## Gravity Model: Explanation of Variables

## Source details

**Canonical URL:** [Gravity Model: Explanation of Variables](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1031.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1031.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1031.pdf.json)

---

### I. Introduction
- Objective: Estimate gravity models separately for exports and imports to assess whether MENA trade volumes are significantly lower than expected given economic, cultural, and geographical characteristics.
- Key approach:
  - Establish a ‘baseline’ standard gravity model.
  - Compare against an ‘augmented’ gravity model that adds four variables from the World Bank’s Business Enterprise surveys.
- Main empirical preview:
  - Standard gravity variables cannot explain a significant part of MENA’s trade performance, particularly on exports.
  - The ‘augmented’ gravity model shows the survey variables are highly significant in explaining MENA’s underperformance in trade.
- Caveats on survey data:
  - Representativeness of respondents is uncertain.
  - Surveys reflect perceptions, which may differ from reality.
  - Openness to respond may vary across countries (e.g., autocratic or closed regimes).

### II. Does the MENA Region Trade Too Little?
- Empirical pattern:
  - Non-oil exports are significantly lower as a share of GDP compared to all other non-advanced regions.
  - MENA imports are lower as a share of GDP compared to the same regions, with the exception of sub-Saharan Africa.
- Select literature findings and quantitative estimates:
  - Al-Atrash and Yousef (2000): intra-Arab trade and Arab trade with the rest of the world are lower than predicted by gravity equations; intra-Arab trade should be about 10-15 percent higher than observed.
  - Bannister and Erickson von Allmen (2001): Palestinian exports to the rest of the world are almost 80 percent below what would be expected given Palestinian characteristics.
  - Iqbal and Nabli (2007): non-oil exports of MENA countries are, on average, one-third of the levels that would be expected on the basis of per capita incomes, resource endowments, and population sizes.
    - Only Jordan and Morocco have non-oil export levels close to predictions.
    - The world’s three biggest underperformers in non-oil exports are MENA countries (Algeria, Iran, and Egypt).
    - Per capita manufacturing imports in the MENA region are about half of what would be predicted on the basis of per capita incomes and population sizes.
  - Note: Results for Egypt differ across studies due to methodology and time period differences.

### III. Key Constraints on Trade in the MENA Region
- Trade policy and protection:
  - Trade regimes in the MENA region are among the most protective in the world, with tariff rates that are high and dispersed.
  - Nontariff barriers, including lengthy processes to comply with customs and quality control standards, are widespread.
  - Persistent overvaluation of exchange rates has compounded trade-impeding effects.
- Cross-country heterogeneity:
  - Algeria, Morocco, Pakistan, Jordan, and Tunisia had tariff rates (unweighted) averaging over 10 percent in 2005/2006, with Tunisia’s being almost 23 percent.
  - Several Gulf states have very low tariff rates and virtually no other barriers to trade (other than on goods from Israel).
- Infrastructure, transport, and logistics:
  - High transport, logistics, and communications costs impede trade.
  - Lack of adequate infrastructure is a major investor concern in the region (an important exception being the Gulf countries).
  - Institutional failures that do not align prices with costs and lack of an enabling environment hinder private provision of infrastructure.
- Human capital and skills:
  - Lack of skilled workers at internationally competitive wages is a significant investor concern.
  - Low human skill/natural resource ratios hinder export diversification and competitiveness in manufacturing and processing.
- Policy implication summary:
  - In addition to standard trade liberalization, MENA countries need strategies to address the human skill gap and to develop adequate physical infrastructure to exploit trade potential.

