistan: -1.5
  - Tunisia: -10.1
- Remittances compared to exports of goods and services:
  - Egypt: more than 15 percent of exports of goods and services
  - Jordan: more than 40 percent of exports of goods and services
  - Morocco: about 25 percent of exports of goods and services
  - Pakistan: about 22 percent of exports of goods and services
  - Tunisia: about 11 percent of exports of goods and services
- Additional balance-of-payments context (2000-04 averages, selected items, percent of GDP):
  - Current Account Balance: Egypt 1.6; Jordan 3.7; Morocco 2.6; Pakistan 2.2; Tunisia -3.3
  - Exports of goods: Egypt 10.1; Jordan 29.3; Morocco 20.8; Pakistan 13.6; Tunisia 32.6
  - Imports of goods: Egypt -17.6; Jordan -53.1; Morocco -31.1; Pakistan -15.1; Tunisia -42.7
  - Services, net: Egypt 4.2; Jordan -1.3; Morocco 5.5; Pakistan -1.2; Tunisia 6.4
  - Current Transfers, net: Egypt 4.8; Jordan 27.2; Morocco 9.3; Pakistan 7.7; Tunisia 5.1
  - Remittances, credit: Egypt 3.5; Jordan 19.9; Morocco 8.2; Pakistan 3.6; Tunisia 4.8
  - Capital and Financial Account, net: Egypt -2.3; Jordan -5.8; Morocco -2.4; Pakistan -2.9; Tunisia 3.4
  - Direct Investment, net: Egypt 0.9; Jordan 4.3; Morocco 3.4; Pakistan 0.8; Tunisia 2.8
  - Reserve Assets, net: Egypt 0.2; Jordan -5.7; Morocco -3.9; Pakistan -2.2; Tunisia -1.3
  - Errors and Omissions, net: Egypt 0.7; Jordan 2.1; Morocco -0.2; Pakistan 0.7; Tunisia -0.1

### Motives to remit and implications for stability
- Altruism / solidarity motives
  - Defined as willingness of workers abroad to provide financial assistance in situations of “hardship.”
  - If important, remittances are expected to be relatively stable in developing countries because recipient families continue to depend on them.
  - Migration pattern changes may alter scope for altruism, but new emigration waves can offset this.
  - Altruistic flows may also reflect self-interested considerations (e.g., leaving someone to manage residential investments), reinforcing potential stability.
- Attachment to homeland
  - Remittances for building assets at home (real estate, small businesses) tend to be persistent.
  - Improved settlement opportunities abroad could reduce attachment, but such changes are likely slow.
- Portfolio diversification motives
  - Would make remittances sensitive to rates of return and potentially volatile, like other capital flows.
- Exchange rate effects
  - Ambiguous net effect: depreciation can both reduce needed transfer amounts (cheaper local goods) and increase purchasing power of expatriates (wealth effect), possibly increasing remittances.
  - Exchange rate also captures unit-of-account differences; therefore it is retained in the empirical specification to control for such changes.
- National policies toward diaspora
  - Institutional features (for example, ministries for nationals abroad) can streamline transactions and contribute to stability.

### Empirical framework
- Long-run relation estimated:
  - Remittances = α Agricultural GDP + β Exchange rate + γ Trend + ε
  - Variables in the estimation:
    - Remittances: logarithm of yearly dollar value of remittance inflows (for Tunisia remittances in local currency are used; see appendix)
    - Agricultural GDP: logarithm of real agricultural GDP in local currency (proxy for “hardship”; a fall interpreted as an increase in hardship)
    - Exchange rate: logarithm of exchange rate (US dollar per national currency)
    - Trend: linear trend to capture other motives to remit
    - ε: stationary error term
- Data coverage:
  - Egypt: 1977-2002
  - Jordan: 1975-2002
  - Morocco: 1975-2002
  - Pakistan: 1976-2002
  - Tunisia: 1976-2002
- Estimation approach:
  - Cointegration techniques applied to logarithms of original variables to estimate long-run relations without differencing and allowing for endogenous regressors.
  - Appendix reports stationarity, cointegration rank tests, and VECM diagnostics for each country.

