## _wp04109 - 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared

## Source details

**Canonical URL:** [_wp04109 - 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04109.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04109.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04109.pdf.json)

---

### Introduction
- Objective: investigate performance of the main trading blocs in the Western Hemisphere by estimating a nonlinear gravity equation that incorporates a novel trade cost factor.
- Sample and period: 64 industrial and developing economies for the period 1978–2001.
- High-level empirical findings:
  - The income elasticity of trade is larger than zero but less than one.
  - Geographical distance and a shared language have become somewhat less relevant over time in explaining trade among countries.
  - Trade among neighboring countries increased in the early 1990s.
  - Trading-bloc performance in the Western Hemisphere:
    - NAFTA: trade integration outweighed trade diversion.
    - MERCOSUR: trade integration and trade diversion went hand in hand.
    - CACM and the Andean Community: evidence points basically to trade diversion.

### Theoretical motivation and model extension
- Base framework: Anderson and Van Wincoop (2001) gravity equation where trade is determined by relative economic mass and relative trade resistance.
- Key theoretical observations and formulations (as given in source):
  - tij is the trade cost factor, tij ≥ 1. (equation (1))
  - Anderson–Van Wincoop gravity (equation (2)):
    - Xij = (GDPi GDPj / GDPw) × t_ij^{1−σ} × Pi^{σ−1} Pj^{σ−1}  (expressed in source notation).
  - Trade cost factor depending on distance and exports (equation (3)):
    - tij = d_ij^{β/(1−σ)} × Xij^{η/(1−σ)}, where β < 0, −1 < η < 1, and 1 − σ < 0.
  - Combined implication (equation (4)) shows trade increases with shorter distance, higher multilateral resistance, and greater relative economic mass.
  - Elasticity of trade w.r.t. relative economic mass equals 1/(1 − η):
    - If 0 < η < 1 then elasticity > 1.
    - If −1 < η < 0 then elasticity < 1.
    - If η = 0 then elasticity = 1.
- Interpretation: allowing tij to depend on Xij reconciles non-unitary income elasticities with the gravity framework via externalities (η).

### Empirical specification
- Nonlinear empirical model estimated (equation (5) in source):
  - Exportijt = f(GDPit, GDPjt, POPit, POPjt, dij, Lij, Aij, Bij, ib PTE, jb PTI, θi, κj) + εij with nonlinear functional form as stated in source.
  - Structural decomposition for μ (as reported):
    - μij = θi + κj + ib PTE + jb PTI + Bij λ3 + Lij λ4 + Aij λ5.
- Rationale for nonlinear specification:
  - Allows inclusion of zero trade flows.
  - Ensures trade → 0 as GDP → 0.
  - Nonlinear regression fits better and error terms behave better than linear alternatives.
- Estimation approach:
  - Nonlinear least squares estimated cross-section for each year (recursive yearly estimation of coefficients).

### Key estimated coefficients and time trends (selected-year values from Table 1)
- Economic mass (income elasticity of trade):
  - Selected-year values: 0.86**, 0.81**, 0.68**, 0.68**, 0.72**, 0.80**, 0.72**, 0.68**, 0.77** (for years 1978, 1980, 1983, 1986, 1989, 1992, 1995, 1998, 2001 respectively).
  - From these estimates the paper infers η = −0.25 (implying net negative externalities).
- Distance:
  - Selected-year values: −0.58**, −0.57**, −0.55**, −0.51**, −0.43**, −0.44**, −0.42**, −0.49**, −0.52**.
  - Comparative static example reported: with estimated distance coefficient −0.51 and distance(A,C) = 3 × distance(A,B), trade(A,C) ≈ 57 percent of trade(A,B).
- Adjacency (Aij):
  - Selected-year values: 0.58**, 0.66**, 0.69**, 0.58**, 0.68**, 1.23**, 1.28**, 1.12**, 1.04**.
  - Noted jump in 1990 to a higher plateau coinciding with surge in RTAs.
- Shared language (Lij):
  - Selected-year values: 0.39**, 0.21**, 0.35**, 0.51**, 0.46**, 0.15**, 0.12**, 0.12**, 0.04**.
  - Shows high volatility around a decreasing trend; small positive value in 2001.
- Population controls:
  - Population - Export Country: 0.14**, 0.26**, 0.35**, 0.35**, 0.41**, 0.32**, 0.39**, 0.51**, 0.36**.
  - Population - Import Country: 0.20**, 0.32**, 0.51**, 0.52**, 0.44**, 0.38**, 0.49**, 0.59**, 0.52**.
- Table 1 summary statistics for selected years:
  - Number of observations: 3,909; 3,848; 3,974; 3,991; 4,005; 4,010; 4,014; 4,017; 4,017.
  - Adjusted R2: 0.95, 0.93, 0.94, 0.97, 0.96, 0.95, 0.96, 0.97, 0.97.
  - Significance notation: * and ** indicate coefficients significantly different from zero at the 5 and 1 percent level, respectively.