### IV. The World Bank’s Business Enterprise Surveys
- Dataset scope:
  - Covers 100,000 businesses in over 100 countries on constraints to business performance and growth.
- Four survey variables used in the augmented gravity model:
  - (i) Percentage of firms that trade that identify transportation as a major or severe constraint.
  - (ii) Percentage of firms that trade that identify customs and trade regulations as a major or severe constraint.
  - (iii) Average time (days) to clear exports through customs.
  - (iv) Average time (days) to clear imports through customs.
- Regional and country-level survey findings (MENA vs. world/OECD and selected countries):
  - Transport constraint:
    - MENA average not much higher than world average, but West Bank and Gaza (WBG), Lebanon, and Algeria report substantially greater difficulties.
  - Customs and trade regulations:
    - MENA average noticeably higher than the world average.
    - Lebanon, Algeria, and WBG have a significantly higher percentage of firms reporting customs and trade regulations as particularly onerous.
    - Morocco and Jordan report significantly lower percentages than the world or MENA averages.
  - Customs clearance times:
    - Average time to clear exports through customs: MENA = 6.3 days, World = 5.6 days, OECD = 5.3 days.
      - High export clearance times noted in Algeria and Lebanon.
      - Lower export clearance times noted in Morocco and Jordan.
    - Average time to clear imports through customs: MENA = 11.4 days, World = 9.0 days, OECD = 5.7 days.
      - Import clearance times considerably above averages in WBG and Algeria.
      - Morocco and Jordan appear to have very efficient customs clearance procedures.
- Data source note:
  - These surveys are separate from the World Bank’s ‘Doing Business’ Indicators and can be accessed through https://www.enterprisesurveys.org.

### V. Specification of the Gravity Model
- Modeling strategy:
  - Estimate standard gravity models as a baseline, then augment with the four Business Enterprise survey variables to assess explanatory power for MENA trade underperformance.
- Theoretical foundations:
  - Gravity model predicts bilateral trade flows based on GDP sizes and distance; trade depends positively on GDP sizes and negatively on distance.
  - Consistency of the gravity model with theory:
    - Deardorff (1998): consistency with Hecksher-Ohlin theory.
    - Helpman (1984, 1987), Bergstrand (1985), Helpman and Krugman (1987), Feenstra, Markusen and Rose (2001): derivations from models of differentiated products, imperfect markets, increasing returns, and differences in tastes.
- Variables included (standard and extended):
  - Standard gravity variables:
    - GDP of exporter and importer.
    - Distance between trading partners.
  - Additional controls:
    - Population.
    - Common language dummy.
    - Border dummy.
    - Landlocked/seaport access dummy.
    - Overall trade restrictiveness index.
    - Regional trade arrangement dummies.
  - Augmented model:
    - Adds LOG(TRANS i), LOG(EXPCLR i), LOG(IMPCLR i), and customs/trade regulation percentage (dropped later for lack of robustness).

### Model specification (formal)
- Multiplicative form (equation (1)):
  - Tij = α0 GDPiα1 GDPjα2 POPiα3 POPjα4 DISTANCEijα5 Wijα6 eij
  - Definitions:
    - Tij: flow of trade between countries i and j.
    - GDPi, GDPj: GDP of countries i and j.
    - POPi, POPj: populations.
    - DISTANCEij: linear distance between capitals.
    - Wij: other factors influencing bilateral trade.
    - eij: log normally distributed error term.
- Linearized (log) form (equation (2)):
  - Log(Tij) = α0 + α1 Log(GDPi) + α2 Log(GDPj) + α3 Log(POPi) + α4 Log(POPj) + α5 Log(DISTANCEij) + β1 LANG + β2 BORDER + β3 REPLL i + β4 TRI i (β4 TRI j) + β5 MENA + ∑ ni βi REGIONij + eij

### Dummy and survey variables (used in estimation)
- LANG: dummy = 1 if countries i and j share a common language (English, French, Arabic, Portuguese or Spanish), zero otherwise.
- BORDER: dummy = 1 if countries i and j share a common border, zero otherwise.
- REPLL i: dummy = 1 if country i is landlocked, zero otherwise.
- TRI i / TRI j: IMF’s overall trade restrictiveness index for country i / country j.
- REGIONij: series of dummies = 1 if countries i and j belong to a preferential trading arrangement (including ASEAN, COMESA, EU and MERCOSUR), zero otherwise.
- MENA: dummy = 1 if country i is in the Middle East and North Africa region, zero otherwise.
- Survey-augmented variables:
  - LXij: Log of exports, in current US dollars, from country i to country j
  - LMij: Log of imports, in current US dollars, of country i from country j
  - LGDPi / LGDPj: Log of GDP, in current US dollars
  - LPOPi / LPOPj: Log of population
  - LDISTANCE: Log of distance (km) between capital cities
  - LOG(TRANSi): Percent of firms in Country i that trade identifying transportation as a major constraint
  - LOG(EXPCLRi): average time (days) to clear exports through customs in Country i
  - LOG(IMPCLRi): average time (days) to clear imports through customs in Country i
  - ASEAN, COMESA, EU, MERCOSUR: region dummies as defined above