### Key empirical findings
- The coefficient on “Agricultural GDP” (α) is negative and statistically significant for all five countries studied (Egypt, Jordan, Morocco, Pakistan, Tunisia), suggesting:
  - In the long run, when agricultural GDP falls (interpreted as increased hardship), remittances tend to increase.
  - This negative long-run correlation is interpreted as evidence consistent with altruistic motives in remitters.
- Exclusion tests indicate “Agricultural GDP” significantly enters the long-run relation for each country (see appendix tables referenced in the chapter).
- The linear trend capturing other motives to remit was significant in the estimates, implying other motives besides altruism also matter.
- Remarks on magnitudes:
  - The magnitude of α is notably larger for Egypt and Pakistan than for the other three countries; this could indicate omitted variables or instability due to sample size limitations.
  - The sign and magnitude of the coefficient on the Exchange rate are more difficult to interpret because of offsetting theoretical effects.

### Conclusions and policy implications
- Evidence supports the view that altruism has been an important determinant of remittances to Egypt, Jordan, Morocco, Pakistan, and Tunisia over the sample period.
- Because altruistic/solidarity motives partly drive remittances, a sudden drop or reversal of these flows is considered unlikely in the foreseeable future; many recipients are likely to continue depending on them.
- Remittances are therefore likely to continue mitigating external vulnerabilities for the countries studied and to influence liquidity conditions and monetary policy considerations (especially where remittances are large relative to exports or where exchange rate regimes are managed).
- Limitations and areas for further research:
  - Use of a linear trend to capture all non-altruistic motives is a simplifying assumption; explicit inclusion of other determinants (wages abroad, nonagricultural GDP, interest differentials) may refine understanding.
  - The relative importance of altruistic motives versus other motives could not be precisely quantified within this framework; the significant trend term suggests other motives deserve further investigation.

*Source: IMF working paper chapter "2. Long-Run Determinants of Remittances"*

### REFERENCES

### _wp06130 - REFERENCES

### Major empirical findings on remittances (from referenced studies)
- Numerous empirical works examine motives, determinants, and impacts of remittances (examples in the reference list include Abel and Kotlikoff 1988; Adams 1991, 1998; Agarwal and Horowitz 2002; Bouhga-Hagbe 2004; Chami, Fullenkamp, and Jahjah 2005; Durand et al. 1996; Lucas and Stark 1985, 1988; Poirine 1997; Taylor 1992, 1999).
- Methodological references used in the econometric appendix include tests and techniques: Elliott-Rothenberg-Stock DF-GLS test (Elliott, Rothenberg, and Stock, 1996), Johansen (1991) cointegration tests, Engle and Granger (1987) error-correction representation, Granger and Newbold (1974), Johansen and Juselius (1992, 1994), Dickey and Rossana (1994), Kremers, Ericsson, and Dolado (1992), Mackinnon (1996), Phillips (1986), and unit-root/cointegration distribution references.

### Appendix — Econometric analysis: common features across countries
- Data source and sample periods:
  - Egypt: "The data are taken from the World Development Indicators database and cover the period 1977-2002."
  - Jordan: "The data are taken from the World Development Indicators database and cover the period 1975-2002."
  - Morocco: "The data are taken from the World Development Indicators database and cover the period 1975-2002."
- For Egypt, Jordan, and Morocco, all variables used in the analysis are reported as integrated of order one (I(1)) based on the Elliott-Rothenberg-Stock DF-GLS test.