### Trading blocs — Findings by bloc (selected-year values and interpretation)
- General:
  - Four Western Hemisphere blocs analyzed: MERCOSUR, NAFTA, CACM, Andean Community.
  - Bloc-specific dummy variables: Bij (integration), ib PTE (propensity to export), jb PTI (propensity to import).
  - Interpretation guidance: decreases in a bloc’s propensity to export or import concomitant with formation of a trade bloc can indicate trade diversion (import substitution).

- MERCOSUR (Argentina, Brazil, Paraguay, Uruguay)
  - Level of Integration (selected-year): 0.09, 0.16, 0.29, 1.18*, 1.77**, 1.29**, 1.72**, 2.08**, 1.57**.
  - Propensity to Import: 0.33*, 0.24, 0.45, 0.17, −0.74*, −0.76**, −0.84**, −1.00**, −0.60**.
  - Propensity to Export: 0.08, −0.68**, 0.56*, 0.23, −0.56*, −0.64**, −1.15**, −1.26**, −0.56**.
  - Interpretation:
    - Integration rose since early 1980s, leveled off since mid-1990s, decline since 1998 possibly due to macro shocks (e.g., Brazilian real devaluation January 1999).
    - Propensity to import decreased sharply after early 1980s.
    - Propensity to export increased sharply in 1980s (debt-crisis-related), then declined.
    - Net assessment: integration and trade diversion went hand in hand.

- NAFTA (Canada, United States, Mexico)
  - Level of Integration (selected-year): −0.23*, 0.06, −0.12, −0.32**, −0.06, −0.16*, 0.07, 0.28**, 0.37**.
  - Propensity to Import: 0.29, 0.09, 1.07*, 1.32**, 0.57, 0.07, 0.04, −0.25*, −0.35**.
  - Propensity to Export: 0.12, −0.39*, 0.66*, 0.87**, 0.07, −0.09, −0.09, −0.15, −0.22.
  - Interpretation:
    - Level of integration below zero in 1994 before NAFTA; increased with Mexico’s liberalization; temporary reversal around 1990; upward trend resumes after 1994.
    - Weak indication of trade diversion in 1990s versus 1980s; stronger evidence of integration particularly after 1997.

- CACM (Central American Common Market)
  - Level of Integration (selected-year): 3.33*, 3.50*, 2.68, 1.14, 1.67, 1.97, 2.23, 1.53, 1.44.
  - Propensity to Import: 0.13, 0.06, −1.72**, −2.41**, −2.17**, −2.11**, −1.84**, −1.55**, −1.65**.
  - Propensity to Export: 0.73, 0.42, 0.75, 0.73, 0.35, 0.42, −0.05, 0.62, 1.05**.
  - Interpretation:
    - Level of integration flat and not significantly different from zero during last two decades.
    - Propensity to export flat and not significantly different from zero.
    - Propensity to import dropped in early 1980s then increased in late 1980s.
    - Net assessment: evidence points to trade diversion without further integration.

- Andean Community
  - Level of Integration (selected-year): −0.24, −0.12, 0.33, 0.45, 0.83, 1.08, 1.91*, 1.86*, 1.58*.
  - Propensity to Import: 1.27**, 1.05**, 1.45**, 0.41**, 0.64, 0.78**, 0.23, 0.03, −0.04.
  - Propensity to Export: 0.94**, 0.59**, 1.08**, 0.85**, 0.75**, 0.74**, 0.19, 0.05, 0.54**.
  - Interpretation:
    - Level of integration increased around November 1990 revival but remained insignificant until 1994; increase appears one-time.
    - Propensities to export and import decreased around revival time, indicating some trade diversion.
    - Net assessment: more trade diversion than integration.

