## 1. Preferred Trading Arrangements

## Source details

**Canonical URL:** [1. Preferred Trading Arrangements](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0866.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0866.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0866.pdf.json)

---

### Introduction: research question, data, and empirical setup
- Purpose: Reexamine Trade Creation and Trade Diversion effects of Preferential Trading Arrangements (PTAs) using Bayesian Model Averaging (BMA) to explicitly incorporate model uncertainty; revisit the 12 major PTAs examined by Ghosh and Yamarik (2004) using the same baseline dataset and extend their approach.
- Preferred specification: Second specification that includes country-pair fixed effects to better account for “natural trading partner” effects.
- Dataset and sample:
  - Coverage: 186 countries; five year intervals from 1970-95.
  - Bilateral pairs: 3,420 bilateral trade pairs; 14,522 observations.
- PTA treatment variables:
  - 12 PTAijt dummies (PTAijt = 1 when both partners are members of the same PTA).
  - 12 PTAit dummies (PTAit = 1 when only one partner is a member of the PTA).
- Dependent variable: average bilateral trade in U.S. dollars deflated by the American GDP chain price index.
- Two core regression specifications:
  - Benchmark: pooled gravity specification without country-pair fixed effects (to replicate Ghosh and Yamarik, 2004).
  - Preferred: gravity specification augmented with country-pair fixed effects (time-invariant regressors absorbed; relies on time-series variation).

### Methodology: Bayesian Model Averaging (BMA) and inference
- Rationale for BMA:
  - Theory generates many competing candidate regressors; BMA provides a probability distribution over model space and parameter space and weights each model by posterior model probability.
  - BMA examines the entire model space (no arbitrary model-size restrictions) and weights models according to quality; contrasts with Extreme Bound Analysis (EBA) which weights models equally.
  - Theoretical properties: minimizes total error rate (Type I + Type II), minimizes mean squared error (MSE) of point estimates and predictions, and yields optimal predictive distributions relative to alternatives.
- Posterior inclusion probability (PIP) interpretation (Jeffreys / Kass and Raftery):
  - < 50%: evidence against an effect.
  - 50-75%: weak evidence for an effect.
  - 75-95%: positive evidence for an effect.
  - 95-99%: strong evidence for an effect.
  - > 99%: decisive evidence for an effect.
  - Regressor termed “effective” if PIP > 50%.
- Major candidate control categories:
  - Historical ties: Common Language, Common Colonizer, Colony.
  - Geography: Distance, Remoteness, Area, Border, Landlocked, Island.
  - Development/Factor Endowments: Per Capita GDP, Schooling, Population Density.
  - Economic Policy: Sachs-Warner Trade Openness, Currency Union, Floating FX Rate, FX Volatility, capital account openness.
- PTA theory uncertainty highlighted: Viner’s Trade Creation vs Trade Diversion, Krugman’s models, Frankel/Stein/Wei natural trading partner hypotheses, Deardorff and Stern on comparative advantage, Baier and Bergstrand extensions, Freund on Open Bloc Trade Creation, Bond and Syropoulos on market power external tariff effects.

### Main empirical findings (summary)
- Finding 1 — Benchmark (Specification 1, pooled OLS without country-pair fixed effects):
  - BMA finds PTAs exhibiting strong Trade Creation, Trade Diversion and Open Bloc effects for multiple PTAs.
  - Four PTAs show large Trade Creation effects in Specification 1: CACM, CARICOM, APEC, LAIA.
  - Ample evidence for Open Bloc Trade Creation: EU, APEC, EFTA, MERCOSUR.
  - Widespread evidence for Trade Diversion: LAIA, ANZCERTA, NAFTA, CARICOM, CACM, AP.
  - BMA and EBA select different models; BMA picks models not captured by EBA.
- Finding 2 — After including country-pair fixed effects (Specification 2):
  - Most previously large PTA effects are attributable to natural trading partnerships (time-invariant unobservables).
  - Trade Creation is muted and all but one Trade Diversion effect disappear once country-pair fixed effects are included.
  - Controlling for natural trading partner effects reveals strong Open Bloc Trade Creation for a number of major PTAs.
  - Only one robust within-PTA Trade Creation effect remains: EU (see numeric results below).
- Finding 3 — Interactions and breadth:
  - A PTA’s impact depends on its own Trade Creation/Diversion effects and on its trading partner’s PTA effects; Open Bloc interactions matter.
  - Empirical strategy must examine as many PTAs as possible to capture Open Bloc interactions correctly.
  - Inclusion of country-pair fixed effects adds 3,420 additional country-pair fixed effects in the second part of the analysis.