### A. Egypt — econometric results and diagnostics
- Stationarity (Elliott-Rothenberg-Stock DF-GLS test) conclusions:
  - In levels: Remittances, Agricultural GDP, exchange rate — "Unit root".
  - In first difference: D(Remittances), D(Agricultural GDP), D(exchange rate) — "Stationary".
- VECM specifications and cointegration:
  - Initial VECM: four lags. Johansen (1991) tests suggested existence of a long-run relation (presented in equation (1) and Table 2).
  - Exclusion test on exchange rate: χ2 (1) = 2.22, (prob=0.14) → exchange rate may not significantly enter long-run relation.
  - Second VECM (without exchange rate): one lag included.
  - Johansen Trace test: existence of a long-run relation at 10 percent level of significance (Table A2).
  - Johansen Max-Eigen test: existence of a long-run relation at 5 percent level of significance (Table A3).
  - Exclusion tests indicate all variables included in the cointegration relation are significant at 5 percent level of significance.
  - Conclusion: "Therefore, the model presented in equation (1) could explain the flow of remittances to Egypt."
- Selected table values (Egypt):
  - Table A1: DF-GLS test statistics in levels and first differences (examples shown):
    - Remittances (levels): test statistic -1.97; critical values: 1 percent -2.66, 5 percent -1.96, 10 percent -1.61; Conclusion: Unit root.
    - D(Remittances): test statistic -4.82; 1 percent critical -3.77; Conclusion: Stationary.
  - Table A2 (Cointegration Rank Tests — Trace):
    - "None" eigenvalue 0.57; Trace Statistic 24.24; Critical Value 25.87; Prob. 0.079.
    - "At most 1" eigenvalue 0.15; Trace Statistic 3.86; Critical Value 12.52; Prob. 0.76.
  - Table A3 (Cointegration Rank Tests — Maximum Eigenvalue):
    - "None" eigenvalue 0.57; Max-Eigen Statistic 20.39; Critical Value 19.39; Prob. 0.036.
    - "At most 1" eigenvalue 0.15; Max-Eigen Statistic 3.86; Critical Value 12.52; Prob. 0.76.
  - Table A4 (Statistical Properties of the VECM):
    - Multivariate LM test values by lag: 6.03, 4.88, 1.98, 3.14, 2.45, 6.28 with p-values [0.20][0.30][0.74][0.53][0.65][0.18].
    - Multivariate Normality JB 4.93 [0.29]; Skewness 4.41 [0.11]; Excess kurtosis 0.51 [0.77].
    - Univariate tests — Weak exogeneity for Remittances and Agricultural GDP: 5.86 and 5.87 [0.02] [0.02].
    - Univariate exclusion tests for Remittances and Agricultural GDP: 16.46 [0.00] and 7.79 [0.00].
    - Univariate Normality JB for Remittances and Agricultural GDP: 2.60 and 2.32 [0.27][0.31].

### B. Jordan — econometric results and diagnostics
- Stationarity (Elliott-Rothenberg-Stock DF-GLS test) conclusions:
  - In levels: Remittances, Agricultural GDP, exchange rate — "Unit root".
  - In first difference: D(Remittances), D(Agricultural GDP), D(exchange rate) — "Stationary".
- VECM specifications and cointegration:
  - Initial VECM: four lags. Johansen tests suggested existence of a long-run relation (equation (1) and Table 2).
  - Exclusion test on exchange rate: χ2 (1) = 1.53, (prob=0.21) → exchange rate may not significantly enter long-run relation.
  - Second VECM (without exchange rate): four lags included.
  - Both Johansen Trace and Max-Eigen tests suggested existence of a long-run relation at 5 percent level of significance (Tables A6 and A7).
  - Exclusion tests suggest all variables included in the cointegration relation are significant at 1 percent level of significance.
  - Conclusion: "Therefore, the model presented in equation (1) could explain the flow of remittances to Jordan."
- Selected table values (Jordan):
  - Table A5: DF-GLS test statistics (examples):
    - Remittances (levels): test statistic -1.18; critical values: 1 percent -2.65, 5 percent -1.95, 10 percent -1.61; Conclusion: Unit root.
    - D(Remittances): test statistic -2.74; Conclusion: Stationary.
  - Table A6 (Cointegration Rank Tests — Trace):
    - "None" eigenvalue 0.65; Trace Statistic 36.36; Critical Value 25.87; Prob. 0.002.
    - "At most 1" eigenvalue 0.37; Trace Statistic 10.93; Critical Value 12.52; Prob. 0.091.
  - Table A7 (Cointegration Rank Tests — Max-Eigen):
    - "None" eigenvalue 0.65; Max-Eigen Statistic 25.43; Critical Value 19.39; Prob. 0.006.
    - "At most 1" eigenvalue 0.37; Max-Eigen Statistic 10.93; Critical Value 12.52; Prob. 0.091.
  - Table A8 (Statistical Properties of the VECM):
    - Multivariate LM test values by lag: 4.31, 4.80, 3.86, 1.43, 1.51, 1.43 with p-values [0.37][0.31][0.42][0.84][0.82][0.84].
    - Multivariate Normality JB 5.35 [0.25]; Skewness 0.56 [0.76]; Excess kurtosis 4.79 [0.09].
    - Univariate tests — Weak exogeneity for Remittances and Agricultural GDP: 8.54 and 9.64 [0.00] [0.00].
    - Univariate exclusion tests for Remittances and Agricultural GDP: 9.80 [0.00] and 6.8 [0.01].
    - Univariate Normality JB for Remittances and Agricultural GDP: 1.74 and 3.61 [0.42][0.16].