### Economic interpretation of externalities and bilateral resistance
- Trade cost factor depends on distance and volume of trade; exports can reduce tij if 0 < η < 1 (positive externalities) or increase tij if −1 < η < 0 (negative externalities).
- Income elasticity implied by specification: 1/(1 − η).
  - Larger than one if positive externalities dominate.
  - Smaller than one (but larger than zero) if negative externalities dominate.
  - Equal to one if positive and negative externalities offset.
- Cost of bilateral resistance Cij defined as proportion of trade lost because of bilateral frictions relative to relative economic mass (expressed in source notation).
  - Cij depends positively on distance (∂C/∂d > 0).
  - Sensitivity of Cij to world GDP and parameters σ and η can be positive or negative depending on parameter values.

### Data, sample, and data sources
- Sample: 64 industrial and developing economies, 1978–2001.
- Data sources as listed in source:
  - Nominal GDP, GDP per capita, population: IMF World Economic Outlook Database.
  - Bilateral trade (exports from country i to country j): IMF Direction of Trade Database.
  - Trading bloc definitions: WTO (2000).
  - Language and adjacency variables: provided by CSTB.
  - Distance (between capitals): Centre d'Etudes Prospectives et d'Informations Internationales (www.cepii.fr/anglaisgraph/bdd/distances.htm).
- Trade definition used: Tij = Xij + Xji (partner country export data).

### Conclusions and policy-relevant implications
- Main empirical conclusions:
  - Distance and shared language have become somewhat less relevant over time but remain important; adjacency became more important in early 1990s, coinciding with surge in RTAs.
  - Estimated economic mass coefficient ≈ 0.8 supports net negative externalities (η inferred ≈ −0.25), implying more exports generate net frictions that increase the trade cost factor and slow trade expansion.
- Potential channels for negative externalities (identified as promising avenues for further research and policy):
  - Intrusive regulation, infrastructure and institutional restrictions (bottlenecks), external pecuniary and technical diseconomies, trade financing constraints, local bias, special interest politics.
- Bloc-specific policy-relevant conclusions:
  - MERCOSUR: integration accompanied by trade diversion; reflects simultaneous liberalization and RTA formation.
  - NAFTA: stronger evidence of integration (particularly after 1997) and weak evidence of trade diversion.
  - CACM and Andean Community: evidence points to more trade diversion than integration.

*Source: _wp04109 - 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared*

### 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared

### _wp04109 - 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared

### Introduction
- Since the early 1990s there has been an upsurge of trading blocs worldwide: the World Trade Organization (WTO) reports a total of 240 regional trade agreements (RTAs) globally, of which 70 percent were “in force” as of July 2000.
- Paper objective: investigate performance of the main trading blocs in the Western Hemisphere by estimating a nonlinear gravity equation that incorporates a novel trade cost factor.
- Sample and period: 64 industrial and developing economies for the period 1978–2001.
- Key empirical results summarized:
  - The income elasticity of trade is larger than zero but less than one.
  - Geographical distance and a shared language have become somewhat less relevant over time in explaining trade among countries.
  - Trade among neighboring countries increased in the early 1990s.
  - Trading-bloc performance in the Western Hemisphere:
    - NAFTA: trade integration outweighed trade diversion.
    - MERCOSUR: trade integration and trade diversion went hand in hand.
    - CACM and the Andean Community: evidence points basically to trade diversion.