### Selected quantitative results (preserving reported numeric values)
- APEC_ijt:
  - Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.48***; Post. Std. Dev. 0.15; Post Mean in % 338%.
  - Spec 2 Post. Incl. Prob. 0.01; Post. Mean 0.14; Post. Std. Dev. 0.12; Post Mean in % 16%.
- CARICOM_ijt:
  - Spec 1 Post. Incl. Prob. 1.00; Post. Mean 2.08***; Post. Std. Dev. 0.41; Post Mean in % 702%.
  - Spec 2 Post. Incl. Prob. 0.00; Post. Mean 0.12; Post. Std. Dev. 0.71; Post Mean in % 13%.
- CACM_ijt:
  - Spec 1 Post. Incl. Prob. 1.00; Post. Mean 2.25***; Post. Std. Dev. 0.23; Post Mean in % 848%.
  - Spec 2 Post. Incl. Prob. 0% (no variation).
- EU_ijt:
  - Spec 1 Post. Incl. Prob. 0.00; Post. Mean 0.03; Post. Std. Dev. 0.14; Post Mean in % 3%.
  - Spec 2 Post. Incl. Prob. 1.00; Post. Mean 0.66***; Post. Std. Dev. 0.17; Post Mean in % 93%.
- NAFTA_it (Trade Diversion/Open Bloc):
  - Spec 1 Post. Incl. Prob. 1.00; Post. Mean -0.63***; Post. Std. Dev. 0.10; Post Mean in % -47%.
  - Spec 2 Post. Incl. Prob. 0.99; Post. Mean -0.31***; Post. Std. Dev. 0.08; Post Mean in % -26%.
- CU_ijt (Currency Union):
  - Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.40***; Post. Std. Dev. 0.29; Post Mean in % 307%.
  - Spec 2 Post. Incl. Prob. 0.01; Post. Mean -0.64; Post. Std. Dev. 0.67; Post Mean in % -47%.
- Other control-variable findings:
  - abs(gdp_DIFF): Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.18***; Post. Std. Dev. 0.02. Spec 2 Post. Incl. Prob. 1.00; Post. Mean -0.31***; Post. Std. Dev. 0.05.
  - abs(DENS_DIFF): Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.23***; Post. Std. Dev. 0.01. Spec 2 Post. Incl. Prob. 0.75; Post. Mean 0.25**; Post. Std. Dev. 0.09.
  - BORDER_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.53***; Post. Std. Dev. 0.10; Post Mean in % 69%.
  - REMOTE_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 342***; Post. Std. Dev. 39.79; Post Mean in % 342%.
  - LANDLOCK_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean -0.42***; Post. Std. Dev. 0.04; Post Mean in % -34%.
  - COMLANG_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.47***; Post. Std. Dev. 0.05; Post Mean in % 60%.
  - COLONY_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.44***; Post. Std. Dev. 0.12; Post Mean in % 320%.

### Net PTA and cross-PTA interaction results (selected entries from Table 4 and Table 4a rules)
- Selected net effects after country-pair fixed effects (Table 4 selected cells):
  - NAFTA with NAFTA: 0%
  - ANZCERTA on NAFTA: -26%
  - AFTA on ANZCERTA: 49%
  - APEC on NAFTA: -26%
  - EU on APEC: 87%
  - EU within EU: 93%
  - EFTA with APEC: 142%
  - Rest of the World diagonal: 0%
- Note: Table 4a provides algebraic calculation rules for net trade effects combining RTA_i and RTA_ij coefficients; net effects account for overlapping PTA membership and membership changes.

### Additional empirical findings: exchange rate policy, factor endowments, and robustness
- Exchange rate policy variables:
  - Many exchange rate policy variables not significant in either specification.
  - Currency Union: Spec 1 shows 307% increase in trade (consistent with Rose (2000)); effect evaporates in Spec 2 once country-pair fixed effects and PTA variables are included.
- Factor endowments and hypotheses:
  - Specification 1 supports Heckscher-Ohlin: differences in per capita GDP and population densities strongly associated with greater trade flows; magnitudes indicate bilateral trade increases by roughly 20% when one trading partner is twice as dense or wealthy as the other.
  - Specification 2: density differences still impact trade positively; per capita GDP differences change sign supporting the Linder hypothesis.
  - Conclusion: data provide support for both Heckscher-Ohlin and Linder hypotheses simultaneously; pair fixed effects likely pick up slow-moving factor endowment differences.
- Education differences: not robustly related to trade in either specification.

### Interpretation, methodological contrasts, and implications for empirical practice
- Reconciliation with prior literature:
  - The implausibly large APEC coefficient commonly observed is initially confirmed by BMA in the benchmark, but after controlling for country-pair fixed effects APEC no longer exhibits strong accession effects.
  - European PTAs reveal strong impact only after controlling for country-pair fixed effects, indicating EU countries naturally under-trade relative to standard gravity predictions.
  - NAFTA is identified as the only Trade Diverting PTA; diversion occurs with a select subset of countries and can be muted or reversed when interacting PTAs exert strong Open Bloc effects.
- Methodological contrasts with Ghosh and Yamarik (2004):
  - Ghosh and Yamarik used Extreme Bound Analysis (EBA) and found no robust PTA effects at EBA extreme bounds; BMA’s unrestricted model search and posterior-probability weighting yield different conclusions and superior predictive performance under Kullback-Leibler information divergence.
  - Inclusion of country-pair fixed effects (3,420 additional fixed effects) materially alters inference about natural trading partner effects.
- Empirical practice implications:
  - PTA analyses must separate Trade Creation and Trade Diversion and consider an exhaustive set of PTAs.
  - Any PTA’s trade influence depends on its own direct effect and interactions with other PTAs’ Open Bloc effects.

### Conclusion (core takeaways)
- BMA that accounts for model uncertainty yields different empirical conclusions than EBA.
- PTA-internal Trade Creation is significantly positive under BMA in pooled specifications, but large Trade Creation often reflects natural trading partners rather than causal PTA effects.
- Once country-pair fixed effects are included:
  - Several strong Open Bloc effects remain.
  - Only one robust within-PTA Trade Creation effect persists: EU (93% increase within EU).
  - Trade Diversion is observed in NAFTA, but net bilateral effects can be altered by strong Open Bloc gains from other PTAs.
- Methodological recommendation: properly account for model uncertainty and bilateral unobserved heterogeneity (country-pair fixed effects) and analyze interactions across a comprehensive set of PTAs.