### Data and estimation
- Sample composition:
  - Dataset: 88 countries with survey results, including 8 MENA countries, supplemented with 5 non-survey countries (France, Israel, Italy, the United Kingdom, and the United States).
  - MENA sample: Algeria, Egypt, Jordan, Lebanon, Mauritania, Morocco, Turkey, and the West Bank and Gaza.
  - These countries represent around 66 percent of trade of the MENA countries in the sample.
- Time period:
  - Standard gravity variables: averages over 2005–07.
  - Survey data: cover 2005–07 (depending on availability).
- Censoring and estimation method:
  - Around 23 percent of observations for bilateral trade are censored at zero.
  - Gravity equations estimated using a censored regression model (TOBIT), assuming underlying log trade is a large negative number when observed bilateral trade is zero.

### Empirical results — standard gravity model (main findings)
- Signs and significance:
  - Logs of GDP (both partners): positive and significant.
  - Logs of population: negative; statistically significant only for the trading partner country in the case of imports.
  - Distance: negative and significant.
  - Common language (LANG) and border (BORDER): positive impacts.
  - Landlocked status (REPLL i): significant and negative for exports and imports.
  - IMF trade restrictiveness index (TRI i / TRI j): negative and significant for exports and imports.
  - MENA dummy: negative and significant for exports, not significant for imports.
- MENA export gap:
  - Coefficient on MENA dummy in export equation implies that MENA exports are more than 86 percent below what would be expected given the characteristics of their economies.
- Table 1 selected numeric entries:
  - MENA coefficient (exports): -1.984, Standard Error 0.751, z-stat -2.642, significance at 1 percent level.
  - MENA coefficient (imports): -0.135, Standard Error 0.738, z-stat -0.183, not significant.
  - Total observations: 7,832 (exports), 7,820 (imports).
  - Uncensored observations: 5,679 (exports), 5,748 (imports).

### Augmented gravity model with survey constraints — main findings
- General:
  - Survey variables significantly explain MENA’s relative underperformance in trade.
  - Customs and trade regulations variable was not robust and often statistically insignificant; it was dropped from the augmented model.
  - Results similar to standard model with two exceptions: population coefficient for reporting country becomes positive and statistically significant for imports.
- Transport constraint:
  - Coefficients on LOG(TRANS i) are negative and statistically significant for exports and imports.
  - Implied elasticity for countries with positive bilateral trade: -0.67 for exports and -0.90 for imports.
  - Applying these elasticities: reducing the transport constraint from the average for the MENA region to the world average could raise exports by 9½ percent and imports by 11½ percent, ceteris paribus.
- Customs clearance times:
  - Estimated coefficients on number of days to clear exports/imports through customs are negative and statistically significant.
  - Implied elasticities for countries with positive bilateral trade: -0.88 for exports and -1.15 for imports.
  - Calculated impacts for MENA moving to world average:
    - Reducing days to clear exports to world average could raise exports by 11 percent, ceteris paribus.
    - Reducing days to clear imports to world average could raise imports by 30½ percent, ceteris paribus.
- Table 2 selected numeric entries:
  - LOG(TRANSi) coefficient: -0.918 (exports), Standard Error 0.254, z-stat -3.610, significant at 1 percent level; -1.222 (imports), Standard Error 0.267, z-stat -4.581, significant at 1 percent level.
  - LOG(EXPCLRi): -1.208 (exports), Standard Error 0.449, z-stat -2.694, significant at 1 percent level.
  - LOG(IMPCLRi): -1.559 (imports), Standard Error 0.444, z-stat -3.506, significant at 1 percent level.
  - MENA coefficient in augmented model: -1.216 (exports), Standard Error 0.743, z-stat -1.635 (not significant at 10 percent); 1.107 (imports), Standard Error 0.739, z-stat 1.497 (not significant).
  - Total observations: 7,742 (exports), 7,636 (imports).
  - Uncensored observations: 5,649 (exports), 5,629 (imports).