### Theoretical motivation and model extension
- Base framework: Anderson and Van Wincoop (2001) gravity equation, where trade is determined by relative economic mass and relative trade resistance.
- Key theoretical observation: standard gravity model implies unitary elasticity of exports with respect to income, but empirical estimates typically find income elasticity between zero and one.
- Model extension hypothesis: exports generate positive and/or negative externalities that affect the trade cost factor; this can reconcile non-unitary income elasticity with the gravity framework.
- Trade cost factor formulation (as stated in the source):
  - ijijp p t ×=,  (equation (1))
    - tij is the trade cost factor, tij ≥ 1.
  - Anderson–Van Wincoop gravity (as given):
    - 1 ij ij w ij GDP GDP t X GDPP P σ−  × =   ×  ,  (equation (2))
      - Xij is consumption of country i good by country j consumers (exports from i to j);
      - GDPw is world GDP;
      - σ > 1 is the elasticity of substitution parameter;
      - Pi and Pj are multilateral resistance variables.
  - Trade cost factor that depends on distance and exports:
    - () () /1 /1 ij ij ij d t X βσ ησ − − − = ,  (equation (3))
      - dij is distance between countries i and j;
      - Xij is exports from i to j;
      - β < 0;
      - –1 < η < 1;
      - 1 – σ < 0.
  - Combined implication (as given):
    - 1 1 1 1 1 1 ij ijij w ij GDP GDP Xd PP GDP σ β η η η − − − −  ×  =    ×   .  (equation (4))
      - Interpretation: trade increases with shorter distance, higher multilateral resistance, and greater relative economic mass.
      - Elasticity of trade with respect to relative economic mass equals 1/(1 – η):
        - If trade generates net positive externalities (0 < η < 1) then elasticity > 1.
        - If trade generates net negative externalities (–1 < η < 0) then elasticity < 1.
        - If positive and negative externalities offset (η = 0) then elasticity = 1.

### Empirical specification
- Nonlinear empirical model (as proposed in the source):
  - () 1,2, ij tt ijtitjtitjtijijt ExportGDPGDPPOPPOPde α μ δδβ γε = × ×××× × +  (equation (5))
    - Exportijt: exports from country i to country j during period t.
    - POP: population.
    - dij: distance between countries i and j.
    - θi: fixed effect for country i.
    - κj: fixed effect for country j.
    - ib PTE and jb PTI: trading bloc–specific dummy variables capturing each bloc’s propensity to export and import, respectively.
    - Bij: trading bloc–specific dummy variable equal to one when both countries are members of the same one of the four blocs in the Western Hemisphere (measures overall trade integration within a bloc).
    - Lij: dummy equal to one if countries i and j share a common language.
    - Aij: dummy equal to one if countries i and j share a common border.
    - εij: error term.
  - Structural decomposition for μ (as reported):
    - ij5 ij4 ij3 jb2 ib1 ji ij A λ L λ B λ PTI λ PTE λ κ θ μ + + + + + + = .
- Rationale for nonlinear specification (from CSTB and authors):
  - Allows use of zero entries in trade data.
  - When GDP goes to zero, trade should be zero—captured only by nonlinear specification.
  - Nonlinear regression has a better fit than the linear one; error terms behave better under nonlinear specification.

### Interpretation of bloc-specific variables and trade diversion/integration
- Bloc propensity variables (PTE, PTI) are analogous to country propensities to export and import and capture trade diversion effects.
- Interpretation guidance given in the source:
  - A decrease in a bloc’s propensity to export or import concomitant with formation of a trade bloc can be interpreted as possible trade diversion (import substitution) related to that trade bloc.
- Empirical summarization for Western Hemisphere blocs (as given in Introduction):
  - NAFTA: trade integration > trade diversion.
  - MERCOSUR: both trade integration and trade diversion present.
  - CACM and Andean Community: evidence mainly of trade diversion.

*Source: _wp04109 - 1. Estimated Coefficients for Economic Mass, Distance, Adjacency, and Shared*

### 1. If j (importer) belongs to bloc 1, but i (exporter) does not belong to bloc 1, then PTI

### _wp04109 - 1. If j (importer) belongs to bloc 1, but i (exporter) does not belong to bloc 1, then PTI

### Methodology and Model Specification
- Nonlinear gravity model estimated across selected countries for each year during 1978–2001 using nonlinear least squares.
- Trade defined using partner country export data (Tij = Xij + Xji).
- Trade cost factor specification:
  - Traditional: tij = d ij^b.
  - Novel specification used in the paper:
    - tij = d ij^(1/(1−σ)) × Xij^(η/(1−σ)), where Xij represents exports from country i to country j, 1 − σ < 0, β < 0, and −1 < η < 1 (expressed in the source as () /1 ij ij ij d t X βσ ησ − − − =).
  - Income elasticity of the gravity equation implied by the specification: 1/(1 − η).
- Estimation approach:
  - Cross-section regression for each year (recursive yearly estimation of coefficients).
  - Rationale: avoids pooling biases and issues with single-year dummies; alternative recursive methods assume coefficients are drawn from same distribution which may not hold.