*Source: _wp0866 - 1. Preferred Trading Arrangements (PDF), pages 19–24.*

### 1. Preferred Trading Arrangements ......................................................................................

### 1. Preferred Trading Arrangements

### Section contents and structure
- Title: "1. Preferred Trading Arrangements" (page 19)
- Related headings appearing in the source's adjacent table of contents:
  - "2. Common Gravity Model Controls, their Estimated Relationship with Bilateral Trade in Past Studies and Variable Description" (page 20)
  - "3. PTA Trade Creation and Trade Diversion" (page 22)
  - "4. Trade Creation and Diversion among Trading Partners" (page 23)
  - "4a. Calculation of Trade Flows among and Within PTAs" (page 23)
  - "References" (page 24)

### Role in the document
- Appears as the first main numbered section in the document's body following the table of contents.
- Positioned immediately before methodological and empirical sections (sections 2–4 and 4a) and the References (page 24).

*Source: _wp0866 - 1. Preferred Trading Arrangements (PDF), pages 19–24.*

### INTRODUCTION

### _wp0866 - INTRODUCTION

### Research question and approach
- Purpose: Reexamine Trade Creation and Trade Diversion effects of Preferential Trading Arrangements (PTAs) using Bayesian Model Averaging (BMA) to explicitly incorporate model uncertainty.
- Comparison benchmark: Revisit the 12 major PTAs examined by Ghosh and Yamarik (2004) using the same baseline dataset and extend their approach.
- Preferred specification: Second specification that includes country-pair fixed effects (introduced to better account for “natural trading partner” effects).

### Data and empirical setup
- Dataset coverage: 186 countries; five year intervals from 1970-95.
- Bilateral pairs and observations: 3,420 bilateral trade pairs; 14,522 observations.
- PTA treatment variables: 12 PTAijt dummies and 12 PTAit dummies (PTAijt = 1 when both partners are members of the same PTA; PTAit = 1 when only one partner is a member of the PTA).
- Dependent variable: average bilateral trade in U.S. dollars deflated by the American GDP chain price index.
- Two core regression specifications:
  - Benchmark (to replicate Ghosh and Yamarik, 2004): pooled gravity specification without country-pair fixed effects.
  - Preferred: gravity specification augmented with country-pair fixed effects to capture time-invariant unobservables (reduced matrix of controls Zijt~ since time-invariant regressors are absorbed).

### Methodology: Bayesian Model Averaging (BMA)
- Rationale: Theoretical PTA literature generates a large and competing set of candidate regressors; BMA provides a probability distribution over model space and parameter space and weights each model by posterior model probability.
- BMA advantages highlighted:
  - Examines the entire model space (no arbitrary restrictions on model size).
  - Weighs models according to quality (posterior model probabilities), unlike Extreme Bound Analysis (EBA) which weights models equally.
  - Theoretical properties: minimizes total error rate (Type I + Type II), minimizes mean squared error (MSE) of point estimates and predictions, and yields optimal predictive distributions relative to alternatives.
- Posterior inclusion probability (PIP) interpretation and thresholds (Jefferies / Kass and Raftery):
  - Posterior probabilities < 50%: evidence against an effect.
  - 50-75%: weak evidence for an effect.
  - 75-95%: positive evidence for an effect.
  - 95-99%: strong evidence for an effect.
  - > 99%: decisive evidence for an effect.
- Inference: A regressor is termed “effective” if its posterior inclusion probability exceeds 50%.

### Theories, controls, and model uncertainty addressed
- Major control categories included in candidate regressor set:
  - Historical ties (Common Language, Common Colonizer, Colony).
  - Geography (Distance, Remoteness, Area, Border, Landlocked, Island).
  - Development/Factor Endowments (Per Capita GDP, Schooling, Population Density).
  - Economic Policy (Sachs-Warner Trade Openness, Currency Union, Floating FX Rate, FX Volatility, capital account openness).
- PTA theory uncertainty: competing theoretical predictions (e.g., Viner’s Trade Creation vs Trade Diversion; Krugman’s two models with different implications; Frankel, Stein and Wei’s natural trading partner hypotheses; Deardorff and Stern on comparative advantage; Baier and Bergstrand extensions; Freund on Open Bloc Trade Creation; Bond and Syropoulos on market power external tariff effects).
- Empirical implication: model averaging needed because inclusion/exclusion of controls substantially changes coefficient estimates in gravity regressions.

### Main empirical findings (threefold)
- Finding 1 (benchmark/specification without country-pair fixed effects replicating Ghosh and Yamarik):
  - PTAs exhibit strong Trade Creation, Trade Diversion and Open Bloc effects for 14 PTAs (as reported in the benchmark specification).
  - Contrast: Ghosh and Yamarik (2004) find no evidence of Trade Creation or Diversion for any PTAs under their Extreme Bounds Analysis; when they relax extreme bounds they find trade creation in CACM, CARICOM, MERCOSUR and APEC. Three of those four are among the PTAs identified here as showing trade creation in the benchmark.
  - BMA and EBA select different models; BMA picks models not captured by EBA.