### Interpretation and comparative evidence
- Interpretation:
  - Reducing transport constraints and improving customs efficiency likely have particularly strong effects on MENA exports relative to other regions, even if impacts on imports are similar across regions.
  - Transport constraints are especially significant in Algeria, Lebanon, and the West Bank and Gaza; for the West Bank and Gaza, transport constraints reflect both infrastructure and restrictions on movement and access.
- Consistency with other studies:
  - Njinkeu, Wilson and Fosso (2008): port and air transport infrastructure quality positively impact African trade; customs and regulatory environments are main obstacles to intra-African trade.
  - Dollar, Hallward-Driemeier, and Mengistae (2005, 2006): power shortages and customs delays are important bottlenecks for exporting firms; sound investment climate (low customs clearance times, reliable infrastructure, good financial services) makes domestic firms more likely to export and attracts foreign investment.

### Conclusions and policy implications
- Main conclusion:
  - Trade volumes in the MENA region are significantly lower than expected given economic, cultural, and geographical characteristics; MENA exports are estimated to be more than 86 percent below expected levels.
  - Standard gravity variables adequately explain MENA import volumes but not exports.
- Policy priorities:
  - Reduce transport constraints:
    - Likely requires long-term action and active participation of the private sector in financing and provision of transport services.
    - Governments could consider greater use of private-public partnerships, particularly where fiscal pressures exist.
  - Improve efficiency of customs clearance procedures:
    - Achievable in the short- to medium-term by streamlining documents required for clearance of exports and imports.
- Expected quantitative benefits:
  - Reducing transport constraint from MENA average to world average: increase exports by approximately 10 percent and imports by over 11 percent.
  - Reducing average days to clear exports to world average: raise exports by around 11 percent.
  - Reducing average days to clear imports to world average: raise imports by over 30 percent.
- Caveats:
  - Interpret survey data with caution due to survey limitations and possible cross-country differences in respondent openness.
  - Despite caveats, results strongly suggest that reducing transport constraints and improving customs clearance efficiency could significantly raise MENA exports and imports.

*Source — IMF working paper section “1. Gravity Model: Explanation of Variables”*

### 1. Gravity Model: Explanation of Variables ........................................................................... 1

### Gravity Model: Explanation of Variables

### I. Introduction
- Objective: Estimate gravity models separately for exports and imports to assess whether MENA trade volumes are significantly lower than expected given economic, cultural, and geographical characteristics.
- Key approach:
  - Establish a ‘baseline’ standard gravity model.
  - Compare against an ‘augmented’ gravity model that adds four variables from the World Bank’s Business Enterprise surveys.
- Main empirical preview:
  - Standard gravity variables cannot explain a significant part of MENA’s trade performance, particularly on exports.
  - The ‘augmented’ gravity model shows the survey variables are highly significant in explaining MENA’s underperformance in trade.
- Caveats on survey data:
  - Representativeness of respondents is uncertain.
  - Surveys reflect perceptions, which may differ from reality.
  - Openness to respond may vary across countries (e.g., autocratic or closed regimes).

### II. Does the MENA Region Trade Too Little?
- Empirical pattern:
  - Non-oil exports are significantly lower as a share of GDP compared to all other non-advanced regions.
  - MENA imports are lower as a share of GDP compared to the same regions, with the exception of sub-Saharan Africa.
- Select literature findings and quantitative estimates:
  - Al-Atrash and Yousef (2000): intra-Arab trade and Arab trade with the rest of the world are lower than predicted by gravity equations; intra-Arab trade should be about 10-15 percent higher than observed.
  - Bannister and Erickson von Allmen (2001): Palestinian exports to the rest of the world are almost 80 percent below what would be expected given Palestinian characteristics.
  - Iqbal and Nabli (2007): non-oil exports of MENA countries are, on average, one-third of the levels that would be expected on the basis of per capita incomes, resource endowments, and population sizes.
    - Only Jordan and Morocco have non-oil export levels close to predictions.
    - The world’s three biggest underperformers in non-oil exports are MENA countries (Algeria, Iran, and Egypt).
    - Per capita manufacturing imports in the MENA region are about half of what would be predicted on the basis of per capita incomes and population sizes.
  - Note: Results for Egypt differ across studies due to methodology and time period differences.

