## wpiea2021064-print-pdf

## Source details

**Canonical URL:** [wpiea2021064-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021064-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021064-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021064-print-pdf.pdf.json)

---

### Introduction and objective
- Reassess empirical identification of policies that foster export diversification and development of superior exports by:
  - Using levels of export categories (non-hydrocarbon/mineral (NHM), manufacturing, services, complex) rather than indices (concentration, sophistication, complexity).
  - Adding gravity-equation and labor cost variables (following Eaton and Kortum (2002)) to commonly used productivity-related independent variables.
- Rationale:
  - Indices such as the Herfindahl-Hirschman Index (HHI), Exports Sophistication Index (ESI), and Economic Complexity Index (ECI) are substantially affected by exogenous commodity price shocks and resource discoveries.
  - Normalizing export categories per population or labor force reduces sensitivity to commodity price fluctuations and improves identification of policy determinants.
- Key empirical claim:
  - Adding horizontal policy variables (governance, educational attainment, infrastructure quality, trade policy openness) to countries’ proximity to international markets explains above 80 percent of cross-country variation in the targeted exports.

### Theoretical and empirical approach
- Theoretical basis:
  - Rooted in EK02 (Eaton and Kortum, 2002) Ricardian general equilibrium model; trade shares implied by a model expression preserved in source.
- Empirical specification:
  - Log-linear panel specification (equation (6) in source) with year fixed effects (μ_t) and country-pair fixed effects (θ_ni).
  - Gravity-aligned regressors include source GDP (X_i,t), importer remoteness index (R_n,t), exporter remoteness index (R_i,t), and importer T-variables (T_n,t).
  - Primary normalization: exports per population (exports per capita). Labor-force normalization discussed and GDP normalization rejected.
- Identification and estimation:
  - Hausman and Taylor (1981) instrumental variable technique used to address correlation of country-pair fixed effects with time-invariant variables.
  - Remote Index (RIi) and Proximity to Markets (PM̂i) indices constructed:
    - RIi = Σ Xn / dni  (equation (8))
    - PM̂i = Σ Xn^(α̂2) dni^(α̂3)  (equation (9))
  - Predicted exports per capita derived from PM and then from PM, T, and w variables using coefficients from bilateral regression (equations (10) and (11)).

### Choice of dependent and independent variables
- Dependent variables recommended:
  - Values of NHM, manufacturing, services, and complex exports per capita (or per labor force).
  - Complex exports defined as products with PCI above zero (upper half of PCI).
- Core independent variables (four emphasized T-variables):
  - Institutional development (governance).
  - Educational attainment.
  - Transport infrastructure quality.
  - Trade policy openness.
- Additional controls:
  - Gravity-related variables: distance, common currency, Free Trade Agreement, common border, common language, common colonizer, past colonial link.
  - Labor costs / labor market flexibility (IMF labor subindex).
  - GDP per capita included as control for endogeneity but excluded from goodness-of-fit calculations when estimating predictive power of policy variables.
- Multilateral resistance handled via remoteness indices; price variables excluded for simplicity.