- Finding 2 (after including country-pair fixed effects):
  - Most previously large PTA effects are attributable to natural trading partnerships (time-invariant unobservables).
  - Trade Creation is muted and all but one Trade Diversion effect disappear once country-pair fixed effects are included.
  - Controlling for natural trading partner effects reveals strong Open Bloc Trade Creation for a number of major PTAs.
  - Interpretation: some PTAs may have been formed because of these unobservables (i.e., countries that naturally trade more may be more likely to form PTAs), but pooled OLS would incorrectly attribute resulting trade changes to PTA formation.

- Finding 3 (interaction and breadth of analysis):
  - The impact of a PTA on bilateral trade depends on its own Trade Creation and Diversion effects and on its trading partner’s PTA effects (interaction across PTAs).
  - Appropriate empirical strategy must examine as many PTAs as possible to capture Open Bloc interactions correctly.
  - Dataset extension: inclusion of country-pair fixed effects adds 3,420 additional country-pair fixed effects in the second part of the analysis.

- Additional notable results and reconciliations with literature:
  - The implausibly large APEC coefficient commonly observed is initially confirmed by BMA in the benchmark, but after controlling for country-pair fixed effects the consultative APEC no longer exhibits strong effects on trade flows.
  - European PTAs reveal strong impact only after controlling for country-pair fixed effects, showing that EU countries naturally under-trade relative to the standard gravity model predictions (negative residuals for European country pairs are a recurrent feature of the literature).
  - NAFTA is identified as the only Trade Diverting PTA, and this diversion occurs only with a select subset of countries; due to strong Open Bloc effects among Asian and European countries, their trade with NAFTA actually increased over the sample period.

### Methodological contrasts with prior work
- Ghosh and Yamarik (2004) used Extreme Bound Analysis (EBA) and found no robust PTA effects at EBA extreme bounds; they detect limited trade creation under relaxed bounds (CACM, CARICOM, MERCOSUR, APEC).
- Key differences leading to divergent results:
  - BMA performs an unrestricted search over model space; EBA examines a limited model space and weights models equally.
  - BMA’s model weighting by posterior probability gives more efficient inference and avoids over-stringency of EBA.
  - Dataset expansion to include 3,420 additional country-pair fixed effects alters inference about natural trading partner effects relative to Ghosh and Yamarik (2004).

### Structure of the remainder of the paper (as presented)
- Section 2: discussion of BMA methodology and framework.
- Section 3: description of data.
- Section 4: results and discussion.
- Section 5: conclusion.

*Source: _wp0866 - INTRODUCTION*

### 1995. In addition to the basic gravity and trade agreement variables, 16 controls variables

### _wp0866 - 1995. In addition to the basic gravity and trade agreement variables, 16 controls variables

### IV. RESULTS — A. PTA Trade Creation/Diversion: Differences due to Methodologies
- Objective: contrast BMA results with Ghosh and Yamarik (2004) who found no robust PTA effects.
- Methodology:
  - Bayesian Model Averaging (BMA) applied to Specification 1 (logs) with reported implied percentage changes.
  - Coefficients on PTA_ijt interpreted as percentage changes in trade volume between two PTA members relative to non-members (positive = Trade Creation).
  - Coefficients on PTA_it reflect percentage changes when only one partner belongs to a PTA (positive = Open Bloc Trade Creation; negative = Trade Diversion).
- Key findings:
  - BMA provides clear evidence that PTAs generate Trade Creation, Open Bloc Trade Creation, and Trade Diversion, contradicting Ghosh and Yamarik (2004).
  - Four PTAs show large Trade Creation effects in Specification 1: CACM, CARICOM, APEC, LAIA.
  - Ample evidence for Open Bloc Trade Creation: EU, APEC, EFTA, MERCOSUR.
  - Widespread evidence for Trade Diversion: LAIA, ANZCERTA, NAFTA, CARICOM, CACM, AP.
  - Most theory-based controls (except Exchange Rate Regimes and Volatility, Schooling Differences, and the Island Dummy) are highly significant with familiar signs and magnitudes.
- Interpretation: expanding the search space to the entire model set and using theory-based model averaging changes conclusions relative to Extreme Bound Analysis; BMA yields superior predictive performance under Kullback-Leibler information divergence.

### IV. RESULTS — B. PTA Trade Creation/Diversion: Accounting for Natural Trading Partners
- Rationale: Specification 1 may be contaminated by omitted variable bias because control variables Z_ijt may not fully account for "natural trading partner effects".
- Specification 2: introduces country-pair fixed effects to capture unobserved time-invariant bilateral heterogeneity; relies on time-series variation (effects of forming or joining PTAs).
- Comparative implications:
  - Substantial coefficient bias in Specification 1 when natural trading partner effects are not fully addressed.
  - The 16 geography/history/culture regressors from previous studies do not fully account for systematic bilateral heterogeneity.
- Effects observed after country-pair fixed effects (Specification 2):
  - PTAs still have important effects but reduced and mainly limited to Trade Diversion/Open Bloc effects.
  - Massive Trade Creation by CARICOM and APEC in Specification 1 is entirely negated in Specification 2.
  - EU exhibits strong Trade Creation in Specification 2, though Specification 1 showed no such evidence.
  - Of 9 Trade Diversion / Open Bloc effects in Specification 1, only three remain significant in Specification 2.
  - Most Trade Diversion coefficients lose significance once country-pair fixed effects are included.
- Interpretation examples:
  - EU: specification 2 purges systematic under-trading relative to gravity predictions and isolates true EU accession effect.
  - APEC: positive coefficient in Specification 1 driven by over-trading of geographically distant APEC countries (Linnemann, 1966); Specification 2 shows no considerable APEC accession effect.