### Key Estimated Coefficients and Time Trends
- Economic mass (income elasticity of trade):
  - Estimated coefficient hovers around 0.8 and is statistically significant for all years.
  - Table 1 selected-year values for Economic mass: 0.86**, 0.81**, 0.68**, 0.68**, 0.72**, 0.80**, 0.72**, 0.68**, 0.77** (for years 1978, 1980, 1983, 1986, 1989, 1992, 1995, 1998, 2001 respectively).
  - From this estimate infer a value of −0.25 for parameter η in equations (3) and (4) (implying net negative externalities).
- Distance:
  - Distance coefficient fell (in absolute value) from −0.57 (average of 1978–79) to −0.51 (average of 2000–01).
  - Table 1 selected-year Distance values: −0.58**, −0.57**, −0.55**, −0.51**, −0.43**, −0.44**, −0.42**, −0.49**, −0.52**.
  - Comparative static: with estimated distance coefficient −0.51 and assumption that distance(A,C) = 3 × distance(A,B), trade(A,C) ≈ 57 percent of trade(A,B).
- Adjacency:
  - Coefficient jumped in 1990 to a higher plateau.
  - Table 1 Adjacency values: 0.58**, 0.66**, 0.69**, 0.58**, 0.68**, 1.23**, 1.28**, 1.12**, 1.04**.
- Language (shared language):
  - Exhibits high volatility around a decreasing trend; small positive value in 2001.
  - Table 1 Language values: 0.39**, 0.21**, 0.35**, 0.51**, 0.46**, 0.15**, 0.12**, 0.12**, 0.04**.
- Population controls:
  - Population - Export Country values: 0.14**, 0.26**, 0.35**, 0.35**, 0.41**, 0.32**, 0.39**, 0.51**, 0.36**.
  - Population - Import Country values: 0.20**, 0.32**, 0.51**, 0.52**, 0.44**, 0.38**, 0.49**, 0.59**, 0.52**.

### Trading Blocs — Findings by Bloc
- General note:
  - Four Western Hemisphere trading blocs analyzed: MERCOSUR, NAFTA, CACM, Andean Community.
  - Other large trading blocs outside region were also considered; inclusion does not change results significantly.

- MERCOSUR (Argentina, Brazil, Paraguay, Uruguay; founded 1991 but bilateral protocols earlier):
  - Level of Integration (Table 1 top panel MERCOSUR): 0.09, 0.16, 0.29, 1.18*, 1.77**, 1.29**, 1.72**, 2.08**, 1.57** (1978–2001 selected years).
  - Bloc's propensity to Import: 0.33*, 0.24, 0.45, 0.17, −0.74*, −0.76**, −0.84**, −1.00**, −0.60**.
  - Bloc's propensity to Export: 0.08, −0.68**, 0.56*, 0.23, −0.56*, −0.64**, −1.15**, −1.26**, −0.56**.
  - Interpretation:
    - Integration increased since early 1980s and leveled off since mid-1990s; decline since 1998 possibly due to macro shocks (e.g., Brazilian real devaluation January 1999).
    - Propensity to import decreased sharply after early 1980s (closing doors to rest of world).
    - Propensity to export increased sharply in 1980s (debt-crisis-related export increases).
    - Evidence: integration and trade diversion went hand in hand in MERCOSUR.

- NAFTA (Canada, United States, Mexico; Mexico added 1994):
  - Level of Integration values: −0.23*, 0.06, −0.12, −0.32**, −0.06, −0.16*, 0.07, 0.28**, 0.37**.
  - Bloc's propensity to Import: 0.29, 0.09, 1.07*, 1.32**, 0.57, 0.07, 0.04, −0.25*, −0.35**.
  - Bloc's propensity to Export: 0.12, −0.39*, 0.66*, 0.87**, 0.07, −0.09, −0.09, −0.15, −0.22.
  - Interpretation:
    - Level of integration below zero in 1994 before NAFTA; increased after mid-1980s with Mexico’s liberalization; temporary reversal after 1990 pre-NAFTA doubts; upward trend resumes after 1994.
    - Weak indication of trade diversion in 1990s versus 1980s; stronger evidence of integration particularly after 1997.