### Major empirical findings: predictive power and drivers
- Proximity and distance:
  - PM alone explains about a quarter of the variation in NHM exports per capita (correlation = 0.54; fitted-line R-squared ≈ 0.25).
  - Cutting distance to markets by half is associated with a 150 percent increase in NHM exports (distance coefficient ≈ -1.247 to -1.471 across specifications).
  - Doubling PM is associated with a 15 percentage points higher annual growth in NHM exports (dynamic regression).
- Combined predictive power:
  - Combining PM with horizontal policy variables raises explanatory power to R-squared up to 0.82 and correlation up to 0.90.
  - The combination of horizontal policy variables with proximity to international markets explains above 80 percent of cross-country variation in the targeted exports.
- Policy variable magnitudes (selected elasticities and associations):
  - Education (UN Index): one standard deviation increase associated with a 170 percent increase in NHM exports (column 2).
  - Governance (WB Index): one standard deviation increase associated with a 65 percent increase in NHM exports (column 2).
  - Infrastructure (GCR Index): one standard deviation increase associated with a 20 percent increase in NHM exports (column 2).
  - Reducing average import tariff from 15 to 5 percent associated with a 45 percent increase in NHM exports (column 2).
  - IMF labor subindex: one standard deviation increase associated with a 45 percent increase in NHM exports (column 3).
  - Doing Business Score: one standard deviation increase associated with a 15 percent increase in NHM exports.
  - Technological readiness (GCR Index): one standard deviation increase associated with a 10 percent increase in NHM exports.
- Selected regression coefficient highlights (exact reported values from panel regressions):
  - Log GDP reporter: 1.527*** (col 1); 0.660*** (col 2); 0.160 (col 3); 0.825*** (col 4)
  - Log GDP partner: 0.386***; 0.220; 0.030; 0.159*
  - Log distance: -1.471***; -1.247***; -1.240***; -1.329***
  - Governance (WB Index): 0.457***; 0.176**
  - Education (UN Index): 4.854***; 3.550***
  - Infrastructure (GCR Index): 0.169***; 0.283***
  - Average Tariff: -0.0299***; -0.0429***
  - Labor market flexibility (IMF Index): 2.861***
  - Trade liberalization (Wacziarg and Welch, 2003): 0.517***
  - Observations: 169,968 (col 1); 44,989 (col 2); 31,253 (col 3); 90,352 (col 4)
  - Rho: 0.76; 0.94; 0.96; 0.86
- Sectoral differences:
  - Distance less relevant for services (Log distance for Services: -0.501**).
  - Governance and education matter across NHM, complex, manufacturing, and services.
  - Infrastructure and tariffs less important for services.
  - Labor market flexibility more important for complex and manufacturing exports than for NHM.
- Outliers and interpretation:
  - Notable upward outliers with higher-than-predicted NHM or complex exports given PM: Australia (AUS), Chile (CHL), New Zealand (NZL), Korea (KOR), Japan (JAP), Malaysia (MYS).
  - Outliers may reflect GVC participation with high imported content, reexports in small trading hubs, or oil exporters with subsidized industries.
- Robustness and caveats:
  - RI and PM are second-best approaches to multilateral resistance; exporter-time and importer-time fixed effects would be first-best but compromise identification of time-invariant determinants.
  - Zero trade flows excluded due to log dependent variables; omission especially affects service export regressions.
  - Analysis centers on gross exports; domestic value added differs by region (OECD TIVA examples: Australia 81 percent, Chile 88 percent, EAHI and EAEM around 60 percent).

### Policy implications and recommendations
- Central policy priority: shorten effective distance to other economies by enhancing connectedness:
  - Reduce trade policy barriers and strengthen trade facilitation.
  - Strengthen transport infrastructure and port/electricity infrastructure.
  - Invest in top-notch communication technology (particularly internet connectivity) to support the digital economy.
  - Foster technological diffusion (including educational exchange programs).
- Strengthen horizontal policy areas shown statistically important:
  - Institutions (government effectiveness; control of corruption).
  - Education (with emphasis on secondary and tertiary education).
  - Technological readiness and infrastructure.
  - Reduce tariffs and liberalize trade where feasible.
  - Improve Doing Business indicators and remove excessively restrictive regulations, including labor market reforms where appropriate.
- Sequencing and complementarities:
  - Short-to-medium term reforms can yield significant payoffs while longer-term institution-building continues.
  - Horizontal policy strength increases the likelihood of effective sector-specific (vertical) policies; sector-specific interventions are second-best tools and carry risks (fiscal erosion, rent seeking, distortions).
  - Special Economic Zones (SEZs) can be useful but are not a panacea.
- Caution advised:
  - Avoid “hard industrial policies” (tax incentives, subsidized credit, exchange rate manipulation, sector-specific protection) given limited robust evidence of effectiveness and associated risks.

### Measurement recommendation and concluding synthesis
- Measurement:
  - For empirical identification of policies that foster diversification and superior exports, prioritize direct measures of targeted export categories per capita (NHM, manufacturing, services, complex) rather than indices sensitive to HM exports (HHI, ESI, ECI).
- Synthesis of main empirical messages:
  - PM (distance/proximity) is a central determinant: explains about a quarter of cross-country variation in NHM exports per capita by itself; halving distance increases NHM exports by about 150 percent.
  - Service exports, being less distance-sensitive, offer a more feasible diversification route for remote countries.
  - Combining PM with horizontal policy variables produces very high predictive power (R-squared up to 0.82), implying that strong horizontal policies can substantially offset remoteness.
  - Priority policy focus: enhance connectedness and strengthen horizontal policies (institutions, education, technological readiness, infrastructure), complemented carefully by selective sector-level measures when justified.

*Source: wpiea2021064-print-pdf - Annex Tables and Figures, and main text summarizing methodology, empirical findings, and policy implications*

### Annex Tables

### Annex Tables

### Table index

- Table A.1: Determinants of Economic Complexity Index (ECI) and ECI Plus ..................... 32
- Table A.2: List of Countries by Regional Group .................................................................... 33
- Table A.3: Determinants of exports by regression specification ............................................ 37
- Table A.4: Summary Results by Study ................................................................................... 45