### IV. RESULTS — C. Quantifying PTAs’ Open Bloc and Trade Creation Effects
- Approach: detailed evaluation of each PTA coefficient; net effects require accounting for overlapping PTA membership and membership changes.
- Table 4 implications:
  - On-diagonal (within-PTA Trade Creation) is weak after country-pair fixed effects; only exception: EU with 93% increase in bilateral trade among member countries.
  - Off-diagonals show Trade Diversion / Open Bloc Trade Creation: cross-PTA trade is often Open Bloc Trade Creating and amplified by interactions between PTAs’ Open Bloc effects.
  - NAFTA Trade Diversion is significantly muted or overturned when NAFTA countries trade with other PTAs because of other PTAs’ strong Open Bloc effects.
  - No Open Bloc effects emanate from the EU in trade with non-member countries, but trade with almost every other PTA increased due to strong Open Bloc effects of those PTAs.
  - Strong trade gains for Pacific Rim and Asian countries’ membership in AFTA, APEC and ANZCERTA are observed without inflated Trade Creation coefficients typical in prior literature.

- Select exact quantitative results from Table 3 (Specification 1 and Specification 2 excerpts shown as reported):
  - APEC_ijt: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.48***; Post. Std. Dev. 0.15; Post Mean in % 338%. Spec 2 Post. Incl. Prob. 0.01; Post. Mean 0.14; Post. Std. Dev. 0.12; Post Mean in % 16%.
  - CARICOM_ijt: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 2.08***; Post. Std. Dev. 0.41; Post Mean in % 702%. Spec 2 Post. Incl. Prob. 0.00; Post. Mean 0.12; Post. Std. Dev. 0.71; Post Mean in % 13%.
  - CACM_ijt: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 2.25***; Post. Std. Dev. 0.23; Post Mean in % 848%. Spec 2 Post. Incl. Prob. 0% (no variation).
  - EU_ijt: Spec 1 Post. Incl. Prob. 0.00; Post. Mean 0.03; Post. Std. Dev. 0.14; Post Mean in % 3%. Spec 2 Post. Incl. Prob. 1.00; Post. Mean 0.66***; Post. Std. Dev. 0.17; Post Mean in % 93%.
  - NAFTA_it (Trade Diversion/Open Bloc): Spec 1 Post. Incl. Prob. 1.00; Post. Mean -0.63***; Post. Std. Dev. 0.10; Post Mean in % -47%. Spec 2 Post. Incl. Prob. 0.99; Post. Mean -0.31***; Post. Std. Dev. 0.08; Post Mean in % -26%.
  - CU_ijt (Currency Union): Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.40***; Post. Std. Dev. 0.29; Post Mean in % 307%. Spec 2 Post. Incl. Prob. 0.01; Post. Mean -0.64; Post. Std. Dev. 0.67; Post Mean in % -47%.
  - abs(gdp_DIFF): Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.18***; Post. Std. Dev. 0.02. Spec 2 Post. Incl. Prob. 1.00; Post. Mean -0.31***; Post. Std. Dev. 0.05.
  - abs(DENS_DIFF): Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.23***; Post. Std. Dev. 0.01. Spec 2 Post. Incl. Prob. 0.75; Post. Mean 0.25**; Post. Std. Dev. 0.09.
  - BORDER_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.53***; Post. Std. Dev. 0.10; Post Mean in % 69%.
  - REMOTE_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 342***; Post. Std. Dev. 39.79; Post Mean in % 342%.
  - LANDLOCK_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean -0.42***; Post. Std. Dev. 0.04; Post Mean in % -34%.
  - COMLANG_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 0.47***; Post. Std. Dev. 0.05; Post Mean in % 60%.
  - COLONY_ij: Spec 1 Post. Incl. Prob. 1.00; Post. Mean 1.44***; Post. Std. Dev. 0.12; Post Mean in % 320%.

- Table 4 (net effects after country-pair fixed effects) — selected cells:
  - NAFTA with NAFTA: 0%
  - ANZCERTA on NAFTA: -26%
  - AFTA on ANZCERTA: 49%
  - APEC on NAFTA: -26%
  - EU on APEC: 87%
  - EU within EU: 93%
  - EFTA with APEC: 142%
  - Rest of the World diagonal: 0%

- Note: Table 4a provides the algebraic calculation rules for net trade effects combining RTA_i and RTA_ij coefficients.

### IV. RESULTS — D. Beyond PTAs: Trade Creation/Diversion and Economic Policy/Factor Endowments
- Trade Openness: key variable in both specifications.
- Exchange rate policy variables:
  - Many exchange rate policy variables not significant in either specification.
  - Currency Union variable: Spec 1 shows extraordinarily large 307% increase in trade (consistent with Rose (2000)); effect evaporates in Specification 2 once country-pair fixed effects and PTA variables are included.
  - Interpretation: natural trade partners may synchronize business cycles and have greater incentives to join currency unions (Frankel and Rose, 1998), creating a spurious effect in Spec 1.
- Factor endowments:
  - Specification 1 supports Heckscher-Ohlin: differences in per capita GDP and population densities strongly associated with greater trade flows. Magnitudes indicate bilateral trade increases by roughly 20% when one trading partner is twice as dense or wealthy as the other.
  - Specification 2: density differences still impact trade positively; per capita GDP differences change sign supporting Linder hypothesis.
  - Conclusion: data provide support for both Heckscher-Ohlin and Linder hypotheses simultaneously; pair fixed effects likely pick up slow-moving factor endowment differences.
- Education differences: not robustly related to trade in either specification.