- CACM (Central American Common Market; founded 1961):
  - Level of Integration values: 3.33*, 3.50*, 2.68, 1.14, 1.67, 1.97, 2.23, 1.53, 1.44.
  - Bloc's propensity to Import: 0.13, 0.06, −1.72**, −2.41**, −2.17**, −2.11**, −1.84**, −1.55**, −1.65**.
  - Bloc's propensity to Export: 0.73, 0.42, 0.75, 0.73, 0.35, 0.42, −0.05, 0.62, 1.05**.
  - Interpretation:
    - Level of integration flat and not significantly different from zero during last two decades.
    - Propensity to export flat and not significantly different from zero.
    - Propensity to import dropped in early 1980s and increased in late 1980s.
    - Evidence points to trade diversion without further integration.

- Andean Community:
  - Level of Integration values: −0.24, −0.12, 0.33, 0.45, 0.83, 1.08, 1.91*, 1.86*, 1.58*.
  - Bloc's propensity to Import: 1.27**, 1.05**, 1.45**, 0.41**, 0.64, 0.78**, 0.23, 0.03, −0.04.
  - Bloc's propensity to Export: 0.94**, 0.59**, 1.08**, 0.85**, 0.75**, 0.74**, 0.19, 0.05, 0.54**.
  - Interpretation:
    - Level of integration increased around November 1990 revival but remained insignificant until 1994; increase appears one-time.
    - Propensities to export and import decreased around revival time, indicating some trade diversion; overall evidence points to more trade diversion than integration.

### Table 1 Summary Statistics and Significance
- Number of observations by selected years: 3,909; 3,848; 3,974; 3,991; 4,005; 4,010; 4,014; 4,017; 4,017.
- Adjusted R2 values for selected years: 0.95, 0.93, 0.94, 0.97, 0.96, 0.95, 0.96, 0.97, 0.97.
- Significance notation: * and ** indicate coefficients significantly different from zero at the 5 and 1 percent level, respectively.

### Economic Interpretation of Externalities and Bilateral Resistance (Appendix II)
- Trade cost factor depends on distance and volume of trade; more exports can either reduce the trade cost factor if 0 < η < 1 (positive externalities) or increase it if −1 < η < 0 (negative externalities).
- Income elasticity implied by specification: 1/(1 − η).
  - Larger than one if positive externalities dominate.
  - Smaller than one (but larger than zero) if negative externalities dominate.
  - Equal to one if positive and negative externalities offset.
- Cost of bilateral resistance Cij defined as proportion of trade lost because of bilateral frictions relative to relative economic mass:
  - Cij = (Xij^w − Xij) × (GDPi GDPj / GDPw)^{-1} (expressed in source notation).
  - Derived expression (from substitution) shows Cij depends positively on distance (∂C/∂d > 0) and its sensitivity to world GDP and parameters σ and η can be positive or negative depending on parameter signs and ranges.

### Data and Sample
- Sample: 64 industrial and developing economies, period 1978–2001.
- Data sources:
  - Nominal GDP, GDP per capita, population: IMF World Economic Outlook Database.
  - Bilateral trade (exports from country i to country j): IMF Direction of Trade Database.
  - Trading bloc definitions: WTO (2000).
  - Language and adjacency variables: provided by CSTB.
  - Distance (between capitals): Centre d'Etudes Prospectives et d'Informations Internationales (www.cepii.fr/anglaisgraph/bdd/distances.htm).
- Table 2 lists the 64 economies (alphabetical listing as provided in source).

### Conclusions (Section V)
- Distance and shared language have become somewhat less relevant over time but remain important; adjacency became more important in early 1990s, coinciding with surge in RTAs.
- Estimated economic mass coefficient ~0.8 supports the presence of net negative externalities (η inferred ≈ −0.25), implying more exports generate net frictions that increase the trade cost factor and slow trade expansion.
- Channels for negative externalities to operate include intrusive regulation, infrastructure and institutional restrictions (bottlenecks), external pecuniary and technical diseconomies, trade financing constraints, local bias, or special interest politics — identification of channels is a promising avenue for policy-relevant research.
- Bloc-specific policy-relevant conclusions:
  - MERCOSUR: integration accompanied by trade diversion; reflects simultaneous liberalization and RTA formation.
  - NAFTA: stronger evidence of integration (particularly after 1997) and weak evidence of trade diversion.
  - CACM and Andean Community: evidence points to more trade diversion than integration.