*Source: wpiea2021064-print-pdf - Annex Tables*

### Annex Figures

### Annex Figures

### Introduction
- Objective: Reassess empirical identification of policies that foster export diversification and development of superior exports by:
  - Using levels of export categories (non-hydrocarbon/mineral (NHM), manufacturing, services, complex) rather than indices (concentration, sophistication, complexity).
  - Adding gravity equation and labor cost variables (following Eaton and Kortum (2002) Ricardian model) to commonly used productivity-related independent variables.
- Rationale:
  - Indices such as the Herfindahl-Hirschman Index (HHI), Exports Sophistication Index (ESI), and Economic Complexity Index (ECI) are substantially affected by exogenous commodity price shocks and resource discoveries, weakening statistical links to policy variables.
  - Normalizing export categories per population or labor force reduces sensitivity to commodity price fluctuations and improves identification of policy determinants.
- Key empirical claim:
  - Adding horizontal policy variables (governance, educational attainment, infrastructure quality, trade policy openness) to countries’ proximity to international markets explains above 80 percent of cross-country variation in the targeted exports.

### Theoretical and Empirical Considerations
- Problems with index-based dependent variables:
  - HHI example: 퐻퐻퐼_j = ∑ (푥_푠푗 / ∑ 푥_푠푗)^2_푠 — can fluctuate with nominal value changes in HM exports unrelated to domestic policy.
  - ESI example: 퐸푆퐼_j = ∑ (푥_푗푠 / ∑ 푥_푗푠) 푃푅푂퐷푌_푠 and 푃푅푂퐷푌_푠 = ∑ ((푥_푗푠 / ∑_푖 푥_푖푗푠) / ∑_푗 (푥_푗푠 / ∑_푖 푥_푖푗푠)) 퐺𝐷𝑃𝑝𝑐_푗 — product sophistication measures can decline mechanically when HM nominal values rise.
  - Empirical illustrations:
    - Chile: copper exports rose from US$ 8 billion in 2003 to a peak of US$ 54 billion in 2011; HHI and ECI moved in ways inconsistent with underlying NHM and complex exports per capita trends.
    - Australia’s ECI is lower than El Salvador and Honduras despite higher institutional and educational quality — likely driven by mineral export shares.
    - Texas example: ECI of 0.29 despite technology leadership, attributed to petroleum endowment and a Product Complexity Index (PCI) of Petroleum Oils of -2.57.
  - Fixed effect regressions (Table A.1 referenced) show ECI strongly associated with resource wealth (Sachs and Warner (1995)), implying commodity exposure confounds ECI–growth links.
- Proposed dependent variables:
  - Use values of NHM, manufacturing, services, and complex exports per capita (or per labor force) to capture diversification and superior export development directly.
  - Complex exports can be defined as those with PCI above zero (upper half of PCI).
- Independent variable considerations:
  - Incorporate gravity-related variables (distance, common currency, Free Trade Agreement, common border, common language, colonizer, past colonial dummy) and labor costs (labor market flexibility) alongside standard T-variables (institutional development, educational attainment, trade policy openness, infrastructure).
  - Acknowledge inclusion of importer T-variables and remoteness indices to control for multilateral resistance.

### Methodology and Data
- Modeling approach:
  - Rooted in EK02 (Eaton and Kortum, 2002) Ricardian general equilibrium model; trade shares implied by:
    - (5) X_ni / X_n = T_i (γ d_ni w_i^β p_i^(1−β) p_n^...)^(−θ) (textual form preserved as in source)
  - Log-linear panel specification (after rearrangement and log-linearization):
    - (6) X_ni,t = α1 log(γ) + α2 log(X_n,t) + α3 log(d_ni) + α4 log T_i,t + α5 log(w_i,t) + θ_ni + μ_t + ε_ni,t
    - Year fixed effects (μ_t) and country-pair fixed effects (θ_ni) are included.
  - Gravity-equation alignment adds:
    - (i) GDP of the source country (X_i,t)
    - (ii) remoteness index of the importer (R_n,t) — inward multilateral resistance control
    - (iii) remoteness index of the exporter (R_i,t) — outward multilateral resistance control
    - (iv) T-variables of the importing country (T_n,t)
- Normalization choices:
  - Primary normalization: exports per population to enable cross-country comparability and reduce sensitivity to commodity price fluctuations.
  - Trade-offs:
    - Normalizing by labor force can control for working-age composition but may be endogenous to policies (labor force participation could rise because of successful policies).
    - Normalizing by GDP is rejected because GDP in HM-abundant countries includes HM exports.
- Variable selection rationale:
  - Four T-variables emphasized: institutional development (governance), educational attainment, transport infrastructure quality, and trade policy openness — found most economically and statistically significant to diversification and complexity.
  - GDP per capita included as a control for endogeneity but excluded from goodness-of-fit calculations when estimating predictive power of policy variables.
- Estimation notes:
  - Price variables excluded for statistical simplicity; multilateral resistance terms are used as reduced-form controls.
  - Potential multicollinearity: log GDP per capita and log GDP of source country have a sample correlation of 0.5; acknowledged but not treated as invalidating main inferences.