### V. CONCLUSION
- BMA incorporates theory uncertainty and yields different empirical conclusions than Extreme Bound Analysis (Ghosh and Yamarik, 2004).
- Summary of core conclusions:
  - PTA-internal Trade Creation is significantly positive under BMA, but large Trade Creation often reflects natural trading partners rather than PTA effects.
  - Once country-pair fixed effects are included, several strong Open Bloc effects remain but only one robust within-PTA Trade Creation effect (EU).
  - Trade Diversion observed in one PTA (NAFTA) after controls; strong evidence for significant Open Bloc Trade Creation among PTAs in Europe and Asia.
- Methodological implications:
  - PTA analyses must separate Trade Creation and Trade Diversion and consider an exhaustive set of PTAs.
  - Any PTA’s trade influence depends on its own direct effect and interactions with other PTAs’ Open Bloc effects (e.g., NAFTA’s Trade Diversion can turn into Trade Creation when considering bilateral flows with AFTA and EFTA).

*Source: _wp0866 - 1995. In addition to the basic gravity and trade agreement variables, 16 controls variables.*

### References

### _wp0866 - References

### Gravity models and determinants of bilateral trade
- Anderson, J., van Wincoop, E., 2003. “Gravity with Gravitas: a solution to the border puzzle”, American Economic Review, 93 (1), pp. 170-192.
- Baier, S.L., Bergstrand, J.H., 2002. “On the endogeneity of international trade flows and free trade agreements”, Manuscript, August 2002.
- Baier, S.L., and Bergstrand, J.H., 2004. “Economic determinants of free trade agreements”, Journal of International Economics, 64, pp. 29-63.
- Baier, S.L., and Bergstrand, J.H., 2007. “Do free trade agreements actually increase members’ international trade?”, Journal of International Economics, 71, pp. 72-95.
- Bergstrand, J., 1985. “The gravity equation in international trade: some microeconomic foundation and empirical evidence”, The Review of Economics and Statistics, 67, pp. 474-481.
- Brun, J., Carrere, C., Guillaumont, P., de Melo, J., 2005. “Has Distance Died? Evidence from a Panel Gravity Model”, World Bank Economic Review, 19 (1), pp. 99-120.
- Carrere, C., 2006. “Revisiting the effects of regional trade agreements on trade flows with proper specification of the gravity model”, European Economic Review, 50, pp. 223-247.
- Egger, P., 2000. “A note on the proper econometric specification of the gravity equation”, Economics Letters, 66, pp. 25-31.
- Egger, P., and Pfaffermayr, M., 2003. “The proper panel econometric specification of the gravity equation: A three-way model with bilateral interaction effects”, Empirical Economics, 28, pp. 571-580.
- Pomoted methodological references addressing heterogeneity and fixed effects in gravity estimates:
  - Andrews, M., Schank, T., Upward, R., 2006. Practical fixed effects estimation methods for the three-way error components model, The Stata Journal, 6(4), pp. 461-481.
  - Cheng, I., Wall, H.J., 2005. “Controlling for Heterogeneity in Gravity Models of Trade and Integration”, Federal Reserve Bank of St. Louis Review, 87 (1), pp. 49-63.
  - Matyas, L., 1997. “Proper Econometric Specification of the Gravity Model”, World Economy, 20 (3), pp. 363-368.

### Preferential trade agreements, regionalism, and welfare effects
- Aitken, N.D., 1973. “The Effect of the EEC and EFTA on European Trade: A Temporal Cross-Section Analysis”, American Economic Review, 63 (5), pp. 881-892.
- Aitken, N.D., and Lowry, W., 1973. “A cross-sectional study of the effects of LASEAN and CACM on Latin American trade, Journal of Common Market Studies, 11, pp. 326-336.
- Baldwin, R., 1995a. “A domino theory of regionalism”, in: Baldwin, R., Haaparanta, P., Kiander, J., eds., Expanding membership of the European Union, Cambridge University Press, Cambridge, MA.
- Baldwin, R., 1995b. “The Eastern Enlargement of the European Union”, European Economic Review, 39 (3-4), pp. 474-481.
- Baldwin, R., 1997. “The Causes of Regionalism”, World Economy, 20 (7), pp. 865-888.
- Bhagwati, J., Panagariya, A., 1996. “Preferential Trading Areas and Multilateralism: Strangers, Friends or Foes?”, in: Bhagwati, J., Panagariya, A., eds., The Economics of Preferential Trade Agreements, AEI Press, Washington, D.C.
- Deardorff, A.V., Stern, R., 1994. “Multilateral Trade Negotiations and Preferential Trade Agreements”, in: Deardorff, A.V., Stern, R., eds., Analytical and Negotiating Issues in the Global Trading System, University of Michigan Press, Ann Arbor, MI.
- Frankel, J., 1997. Preferential Trading Blocs in the World Trading System, Institute for International Economics, Washington, D.C.
- Frankel, J., Stein, E., Wei, S.J., 1995. “Trading blocs and the Americas: the natural, the unnatural and the supernatural?” Journal of Development Economics, 47, pp. 61-95.
- Freund, C., 2000. “Different paths from free trade: the gains from regionalism”, The Quarterly Journal of Economics, 115, pp. 1317–1341.
- Ghosh, S., Yamarik, S., 2004. “Are preferential trade agreements trade creating? An application of extreme bounds analysis”, Journal of International Economics, 63, pp. 369-395.
- Levy, P.I., 1996. “Free Trade Agreements and Inter-Bloc Tariffs”, Mimeo, Yale University.
- Levy, P.I., 1997. “A Political-Economic Analysis of Free Trade Agreements”, American Economic Review, 87 (4), pp. 506-519.
- Magee, C., 2003. “Endogenous preferential trade agreements: an empirical analysis”, Contributions to Economic Analysis and Policy, 2 (1), Berkeley Electronic Press.
- Panagariya, A., 1999. “The Regionalism Debate: An Overview”, World Economy, 22 (4), pp. 280-301.
- Panagariya, A., 2000. “Preferential Trade Liberalization: The Traditional Theory and New Developments”, Journal of Economic Literature, 38 (2), pp. 287-331.
- Soloaga, I., Winters, L.A., 2001. “Preferentialism in the nineties: what effect on trade?” North American Journal of Economics and Finance, 12, pp. 1-29.
- Viner, J., 1950. The Customs Union Issue, Carnegie Endowment for International Peace, New York.