*Source: Own calculation; dependent variable: trade; 64 economies, 1978–2001.*

### REFERENCES

### _wp04109 - REFERENCES

### References
- Anderson, James E. 1999, “Why do Nations Trade (So Little)?” Mimeograph (August), Boston College.

- Anderson, James E., and Eric Van Wincoop, 2003, “Gravity with Gravitas: A Solution to the Border Puzzle,” American Economic Review, Vol. 92 (March), pp. 170–92.

- Bernard, Andrew B., and Joachim Wagner, 1998, “Export Entry and Exit by German Firms,” National Bureau of Economic Research Working Paper Series No. 6538 (Cambridge, Massachusetts: National Bureau of Economic Research).

- Coe, David T., Arvind Subramanian, Natalia T. Tamirisa, and Tikhil Bhavnani, 2002, “The Missing Globalization Puzzle,” IMF Working Paper WP/02/171 (Washington: International Monetary Fund).

- Dell’Ariccia, Giovanni, 1999, “Exchange Rate Fluctuations and Trade Flows: Evidence from the European Union,” IMF Staff Papers, 46, No. 3 (September–December), pp. 315–34.

- Edwards, Sebastian, 1995, Crisis and Reform in Latin America: From Despair to Hope (Washington: World Bank and Oxford University Press).

- Ethier, Wilfred J., 1998, “Regionalism in a Multilateral World,” Journal of Political Economy, Vol. 106, No. 6 (December), pp. 1214–45.

- Frankel, Jeffrey A. (with Ernesto Stein and Shang Jin Wei), 1997, Regional Trading Blocs in the World Economic System (Washington: Institute for International Economics).

- ———, and Shang-Jin Wei, 1993, “Is There a Currency Bloc in the Pacific?” CIDER Working Paper No. C93-025 (Berkeley, California: University of California, Center for International and Development Economics Research).

- Glick, Reuven, 1998, “Contagion and Trade: Why Are Currency Crises Regional?” National Bureau of Economic Research Working Paper No. 6806 (Cambridge, Massachusetts: National Bureau of Economic Research).

- Krueger, Anne O., 1999, “Trade Creation and Trade Diversion Under NAFTA,” National Bureau of Economic Research Working Paper No. 7429 (Cambridge, Massachusetts: National Bureau of Economic Research).

- Obstfeld, Maurice, and Kenneth Rogoff, 2000, “The Six Major Puzzles in International Macroeconomics: Is There a Common Cause?” National Bureau of Economic Research Working Paper No. 7777 (Cambridge, Massachusetts: National Bureau of Economic Research).

- Panagariya, Arvind, 1999, “The Regionalism Debate: An Overview,” World Economy, Vol. 4 (June), pp. 477–511.

- Roberts, Mark J., and James R. Tybout, 1997, “The Decision to Export in Colombia: An Empirical Model of Entry with Sunk Costs,” American Economic Review, Vol. 87 (September), pp. 545–64.

- Soloaga, Isidro, and L. Alan Winters, 1999, “Regionalism in the Nineties: What Effect on Trade?” Centre for Economic Policy Research Discussion Paper Series No. 2183 (London: Centre for Economic Policy Research).

- Summers, Lawrence, 1993, “Regionalism and the World Trading System,” In Policy Implications of Trade and Currency, Proceedings of a Symposium Sponsored by the Federal Reserve Bank of Kansas City, Jackson Hole, Wyoming (Kansas City, Missouri: Federal Reserve Bank of Kansas City).

- World Bank, 2000, “Trade Blocs and Beyond: Political Dreams and Practical Decisions,” Policy Research Report (Washington).

- World Trade Organization, 2000, “Mapping of Regional Trade Agreements” (October).

- Yeats, Alexander J., 1997, “Does MERCOSUR’S Trade Performance Raise Concerns About the Effects of Regional Trade Arrangements?” Policy Research Working Paper 1729 (Washington: World Bank).

- 23 -

*Source: _wp04109 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04109.pdf_