### Empirical Findings (summarized)
- Main statistical relationships:
  - Strong relation between gravity-related variables and NHM, manufacturing, complex, and service exports.
  - Governance, educational attainment, infrastructure quality, and trade policy openness are robustly related to these exports.
  - The combination of horizontal policy variables with proximity to international markets explains above 80 percent of cross-country variation in the targeted exports.
- Robustness and interpretation:
  - Complex exports per capita positively associated with ECI in fixed effects regressions (Table A.1), but complex exports per capita are not significantly correlated with HM exports to GDP; hence complex exports per capita is less affected by natural resource abundance.
  - Empirical results are consistent with GVC literature linking gravity variables to Global Value Chain participation and thus to more complex manufacturing exports.

### Policy Implications (as discussed)
- Policy levers with robust statistical associations to diversification and superior exports:
  - Strengthen governance (institutional development).
  - Increase educational attainment.
  - Improve transport infrastructure quality.
  - Enhance trade policy openness.
- Complementarity with geography:
  - Policies above are important but must be considered alongside countries’ geographic proximity and other gravity-related factors to major markets; proximity explains a large share of cross-country variation in export outcomes.
- Measurement recommendation:
  - For empirical identification of policies that foster diversification and superior exports, prioritize direct measures of targeted export categories per capita (NHM, manufacturing, services, complex) rather than indices susceptible to commodity shocks (HHI, ESI, ECI).

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021064-print-pdf.pdf*

### conclusions of this paper because: (i) the estimated coefficient of log GDP, which is used to calculate the

### wpiea2021064-print-pdf - conclusions of this paper because: (i) the estimated coefficient of log GDP, which is used to calculate the

### Methodology and key indices
- Gravity framework estimated with Hausman and Taylor (1981) instrumental variable technique to address correlation of country-pair fixed effects with time-invariant variables.
- Remote Index (RI) of country i:
  - RIi = Σ Xn / dni  (equation (8))
- Proximity to Markets (PM) index of country i (aggregate gravity pull):
  - PM̂i = Σ Xn^(α̂2) dni^(α̂3)  (equation (9))
  - In illustrative charts, exponents α̂2 and α̂3 are set equal to one.
- Predicted exports per capita based on PM and then on PM, T, and w variables using coefficients from bilateral regression (equations (10) and (11)).

### Major empirical findings (determinants of NHM, complex, manufacturing, and services exports)
- PM and distance:
  - PM alone explains about a quarter of the variation in NHM exports per capita (correlation = 0.54; fitted-line R-squared ≈ 0.25).
  - Cutting distance to markets by half is associated with a 150 percent increase in NHM exports (distance coefficient ≈ -1.247 to -1.471 across specifications).
  - Doubling PM is associated with a 15 percentage points higher annual growth in NHM exports (dynamic regression).
- Trade liberalization:
  - Trade liberalization dummy (Wacziarg and Welch) associated with a 59 percent increase in NHM exports per capita in one specification and an 80 percent increase in NHM exports per capita in another specification (fourth regression).
  - Trade liberalization episodes are followed by a 6 percentage points acceleration in NHM exports growth (dynamic regression).
- Policy (T-) variables: statistical and economic significance
  - Education (UN Index): one standard deviation increase associated with a 170 percent increase in NHM exports (column 2).
  - Governance (WB Index): one standard deviation increase associated with a 65 percent increase in NHM exports (column 2).
  - Infrastructure (GCR Index): one standard deviation increase associated with a 20 percent increase in NHM exports (column 2).
  - Reducing average import tariff from 15 to 5 percent associated with a 45 percent increase in NHM exports (column 2).
  - IMF labor subindex: one standard deviation increase associated with a 45 percent increase in NHM exports (column 3).
  - Doing Business Score: one standard deviation increase associated with a 15 percent increase in NHM exports.
  - Technological readiness (GCR Index): one standard deviation increase associated with a 10 percent increase in NHM exports.
- Subcomponents within policy areas:
  - Government effectiveness: one standard deviation increase associated with a 60 percent higher NHM exports.
  - Control of corruption: one standard deviation increase associated with a 40 percent higher NHM exports.
  - Secondary and tertiary education more significant than primary education.
  - Port and electricity infrastructure most statistically significant among infrastructure subcomponents.
- Sectoral differences:
  - Distance less relevant for services exports; governance and education matter across NHM, complex, manufacturing, and services; infrastructure and tariffs less important for services.
  - Labor market flexibility more important for complex and manufacturing exports than for NHM.
- Predictive power when combining PM and policy variables:
  - R-squared increases up to 0.82 and correlation up to 0.90 when PM and policy explanatory variables (from regression on page 19) are combined.
  - No country with the PM and policy variables typical of most Sub-Saharan African or Latin American countries attains the NHM exports per capita of Japan, Korea, or Malaysia given those PM and policy variables.