### Currency unions, exchange rates, and trade effects
- Brada, J., Mendez, J., 1988. “Exchange rate risk, exchange rate regime and the volume of international trade”, Kyklos, 41, pp. 263–280.
- Chen, Y.C., Rogoff, K., 2006. “Commodity Currencies and Exchange Rate Predictability: A Bayesian Model Averaging Approach”, Mimeo, University of Washington.
- Clark, P., Tamirisa, N., Wei, S.J., Sadikov, A., Zeng, L., 2004. “Exchange Rate Volatility and Trade Flows - Some New Evidence”, Working Paper, International Monetary Fund.
- Frankel, J., Rose, A.K., 1998. “The Endogeneity of the Optimum Currency Area Criteria”, Economic Journal, 108, pp. 1009-1025.
- Frankel, J., and Rose, A.K., 2002. “An estimate of the effect of common currencies on trade and income”, The Quarterly Journal of Economics, 117, pp. 437–466.
- Glick, R., Rose, A.K., 2002. “Does a currency union affect trade? The time-series evidence”, European Economic Review, 46, pp. 1125-1151.
- Rose, A.K., 2000. One money, one market? The Effects of Common Currencies on International Trade”, Economic Policy,15, pp. 7-46.
- Rose, A.K., and van Wincoop, E., 2001. “National Money as a Barrier to International Trade: The Real Case for a Currency Union”, American Economic Review, 91 (2), pp. 386-390.
- Tenreyro, S., Barro, R.J., 2002. “Economic effects of currency unions”, National Bureau of Economic Research Working Paper No. 9435.
- Wei, S.J., and Frankel, J., 1998. “Open Regionalism in World of Continental Trade Blocs”, IMF Staff Papers, 45 (3), pp. 440-453.
- Wei, S.J., and Zhang, Z., 2006. “Collateral Damage: Exchange Controls and International Trade”, Working Paper, International Monetary Fund.

### Trade theory, market structure, and political economy of trade
- Davis, D.R., 1995. “Intra-industry Trade: A Heckscher-Ohlin-Ricardo Approach”, Journal of International Economics, 39, pp. 201-226.
- Deardorff, A.V., 1998. “Determinants of bilateral trade: does gravity work in a classical world”, in: Frankel, J. (Ed.), The Preferentialization of the World Economy. University of Chicago Press, Chicago, pp. 7–22.
- Feenstra, R., 2003. Advanced International Trade: Theory and Evidence. Princeton University Press, New Jersey, USA.
- Feenstra, R.C., Markusen, J.R., Rose, A.K., 2001. “Using the Gravity Equation to Differentiate among Alternative Theories of Trade”, Canadian Journal of Economics, 34 (2), pp. 430-477.
- Helpman, E., Krugman, P.R., 1985. Market Structure and Foreign Trade, MIT Press, Cambridge, MA.
- Grossman, G.M., Helpman, E., 1991. Innovation and Growth in the Global Economy, Cambridge, MIT Press.
- Grossman, G.M., Helpman, E., 1995. “The Politics of Free-Trade Agreements”, American Economic Review, 85 (4), pp. 667-690.
- Krugman, P., 1991a. “Is bilateralism bad?”, in: Helpman, E., Razin, A., eds, International trade and trade policy, MIT Press, Cambridge, MA.
- Krugman, P., 1991b. “The move toward free trade zones”, in: Policy Implications of Trade and Currency Zones, A Symposium Sponsored by the Federal Reserve Bank of Kansas City, Jackson Hole, WY, Aug., pp. 7-42.
- Rivera-Batiz, L.A., Romer, P.M., 1991. “Economic Integration and Endogenous Growth”, Quarterly Journal of Economics, 106 (2), pp. 531-555.
- Splimbergo, A., Stein, E., 1996. “The Welfare Implications of Trading Blocs among Countries with Different Endowments”, National Bureau of Economic Research Working Paper No. 5472.
- Syropoulos, C., 1999. “Customs Unions and Comparative Advantage”, Oxford Economic Papers, 51 (2), pp. 239-266.
- Richardson, M., 1993. “Endogenous Protection and Trade Diversion”, Journal of International Economics, 34 (3-4), pp. 309-324.
- Kemp, M.C., Wan, H.Y., 1976. “An Elementary Proposition Concerning the Formation of Customs Unions”, Journal of International Economics, 6, pp. 95-97.
- Haveman, J.D., 1996. “Some Welfare Effects of Sequential Customs Union Formation”, Canadian Journal of Economics, 29 (4), pp. 941-958.
- Bond, E.W., Syropoulos, C., 1996. “The size of trading blocs, market power and world welfare effects”, Journal of International Economics, 40 (3), pp. 411-437.