### C. Morocco — econometric results and diagnostics
- Stationarity:
  - All variables are integrated of order one (I(1)) by the DF-GLS test (Table A9).
- VECM specifications and cointegration:
  - Initial VECM: 3 lags. Both Johansen and Trace tests suggested existence of a long-run relation at 5 percent level of significance (Tables A10 and A11).
  - In the initial VECM only the first and third lags were significant; a second VECM was estimated including only the first and third lag.
  - Exclusion tests suggest all variables included in the cointegration relation are significant at 5 percent level of significance.
  - Conclusion: "Therefore, the model presented in equation (1) could explain the flow of remittances to Morocco."
- Data period noted: "The data are taken from the World Development Indicators database and cover the period 1975-2002."

*Source: _wp06130 - REFERENCES (IMF PDF)_

### Conclusion

### _wp06130 - Conclusion

### Stationarity and Unit Root Tests (Elliott-Rothenberg-Stock DF-GLS test)
- Test description and notes:
  - The test used is the one proposed by Elliott, Rothenberg, and Stock (1996). The null hypothesis is that the variable has a unit root.
  - The number of lags is selected using the Akaike information criterion. The maximum number of lags used is 6.
  - The critical values are from MacKinnon (1996).
  - For regression equations, the critical values are from MacKinnon (1996).

- Morocco (summary of test outcomes shown):
  - In levels:
    - Remittances (constant): 0-0.45-2.65-1.95-1.61 Unit root
    - Agricultural GDP (constant): 1-1.16-2.66-1.95-1.61 Unit root
    - exchange rate (constant): 1-0.96-2.66-1.95-1.61 Unit root
  - In first difference:
    - D(Remittances) (constant): 1-4.66-2.66-1.95-1.61 Stationary
    - D(Agricultural G D) (constant): 0-10.58-2.66-1.95-1.61 Stationary
    - D(exchange rate) (constant): 4-3.20-2.67-1.95-1.61 Stationary

- Pakistan (Table A13 summary):
  - In levels:
    - Remittances (constant): 0-1.52-2.66-1.95-1.61 Unit root
    - Agricultural GDP (constant): 4-0.73-2.67-1.95-1.61 Unit root
    - exchange rate (constant): 3-1.06-2.67-1.95-1.61 Unit root
  - In first difference:
    - D(Remittances) (constant): 0-2.02-2.66-1.95-1.61 Stationary
    - D(Agricultural G D) (constant): 0-6.27-2.66-1.95-1.61 Stationary
    - D(exchange rate) (constant): 6-2.88-2.69-1.96-1.61 Stationary
  - Data coverage note: The data are taken from the World Development Indicators database and cover the period 1976-2002.