### Key regression coefficient highlights (selected exact values reported)
- Determinants of non-hydrocarbon/mineral (NHM) exports (panel regressions, Hausman and Taylor):
  - Log GDP reporter: 1.527*** (col 1); 0.660*** (col 2); 0.160 (col 3); 0.825*** (col 4)
  - Log GDP partner: 0.386***; 0.220; 0.030; 0.159*
  - Log distance: -1.471***; -1.247***; -1.240***; -1.329***
  - Governance (WB Index): 0.457***; 0.176**
  - Education (UN Index): 4.854***; 3.550***
  - Infrastructure (GCR Index): 0.169***; 0.283***
  - Average Tariff: -0.0299***; -0.0429***
  - Labor market flexibility (IMF Index): 2.861***
  - Trade liberalization (Wacziarg and Welch, 2003): 0.517***
  - Observations: 169,968 (col 1); 44,989 (col 2); 31,253 (col 3); 90,352 (col 4)
  - Rho: 0.76; 0.94; 0.96; 0.86
- Other determinants table (selected):
  - Log distance: -1.247***; -1.265***; -1.237***; -1.223***
  - Doing Business Score: 0.0155***
  - Technological readiness (GCR Index): 0.0976***
  - Observations: 44,989; 44,704; 44,704; 44,110
  - Rho: 0.94; 0.95; 0.95; 0.95
- Determinants by export type (selected coefficients):
  - Log GDP reporter for Services: 2.401***
  - Log distance for Services: -0.501**
  - Governance (WB Index) for Services: 3.186***
  - Observations for Services regression: 5,512; Rho = 0.96

### Analysis of fit and outliers
- PM-only prediction:
  - Correlation between actual NHM exports per capita and PM-predicted values = 0.54; R-squared ≈ 0.25.
- PM + policy variables:
  - Correlation up to 0.90; R-squared up to 0.82.
- Notable upward outliers (countries with higher-than-predicted NHM or complex exports given PM): Australia (AUS), Chile (CHL), New Zealand (NZL), Korea (KOR), Japan (JAP), Malaysia (MYS).
- Remaining positive outliers may reflect:
  - Participation in GVCs with high imported components (CAM, East Asia, EE).
  - Small trading hubs with significant reexports (Hong Kong, Panama, Singapore).
  - Oil exporting countries with subsidized industries (may be economically inefficient models of diversification).

### Robustness, caveats, and data notes
- RI and PM are second-best approaches to multilateral resistance; exporter-time and importer-time fixed effects suggested as first-best but would wash out identification of hypothesized determinants due to multicollinearity.
- Zero trade flows are excluded because dependent variables are in logarithmic terms; omission may affect service export regressions more due to localization and specialization of services.
- Analysis centers on gross exports; domestic value added differs by region (OECD TIVA: Australia 81 percent, Chile 88 percent, EAHI and EAEM around 60 percent).
- Data sources cited include UN Comtrade (EBOPS), CEPII gravity database, Polity IV, World Bank governance indicators, UN Education index, Barro-Lee, World Integrated Trade Solution tariffs, Global Competitiveness Report, Doing Business, and IMF labor subindices.

### Policy implications and recommendations
- Central policy priority: shorten effective distance to other economies by enhancing connectedness:
  - Reduce trade policy barriers and strengthen trade facilitation.
  - Strengthen transport infrastructure and port/electricity infrastructure.
  - Invest in top-notch communication technology (particularly internet connectivity) to support the digital economy.
  - Foster technological diffusion (including educational exchange programs).
- Strengthen horizontal policy areas shown statistically important:
  - Institutions (government effectiveness; control of corruption).
  - Education (with emphasis on secondary and tertiary).
  - Technological readiness and infrastructure.
  - Reduce tariffs and liberalize trade where feasible.
  - Improve Doing Business indicators and remove excessively restrictive regulations, including labor market reforms where appropriate.
- Short-to-medium term reforms can yield significant payoffs even while longer-term institution-building proceeds.
- Sector-specific (vertical) policies:
  - Horizontal policy strength increases the likelihood of effective vertical policies; sector-specific interventions can be considered as second-best tools (e.g., targeted technical education, sector-specific infrastructure, lower restrictions on imported inputs), but with caution due to risks of fiscal erosion, rent seeking, and distortions.
  - Special Economic Zones (SEZs) historically used (example: Mauritius) to bypass economy-wide constraints; useful but not a panacea.
- Caution against “hard industrial policies” (tax incentives, subsidized credit, exchange rate manipulation, sector-specific protection) given limited robust evidence of effectiveness and risks identified in the literature.