### III. Key Constraints on Trade in the MENA Region
- Trade policy and protection:
  - Trade regimes in the MENA region are among the most protective in the world, with tariff rates that are high and dispersed.
  - Nontariff barriers, including lengthy processes to comply with customs and quality control standards, are widespread.
  - Persistent overvaluation of exchange rates has compounded trade-impeding effects.
- Cross-country heterogeneity:
  - Algeria, Morocco, Pakistan, Jordan, and Tunisia had tariff rates (unweighted) averaging over 10 percent in 2005/2006, with Tunisia’s being almost 23 percent.
  - Several Gulf states have very low tariff rates and virtually no other barriers to trade (other than on goods from Israel).
- Infrastructure, transport, and logistics:
  - High transport, logistics, and communications costs impede trade.
  - Lack of adequate infrastructure is a major investor concern in the region (an important exception being the Gulf countries).
  - Institutional failures that do not align prices with costs and lack of an enabling environment hinder private provision of infrastructure.
- Human capital and skills:
  - Lack of skilled workers at internationally competitive wages is a significant investor concern.
  - Low human skill/natural resource ratios hinder export diversification and competitiveness in manufacturing and processing.
- Policy implication summary:
  - In addition to standard trade liberalization, MENA countries need strategies to address the human skill gap and to develop adequate physical infrastructure to exploit trade potential.

### IV. The World Bank’s Business Enterprise Surveys
- Dataset scope:
  - Covers 100,000 businesses in over 100 countries on constraints to business performance and growth.
- Four survey variables used in the augmented gravity model:
  - (i) Percentage of firms that trade that identify transportation as a major or severe constraint.
  - (ii) Percentage of firms that trade that identify customs and trade regulations as a major or severe constraint.
  - (iii) Average time (days) to clear exports through customs.
  - (iv) Average time (days) to clear imports through customs.
- Regional and country-level survey findings (MENA vs. world/OECD and selected countries):
  - Transport constraint:
    - MENA average not much higher than world average, but West Bank and Gaza (WBG), Lebanon, and Algeria report substantially greater difficulties.
  - Customs and trade regulations:
    - MENA average noticeably higher than the world average.
    - Lebanon, Algeria, and WBG have a significantly higher percentage of firms reporting customs and trade regulations as particularly onerous.
    - Morocco and Jordan report significantly lower percentages than the world or MENA averages.
  - Customs clearance times:
    - Average time to clear exports through customs: MENA = 6.3 days, World = 5.6 days, OECD = 5.3 days.
      - High export clearance times noted in Algeria and Lebanon.
      - Lower export clearance times noted in Morocco and Jordan.
    - Average time to clear imports through customs: MENA = 11.4 days, World = 9.0 days, OECD = 5.7 days.
      - Import clearance times considerably above averages in WBG and Algeria.
      - Morocco and Jordan appear to have very efficient customs clearance procedures.
- Data source note:
  - These surveys are separate from the World Bank’s ‘Doing Business’ Indicators and can be accessed through https://www.enterprisesurveys.org.

### V. Specification of the Gravity Model
- Modeling strategy:
  - Estimate standard gravity models as a baseline, then augment with the four Business Enterprise survey variables to assess explanatory power for MENA trade underperformance.
- Theoretical foundations:
  - Gravity model predicts bilateral trade flows based on GDP sizes and distance; trade depends positively on GDP sizes and negatively on distance.
  - Consistency of the gravity model with theory:
    - Deardorff (1998): consistency with Hecksher-Ohlin theory.
    - Helpman (1984, 1987), Bergstrand (1985), Helpman and Krugman (1987), Feenstra, Markusen and Rose (2001): derivations from models of differentiated products, imperfect markets, increasing returns, and differences in tastes.
- Variables included (standard and extended):
  - Standard gravity variables:
    - GDP of exporter and importer.
    - Distance between trading partners.
  - Additional controls included following recent literature:
    - Population (expected sign ambiguous given GDP control).
    - Common language dummy (proxy for cultural proximity).
    - Border dummy (captures neighbor trading propensity).
    - Landlocked/seaport access dummy (direct access to a seaport promotes trade).
    - Overall trade restrictiveness index (higher restrictiveness lowers trade volumes).
    - Regional trade arrangement dummies (capture membership effects on bilateral trade).
  - Augmented model:
    - Adds the four World Bank Business Enterprise survey variables listed above to examine their impact on bilateral trade flows.