- Tunisia (Table A17 summary, local currency analysis):
  - In levels:
    - Remittances (constant): 00.40-2.66-1.95-1.61 Unit root
    - Agricultural GDP (constant): 0-1.07-2.66-1.95-1.61 Unit root
  - In first difference:
    - D(Remittances) (constant): 0-4.21-2.66-1.95-1.61 Stationary
    - D(Agricultural GDP ) (constant): 0-5.14-2.66-1.95-1.61 Stationary
  - Data coverage note: The data are taken from the World Development Indicators database and cover the period 1976-2002.

### Cointegration Tests (Johansen Trace and Maximum Eigenvalue)
- Morocco (Tables A10 and A11):
  - Trace test (Table A10):
    - None: Eigenvalue 0.67, Trace Statistic 50.16, Critical Value 42.92, Prob. 0.008
    - At most 1: Eigenvalue 0.47, Trace Statistic 23.62, Critical Value 25.87, Prob. 0.093
    - At most 2: Eigenvalue 0.30, Trace Statistic 8.44, Critical Value 12.52, Prob. 0.218
  - Maximum eigenvalue test (Table A11):
    - None: Eigenvalue 0.67, Max-Eigen Statistic 26.55, Critical Value 25.82, Prob. 0.040
    - At most 1: Eigenvalue 0.47, Max-Eigen Statistic 15.18, Critical Value 19.39, Prob. 0.184
    - At most 2: Eigenvalue 0.30, Max-Eigen Statistic 8.44, Critical Value 12.52, Prob. 0.218

- Pakistan (Tables A14 and A15):
  - Trace test (Table A14):
    - None: Eigenvalue 0.65, Trace Statistic 39.32, Critical Value 42.92, Prob. 0.11
    - At most 1: Eigenvalue 0.29, Trace Statistic 13.21, Critical Value 25.87, Prob. 0.72
    - At most 2: Eigenvalue 0.17, Trace Statistic 4.64, Critical Value 12.52, Prob. 0.65
  - Maximum eigenvalue test (Table A15):
    - None: Eigenvalue 0.65, Max-Eigen Statistic 26.11, Critical Value 25.82, Prob. 0.046
    - At most 1: Eigenvalue 0.29, Max-Eigen Statistic 8.57, Critical Value 19.39, Prob. 0.77
    - At most 2: Eigenvalue 0.17, Max-Eigen Statistic 4.64, Critical Value 12.52, Prob. 0.64

- Tunisia (Tables A18 and A19):
  - Trace test (Table A18):
    - None: Eigenvalue 0.53, Trace Statistic 27.81, Critical Value 25.87, Prob. 0.028
    - At most 1: Eigenvalue 0.40, Trace Statistic 11.31, Critical Value 12.52, Prob. 0.079
  - Maximum eigenvalue test (Table A19):
    - None: Eigenvalue 0.53, Max-Eigen Statistic 16.50, Critical Value 19.39, Prob. 0.125
    - At most 1: Eigenvalue 0.40, Max-Eigen Statistic 11.31, Critical Value 12.52, Prob. 0.079

### Vector Error Correction Model (VECM) Statistical Properties and Diagnostic Tests
- Morocco (Table A12: Statistical Properties of the VECM):
  - Multivariate LM test statistics (lags 1–6): 13.62, 10.77, 7.50, 10.69, 9.01, 1.02
    - Corresponding p-values in brackets: [ 0.14][0.29][ 0.58][ 0.30] [0.44][ 0.999]
  - Multivariate Normality:
    - Normality JB 9.19 [0.16]
    - Skewness 2.68 [0.44]
    - Excess kurtosis 6.52 [0.09]
  - Univariate tests:
    - Weak exogeneity statistics (Remittances, Agricultural GDP, Exchange rate): 4.81, 6.44, 1.31 with p-values [ 0.03][ 0.01][ 0.25]
    - Exclusion test statistics (Remittances, Agricultural GDP, Exchange rate): 3.96, 4.21, 7.27 with p-values [0.05][0.04][0.01]
    - Normality JB (Remittances, Agricultural GDP, Exchange rate): 3.32, 2.56, 3.32 with p-values [0.19] [0.28][ 0.19]