### Concluding remarks (synthesis)
- Directly analyzing export groups associated with diversification (NHM, manufacturing, complex, services) yields more accurate identification of determinants than commonly used diversification indices, which are sensitive to HM exports.
- PM (distance/proximity) is a central determinant: explains about a quarter of cross-country variation in NHM exports per capita; halving distance increases NHM exports by about 150 percent.
- Service exports, being less distance-sensitive, offer a more feasible diversification route for remote countries.
- Combining PM with horizontal policy variables produces very high predictive power (R-squared up to 0.82), implying that strong horizontal policies can substantially offset remoteness.
- Priority policy focus: enhance connectedness and strengthen horizontal policies (institutions, education, technological readiness, infrastructure), complemented carefully by selective sector-level measures when justified.

*Source: wpiea2021064-print-pdf*

### References

### References

### Cited works
- Albeaik, Saleh, Mary Kaltenberg, Mansour Alsaleh, Cesar Hidalgo, 2017, "Improving the Economic Complexity Index," Papers 1707.05826, arXiv.org, revised Jul 2017.
- Allen, Treb, Costas Arkolakis, Yuta Takahashi, 2020, "Universal Gravity," Journal of Political Economy, University of Chicago Press, vol. 128(2), pages 393-433.
- Anderson James and Eric van Wincoop, 2003, "Gravity with Gravitas: A Solution to the Border Puzzle," American Economic Review, American Economic Association, vol. 93(1), pages 170-192.
- Arkolakis, Costa, Arnaud Costinot and Andrés Rodríguez-Clare, 2012, "New Trade Models, Same Old Gains?" American Economic Review, American Economic Association, vol. 102(1), pages 94-130.
- Bakker, Bas, Manuk Ghazanchyan, Alex Ho, and Vibha Nanda, 2020, “The Lack of Convergence of Latin America compared with CESEE and East and South-east Asia: Is Low Investment to Blame?” IMF Working Papers 20/98, International Monetary Fund.
- Barro, Robert J. and Jong Wha Lee, 2013, "A new data set of educational attainment in the world, 1950–2010," Journal of Development Economics, Elsevier, vol. 104(C), pages 184-198, retrieved from: < http://www.barrolee.com/>.
- Cadestin, Charles, Julien Gourdon and Przemyslaw Kowalski, 2016, "Participation in Global Value Chains in Latin America: Implications for Trade and Trade-Related Policy," OECD Trade Policy Papers 192, OECD Publishing.
- Caves, Richard E. and Uekusa Masu, 1976, Industrial Organization in Japan,: The Brookings Institution, 1976. Pp. xi, 169, Washington, D.C.
- Cherif, Reda and Hasanov Fuad, 2014, “Soaring of the Gulf Falcons: Diversification in the GCC Oil Exporters in Seven Propositions,” IMF Working Paper WP/14/177, Washington DC.
- Cherif, Reda, and Fuad Hasanov, 2019, "Principles of True Industrial Policy," Journal of Globalization and Development, De Gruyter, vol. 10(1), pages 1-22, June.
- Ding, Xiaodan and Metodij Hadzi-Vaskov, 2017, "Composition of Trade in Latin America and the Caribbean," IMF Working Papers 2017/042, International Monetary Fund.
- Eaton, Jonatahan and Samuel Kortum, 2002, "Technology, Geography, and Trade," Econometrica, Econometric Society, vol. 70(5), pages 1741-1779.
- Giri, Rahul, Saad N Quayyum, Rujun Yin, 2019, "Understanding Export Diversification: Key Drivers and Policy Implications," IMF Working Papers 19/105, International Monetary Fund.
- Haddad, Mona, Jamus Jerome Lim, Cosimo Pancaro and Christian Saborowski, 2012, “Trade openness reduces growth volatility when countries are well diversified,” ECB Working Paper No. 1491.
- Hallak, Juan Carlos, 2006, "Product Quality And The Direction Of Trade," Journal of International Economics, Vol 68 pp. 238-265.
- Harrison, Ann and Andrés Rodríguez-Clare, 2010, “Trade, Foreign Investment, and Industrial Policy for Developing Countries,” in Dani Rodrik and Mark Rosenzweig, editors: Handbook of Development Economics, Vol. 5, The Netherlands: North-Holland, 2010, pp. 4039-4214. ISBN: 978-0-444-52944-2