### Econometric methods, model uncertainty, and Bayesian model averaging
- Leamer, E.E., 1978. Specification Searches, J. Wiley, New York, NY.
- Leamer, E.E., 1983. “Let's Take the Con Out of Econometrics”, American Economic Review, 73, pp. 31-43.
- Sala-i-Martin, X.G., 1997. “I Just Ran Two Million Regressions,” American Economic Review, Papers and Proceedings, 87, pp.178-183.
- Fernandez, C., Ley, E., Steel, M., 2001a. “Model Uncertainty in Cross-Country Growth Regressions,” Journal of Applied Econometrics, 16, pp. 563-576.
- Fernandez, C., Ley, E., Steel, M., 2001b. “Benchmark Priors for Bayesian Model Averaging,” Journal of Econometrics, 100, pp. 381-427.
- Raftery, A.E., Madigan, D., Hoeting, J.A., 1997. "Bayesian Model Averaging for Linear Regression Models," Journal of the American Statistical Association, 92, pp. 179-191.
- Raftery, A.E., Zheng, Y., 2002. Long-Run Performance of Bayesian Model Averaging, Journal of the American Statistical Association, 98, pp. 931-937.
- Raftery, A.E., Zheng, Y., 2003. “Discussion: Performance of Bayesian Model Averaging”, Journal of the American Statistical Association, 98 (464), Theory and Methods, pp. 931-938.
- Kass, R.E., Raftery, A.E., 1995. “Bayes Factors”, Journal of the American Statistical Association, 90 (430), pp. 773-795.
- Eicher, T.S., Papageorgiou, C., Raftery, A.E., 2007. “Evaluating the Impact of Priors in Economic Applications of Bayesian Model Averaging”, Working Paper, University of Washington.
- Eicher, T.S., Papageorgiou, C., Roehn, O., 2007. “Unravelling the Fortunes of the Fortunate: An Iterative Bayesian Model Averaging (IBMA) Approach”, Journal of Macroeconomics, forthcoming.
- Hjort, N.L., Claeskens, G., 2003. “Frequentist Model Average Estimators”, Journal of the American Statistical Association, 98 (464), pp. 879-899.
- Hansen, B.E., 2007. ”Least Squares Model Averaging," Econometrica, forthcoming.
- Wright, J.H., 2003. “Bayesian Model Averaging and Exchange Rate Forecasts”, Board of Governors of the Federal Reserve System Discussion Paper 779.
- Jeffries, H., 1961. Theory of Probability, 3rd Edition, Clarendon Press, Oxford.
- Greene, W.H., 2003. Econometric Analysis, 5th Edition, Prentice-Hall, Princeton, NJ.
- Hendry, D., 1995. Dynamic Econometrics, Oxford University Press, Oxford, United Kingdom.

### Regional case studies, Linder hypothesis, and special topics
- Frankel, J., 1992. “Is Japan Creating a Yen Bloc in East Asia and the Pacific?”, National Bureau of Economic Research Working Paper No. 4050.
- Frankel, J., and Wei, S.J., 1993. “Trading Blocs and Currency Blocs”, National Bureau of Economic Research Working Paper No. 4335.
- Frankel, J., and Wei, S.J., 1996. “ASEAN in a Regional Perspective”, Pacific Basin Working Paper No. PB96-02.
- Ponded studies on North-South and regional trade patterns:
  - Coe, D., Hoffmaister, A., 1999. “North-South trade: is Africa unusual”, Journal of African Economies, 8, pp. 228-256.
  - Montenegro, C., Soto, R., 1996. “How distorted is Cuba’s trade? Evidence and predictions from a gravity model”, Journal of International Trade and Economic Development, 5, pp. 45-70.
  - Pollak, J.J., 1996. “Is APEC a Natural Regional Trading Bloc? A Critique of the ‘Gravity Model’ of International Trade”, World Economy, pp. 533-543.
  - Wei, S.J., 1996. “Intra-national versus International Trade: How Stubborn are Nations in Global Integration?”, National Bureau of Economic Research Working Paper No. 5531.
  - Hummels, D., Levinsohn, J., 1995. “Monopolistic Competition and International Trade: Reconsidering the Evidence”, Quarterly Journal of Economics, 110 (3), pp. 799-836.
  - Thursby, J., Thursby, M., 1987. “Bilateral trade flows, the Linder hypothesis, and exchange risk”, The Review of Economics and Statistics, 69, pp. 488-495.
  - Redding, S., Venables, A.J., 2004. “Economic Geography and International Inequality”, Journal of International Economics, 62 (1), pp. 53-82.
- Other empirical and theoretical contributions relevant to trade integration and political economy:
  - Groosman and Helpman, various works cited: Grossman, G.M., Helpman, E., 1991; Grossman, G.M., Helpman, E., 1995.
  - Subramanian, A., Wei, S.J., 2007. “The WTO Promotes Trade, Strongly But Unevenly”, Journal of International Economics, forthcoming.
  - Glick, R., Rose, A.K., 2002. “Does a currency union affect trade? The time-series evidence”, European Economic Review, 46, pp. 1125-1151.

*Italicized: Source — _wp0866 - References (pdf)._*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0866.pdf_