- Pakistan (Table A16: Statistical Properties of the VECM):
  - Multivariate LM test statistics (lags 1–6): 3.15, 6.30, 11.24, 5.43, 10.61, 10.88
    - Corresponding p-values (formatted inline): 0.96 0.71 0.26 0.80 0.30 0.28
  - Multivariate Normality:
    - Normality JB 7.23 [0.30]
    - Skewness 1.57 [0.67]
    - Excess kurtosis 5.66 [0.13]
  - Univariate tests:
    - Weak exogeneity (Remittances, Agricultural GDP, Exchange rates): 4.02, 8.02, 1.51 with p-values [ 0.04][ 0.00][ 0.22]
    - Exclusion test (Remittances, Agricultural GDP, Exchange rates): 10.16, 17.44, 1.20 with p-values [0.00][ 0.00][ 0.27]
    - Normality JB (Remittances, Agricultural GDP, Exchange rates): 1.83, 2.83, 2.57 with p-values [0.40][0.24][0.28]

- Tunisia (Table A20: Statistical Properties of the VECM):
  - Multivariate LM test statistics (lags 1–6): 2.97, 2.73, 1.70, 0.38, 0.92, 2.75
    - Corresponding p-values in brackets: [ 0.56][ 0.60][ 0.79][ 0.98][ 0.92][ 0.60]
  - Multivariate Normality:
    - Normality JB 8.18 [0.09]
    - Skewness 0.06 [0.97]
    - Excess curtosis 8.12 [0.02]
  - Univariate tests:
    - Weak exogeneity (Remittances, Agricultural GDP): 3.00, 1.03 with p-values [0.08] [0.31]
    - Exclusion test (Remittances, Agricultural GDP): 5.07, 3.27 with p-values [0.02][0.07]
    - Normality JB (Remittances, Agricultural GDP): 4.43, 3.75 with p-values [0.11][0.15]

### Country-specific Conclusions and Model Interpretations
- Morocco:
  - Johansen trace and maximum-eigenvalue tests provide evidence of cointegration at conventional significance levels (Trace: Prob. 0.008 for none; Max-Eigen: Prob. 0.040 for none).
  - VECM diagnostics indicate residuals are not autocorrelated at most lags and multivariate normality is not rejected at conventional levels.
  - Exclusion and weak exogeneity tests identify significant contributions of variables in the long-run relation (see Univariate tests).

- Pakistan:
  - All variables are integrated of order one.
  - Johansen Trace test did not suggest the existence of a cointegration equation at 10 percent level of significance (Trace none: Prob. 0.11), but the Johansen Maximum Eigenvalue test suggested a long-run relation at 5 percent level (Max-Eigen none: Prob. 0.046).
  - The estimated long-run relation is presented in the main text (equation (1) and Table 2).
  - VECM diagnostics show residuals are not autocorrelated; multivariate JB test does not reject normality.
  - Exclusion tests suggest all variables included in the cointegration relation, with the exception of the exchange rate, are significant at 1 percent level of significance.
  - Additional note: A VECM estimated without the exchange rate produced a model where “Remittances” was weakly exogenous (χ2 (1) = 0.15, prob=0.70), indicating such a model could not be considered as providing a potential explanation for the flow of remittances to Pakistan.

- Tunisia:
  - Initial analyses using dollar-denominated variables produced model uncertainty: remittances were weakly exogenous in a model including exchange rate (χ2 (1) = 0.25, prob=0.62), and agricultural GDP may not enter the long-run relation in a model without exchange rate.
  - An alternative analysis using values in local currency was conducted.
  - All local-currency variables are integrated of order one.
  - Johansen trace test suggested the existence of a long-run relation at 5 percent level (Trace none: Prob. 0.028), while the Johansen eigenvalue (maximum-eigenvalue) test was less conclusive (Max-Eigen none: Prob. 0.125).
  - VECM diagnostics indicate residuals are not autocorrelated and multivariate JB test does not reject normality; exclusion tests suggest all variables included in the cointegration relation are significant.
  - Therefore, the model presented in equation (1) could explain the flow of remittances to Tunisia.

*Source: Author's estimates*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp06130.pdf_