- Hausman, Jerry A. and William E. Taylor, 1981, "Panel Data and Unobservable Individual Effects," Econometrica, Econometric Society, vol. 49(6), pages 1377-1398.
- Hausmann, Ricardo, Jason Hwang and Dani Rodrik, 2006, "What You Export Matters," CEPR Discussion Papers 5444, C.E.P.R. Discussion Papers.
- Hausmann, Ricardo, César A. Hidalgo, Sebastián Bustos, Michele Coscia, Alexander Simoes, Muhammed A. Yildirim, 2013, “The Atlas of Economic Complexity: Mapping Paths to Prosperity,” The MIT Press.
- Hidalgo, Cesar and Ricardo Hausmann, 2009, "The Building Blocks of Economic Complexity," Proceedings of the National Academy of Sciences of the United States of America, 106 (26), pp. 10570-10575.
- Head, Keith, Tierry Mayer, and Jhon Ries, 2010, “The erosion of colonial trade linkages after independence,” Journal of International Economics, 81(1):1–14, 2010.
- Hnatkovska, Viktoria V. and Norman Loayza, 2004, "Volatility and growth," Policy Research Working Paper Series 3184, The World Bank.
- International Monetary Fund, 1980, World Economic Outlook, Washington, D.C: International Monetary Fund.
- International Monetary Fund, 2019, “Reigniting growth in low-income and emerging market economies: what role can structural reforms play,” World Economic Outlook, Chapter 3, October. (Washington).
- OECD, 2018, Trade in Value added. Available at http://www.oecd.org/sti/ind/measuring-trade-in-value-added.htm
- Polity IV, 2014, Polity IV Project, Severn, Maryland: Center for Systematic Peace. Retrieved from: http://www.systemicpeace.org/polity/polity4.htm.
- Raei, Faezeh, Anna Ignatenko, Borislava Mircheva, 2019, "Global Value Chains: What are the Benefits and Why Do Countries Participate?," IMF Working Papers 2019/018, International Monetary Fund.
- Ramey, Garey and Valerie A. Ramey, 1995, “Cross-Country Evidence on the Link Between Volatility and Growth,” The American Economic Review, Vol. 85, No. 5 (Dec., 1995), pp. 1138-1151.
- Rodrik, Dani, 2004, "Industrial Policy for the Twenty-First Century," CEPR Discussion Papers 4767, C.E.P.R. Discussion Papers.
- Rodrik, Dani, 2008, "Normalizing Industrial Policy," World Bank Publications, The World Bank, number 28009, November.
- Rose, Andrew K,2004, Do We Really Know That the WTO Increases Trade? American Economic Review, 94(1):98–114, 2004.
- Sachs, Jeffrey and Andrew M. Warner 1995, “Economic reform and the process of global integration”, Brookings Papers on Economic Activity, 1–118.
- United Nations, 2020, UN Comtrade. Available at <http://comtrade.un.org>.
- United Nations Development Program, 2020, Human Development Reports, Various editions, retrieved from: http://hdr.undp.org/en/content/education-index.
- Wei, Shang-Jin, 1996, “Intra-national versus International Trade: How Stubborn Are Nation States in Globalization?”, NBER working papers no. 5331.
- Wacziarg, Romain and Karen Horn Welch, 2008, "Trade Liberalization and Growth: New Evidence," World Bank Economic Review, World Bank Group, vol. 22(2), pages 187-231.
- Weldemicael, Ermias O., 2012, “Determinants of Export Sophistication,” The University of Melbourne, available at https://www.business.unsw.edu.au/About-Site/Schools-Site/Economics-Site/Documents/E.Weldemicael%20-%20Determinants%20of%20Export%20Sophistication.pdf
- World Bank, 2020a, Worldwide Governance Indicators, various editions, World Bank Publications, World Bank, Washington DC, retrieved from: <https://info.worldbank.org/governance/wgi/>.
- World Bank, 2020b, World Integrated Trade Solution, various editions, World Bank, Washington DC, retrieved from: https://wits.worldbank.org/.
- World Bank, 2020c, Doing Business, various editions, World Bank Publications, The World Bank, Washington DC, retrieved from: https://www.doingbusiness.org/en/doingbusiness.
- World Economic Forum, and Harvard University, 2020, The Global Competitiveness Report, various editions, Geneva: World Economic Forum, retrieved from: https://www.weforum.org/reports/the-global-competitiveness-report-2020.
- Yotov, Yoto, Roberta Piermartini, Jose Antonio Monteiro, and Mario. Larch, 2016, “An advanced guide to trade policy analysis: the structural gravity model,” World Trade Organization.