*Source: IMF working paper section “1. Gravity Model: Explanation of Variables”*

### 22.      More formally, the standard gravity model is given by

### 22.      More formally, the standard gravity model is given by

### Model specification
- Multiplicative form (equation (1)):
  - Tij = α0 GDPiα1 GDPjα2 POPiα3 POPjα4 DISTANCEijα5 Wijα6 eij
  - Tij is the flow of trade between countries i and j.
  - GDPi and GDPj are the GDP of countries i and j.
  - POPi and POPj are the populations of countries i and j.
  - DISTANCEij is the linear distance between the capitals of countries i and j.
  - Wij includes other factors that influence bilateral trade.
  - eij is a log normally distributed error term.
- Linearized (log) form (equation (2)):
  - Log(Tij) = α0 + α1 Log(GDPi) + α2 Log(GDPj) + α3 Log(POPi) + α4 Log(POPj)
    + α5 Log(DISTANCEij) + β1 LANG + β2 BORDER + β3 REPLLi
    + β4 TRI i (β4 TRI j) + β5 MENA + ∑ ni βi REGIONij + eij

### Dummy and survey variables (definitions used in estimation)
- LANG: dummy = 1 if countries i and j share a common language (English, French, Arabic, Portuguese or Spanish), zero otherwise.
- BORDER: dummy = 1 if countries i and j share a common border, zero otherwise.
- REPLL i: dummy = 1 if country i is landlocked, zero otherwise.
- TRI i / TRI j: IMF’s overall trade restrictiveness index for country i / country j.
- REGIONij: series of dummies = 1 if countries i and j belong to a preferential trading arrangement (including ASEAN, COMESA, EU and MERCOSUR), zero otherwise.
- MENA: dummy = 1 if country i is in the Middle East and North Africa region, zero otherwise.
- Survey-augmented variables (Box 1):
  - LXij: Log of exports, in current US dollars, from country i to country j
  - LMij: Log of imports, in current US dollars, of country i from country j
  - LGDPi / LGDPj: Log of GDP, in current US dollars
  - LPOPi / LPOPj: Log of population
  - LDISTANCE: Log of distance (km) between capital cities
  - LOG(TRANSi): Percent of firms in Country i that trade identifying transportation as a major constraint
  - LOG(EXPCLRi): average time (days) to clear exports through customs in Country i
  - LOG(IMPCLRi): average time (days) to clear imports through customs in Country i
  - ASEAN, COMESA, EU, MERCOSUR: region dummies as defined above

### Data and estimation
- Sample composition:
  - Dataset consists of 88 countries for which survey results are available, including 8 countries in the MENA region, supplemented with data for 5 other non-survey countries (France, Israel, Italy, the United Kingdom, and the United States).
  - The MENA sample includes Algeria, Egypt, Jordan, Lebanon, Mauritania, Morocco, Turkey, and the West Bank and Gaza.
  - These countries represent around 66 percent of trade of the MENA countries in the sample.
- Time period:
  - Variables in the standard gravity model use averages over 2005–07.
  - Survey data cover the years 2005–07 (depending on year of availability).
- Censoring and estimation method:
  - Around 23 percent of observations for bilateral trade are censored at zero.
  - Ordinary Least Squares (OLS) would produce biased estimates when dependent variable is censored at zero.
  - Gravity equations are estimated using a censored regression model (TOBIT), adopting the assumption that the underlying value of the log of trade is a large negative number when observed bilateral trade is zero.

### Empirical results — standard gravity model (summary of main findings)
- Sign and significance:
  - Coefficients on logs of GDP of both trading partners: positive and significant.
  - Coefficients on logs of population: negative; statistically significant only for the trading partner country in the case of imports.
  - Distance: negative and significant.
  - Common language (LANG) and border (BORDER): positive impacts on trade volumes.
  - Landlocked status (REPLL i): significant and negative for both exports and imports.
  - IMF trade restrictiveness index (TRI i / TRI j): negative and significant for both exports and imports.
  - MENA dummy: negative and significant for exports, not significant for imports.
- MENA export gap:
  - Coefficient on MENA dummy in export equation implies that MENA exports are more than 86 percent below what would be expected given the characteristics of their economies.
- Table 1 key numeric entries (selected):
  - MENA coefficient (exports): -1.984, Standard Error 0.751, z-stat -2.642, significance at 1 percent level.
  - MENA coefficient (imports): -0.135, Standard Error 0.738, z-stat -0.183, not significant.
  - Total observations: 7,832 (exports), 7,820 (imports).
  - Uncensored observations: 5,679 (exports), 5,748 (imports).

### Augmented gravity model with survey constraints — main findings
- General:
  - Survey variables are significant in explaining MENA’s relative underperformance in trade.
  - The customs and trade regulations variable was not robust and often statistically insignificant; it was dropped from the augmented model.
  - Results broadly similar to standard model with two exceptions: the coefficient on population for reporting country becomes positive and statistically significant in the case of imports.
- Transport constraint:
  - Coefficients on the transport constraint variable are negative and statistically significant for both exports and imports.
  - Implied elasticity for countries with positive bilateral trade: -0.67 for exports and -0.90 for imports.
  - Applying these elasticities: reducing the transport constraint from the average for the MENA region to the world average could raise exports by 9½ percent and imports by 11½ percent, ceteris paribus.
  - When survey constraints are added, the MENA dummy loses statistical significance at the 10 percent level in the export equation.
- Customs clearance times:
  - Estimated coefficients on number of days to clear exports/imports through customs are negative and statistically significant.
  - Implied elasticities for countries with positive bilateral trade: -0.88 for exports and -1.15 for imports.
  - Calculated impacts for MENA moving to world average:
    - Reducing days to clear exports to world average could raise exports by 11 percent, ceteris paribus.
    - Reducing days to clear imports to world average could raise imports by 30½ percent, ceteris paribus.
- Table 2 key numeric entries (selected):
  - LOG(TRANSi) coefficient: -0.918 (exports), Standard Error 0.254, z-stat -3.610, significant at 1 percent level; -1.222 (imports), Standard Error 0.267, z-stat -4.581, significant at 1 percent level.
  - LOG(EXPCLRi): -1.208 (exports), Standard Error 0.449, z-stat -2.694, significant at 1 percent level.
  - LOG(IMPCLRi): -1.559 (imports), Standard Error 0.444, z-stat -3.506, significant at 1 percent level.
  - MENA coefficient in augmented model: -1.216 (exports), Standard Error 0.743, z-stat -1.635 (not significant at 10 percent); 1.107 (imports), Standard Error 0.739, z-stat 1.497 (not significant).
  - Total observations: 7,742 (exports), 7,636 (imports).
  - Uncensored observations: 5,649 (exports), 5,629 (imports).

### Interpretation and comparative evidence
- Interpretation:
  - Reducing transport constraints and improving customs efficiency likely have particularly strong effects on MENA exports relative to other regions, even if impacts on imports are similar across regions.
  - Transport constraints are especially significant in Algeria, Lebanon, and the West Bank and Gaza; for the West Bank and Gaza, transport constraints reflect both infrastructure and restrictions on movement and access.
- Consistency with other studies:
  - Njinkeu, Wilson and Fosso (2008): port and air transport infrastructure quality positively impact African trade; customs and regulatory environments are main obstacles to intra-African trade.
  - Dollar, Hallward-Driemeier, and Mengistae (2005, 2006): power shortages and customs delays are important bottlenecks for exporting firms; sound investment climate (low customs clearance times, reliable infrastructure, good financial services) makes domestic firms more likely to export and attracts foreign investment.

### Conclusions and policy implications
- Main conclusion:
  - Trade volumes in the MENA region are significantly lower than expected given economic, cultural, and geographical characteristics; MENA exports are estimated to be more than 86 percent below expected levels.
  - Standard gravity variables adequately explain MENA import volumes but not exports.
- Policy priorities:
  - Reduce transport constraints:
    - Likely requires long-term action and active participation of the private sector in financing and provision of transport services.
    - Governments could consider greater use of private-public partnerships, particularly where fiscal pressures exist.
  - Improve efficiency of customs clearance procedures:
    - Achievable in the short- to medium-term by streamlining documents required for clearance of exports and imports.
  - Expected quantitative benefits:
    - Reducing transport constraint from MENA average to world average: increase exports by approximately 10 percent and imports by over 11 percent.
    - Reducing average days to clear exports to world average: raise exports by around 11 percent.
    - Reducing average days to clear imports to world average: raise imports by over 30 percent.
- Caveats:
  - Survey data interpretations require caution due to usual survey caveats and possible cross-country differences in respondent openness depending on culture and political regime.
  - Despite caveats, results strongly suggest that reducing transport constraints and improving customs clearance efficiency could significantly raise MENA exports and imports.

*Italic: Source — IMF working paper content provided in the supplied PDF excerpt.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp1031.pdf_