### Data sources highlighted in the references
- UN Comtrade (United Nations, 2020)
- World Economic Outlook (International Monetary Fund)
- World Bank: Worldwide Governance Indicators; World Integrated Trade Solution; Doing Business
- OECD Trade in Value Added
- Polity IV Project (Polity IV, 2014)
- Human Development Reports (United Nations Development Program, 2020)
- Global Competitiveness Report (World Economic Forum and Harvard University, 2020)

### Tables, figures, and reported numeric entries in the References appendix
- Table A.1: Determinants of Economic Complexity Index (ECI) and ECI Plus
  - Dependent Variable coefficients (excerpt): 0.077***, 0.079***, 0.081***, 0.083***, -0.162***, -0.292***
  - Constant entries (excerpt): -0.174***, -0.184***, -0.175***, -0.175***
  - Observations: 4,370; 4,207; 4,274; 4,207
  - R-squared: 0.47; 0.44; 0.35; 0.45
  - Additional labels: Log Complex Exports per Capita; Hydrocarbon and Mineral Exports-to-GDP (Sachs and Warner, 1995); ECI Plus; ECI
  - Sources: Hausmann and others (2013), World Economic Outlook (IMF), UN Comtrade, and author's calculations.
  - Notes: * p<0.1, ** p<0.05, *** p<0.01. Fixed effects regression.

- Table A.2: List of Countries by Regional Group (regional group names and many country entries preserved in source; includes region codes such as AND, ARB, CA, CAR, EU, IND, ME, NAM, OCE, SCND, SCC, SSA, etc.)

- Table A.3: Determinants of exports by regression specification
  - Dependent Variable: Log of non-hydrocarbon/mineral exports
  - Regression specification types: Hausman-Taylor, Pooled OLS, Between Effects, Random Effects, Fixed Effects
  - Selected coefficient values (excerpt):
    - Log GDP reporter: 0.756***, 0.484***, 0.756***, 1.354***, 1.345***, 1.369***, -0.548***
    - Log GDP partner: 0.858***, 0.960***, 0.858***, 0.920***, 0.904***, 0.925***, 0.347***
    - Log distance: -1.279***, -0.616***, -1.279***, -1.420***, -1.444***, -1.465***
    - Common currency dummy: 0.220, .368**, 0.22, -0.193*, -0.27, -0.01
    - Common border dummy: 1.888***, 2.999***, 1.888***, 1.308***, 1.327***, 1.260***
    - Common language dummy: 0.617***, 0.899***, 0.617***, 0.676***, 0.773***, 0.638***
    - Common colonizer dummy: 0.339**, 0.327**, 0.339**, 0.602***, 0.561***, 0.444***
    - Past colonial link dummy: 1.228***, 1.309***, 1.228***, 0.526***, 0.482***, 0.790***
    - Log GDP per capita: -0.15, -0.04, -0.15, -0.918***, -0.973***, -0.680***, 0.866***
    - Governance (WB Index): 0.484***, 0.422***, 0.484***, 0.634***, 0.505***, 0.789***, 0.099*
    - Education (UN Index): 5.099***, 3.924***, 5.099***, 1.166***, 1.498***, 0.799***, 1.031**
    - Infrastructure (GCR Index): 0.175***, 0.166***, 0.175***, 0.694***, 0.864***, 0.307***, 0.113***
    - Average Tariff: -0.0310***, -0.0304***, -0.0310***, -0.0722***, -0.0784***, -0.0421***, -0.0197***
    - Labor market flexiblity (GCR Index): 0.02, 0.04, 0.02, -0.334***, -0.374***, -0.04, 0.03
  - Constant entries (excerpt): -0.86, -19.34***, -0.86, -11.67***, -12.17***, -11.28***, -7.223**
  - Observations: 44,989 (multiple columns); in one column 44,989; another shows 44,989 and 44,989, and one shows 44,989, 44,989, 44,989, 44,989, 44,989, 44,989, 44,989, 989
  - Notes: * p<0.1, ** p<0.05, *** p<0.01. Panel regressions based on Hausman and Taylor (1981) technique with groups consisting of all combinations of reporter and partner countries in UN Comtrade database. Observations are non-overlapping 5-year averages within the 1962-2018 period, depending on data availability. Regression specification based on equation (7). Multilateral resistance terms and partner country's policy variables included (coefficients not reported). Dependent variable is the logarithm of the value of exports excluding hydrocarbon and mineral products (SITC2 codes 0-2999, 4000-6772, 6900-8999).

- Table A.4: Summary Results by Study (excerpted comparative entries)
  - Adjusted R-Squared with larger specification: 0.80 to 0.86; 0.70 to 0.75; 0.70 to 0.73; 0.457 to 0.751; 0.457 to 0.751
  - Notes on variable significance across studies: Proximity to Markets; Governance; Education; Infrastructure; Trade Openness (as reported across studies in the table)

### Appendix figure panels listed
- Panel Figure A.1: Index of Determinants of Economic Complexity
- Panel Figure A.2: Actual vs Predicted Exports
- Panel Figure A.3: Deviation of Actual Exports from Predicted-by Distance Exports
- Panel Figure A.4: Distance and Fundamentals by Subregion in 2015-2017
- Panel Figure A.5: Other Explanatory Variables by Subregion in 2015-2017

*References and appendix tables/figures as presented in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021064-print-pdf.pdf_