## _wp11123

## Source details

**Canonical URL:** [_wp11123](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11123.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11123.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11123.pdf.json)

---

### Introduction
- Research questions:
  - Does the composition of FDI matter for current account deficits?
  - Can policies influence the sectoral composition of FDI?
- Scope:
  - Analysis of effects of sectoral distribution of FDI on trade balance via exports and imports in fifteen CESE countries in the run-up to the global financial crisis.
  - Empirical examination of determinants of FDI in the tradable sector across CESE countries.
  - Policy recommendations for host countries to affect sectoral allocation of FDI from the viewpoint of external stability and competitiveness.

### Capital inflows in the CESE countries — key observations
- Capital inflows into CESE were already high in 2003 and increased further thereafter, driven by the prospect of EU accession, ample liquidity, and strong world growth.
- FDI was generally the largest component of capital inflows in the region.
- Country- and subregion-specific patterns:
  - Bulgaria and Romania (EU Balkans) recorded the largest inflows of FDI relative to GDP.
  - The Baltics (Estonia, Latvia, and Lithuania) saw momentum upon EU accession in 2004.
  - Albania, Bosnia & Herzegovina, Croatia, Macedonia, and Serbia (Non-EU Balkans) experienced an increasing trend since 2005, mainly due to large-scale privatization.
  - CEE countries (Czech Republic, Slovak Republic, Hungary, Poland, and Slovenia) had a more moderate increase in FDI after 2003.
- Sectoral composition in 2007 varied markedly across CESE:
  - SEE countries: FDI dominated by nontradable sectors with exceptions (Macedonia and Romania).
  - Estonia and Latvia showed a similar nontradable-dominated pattern.
  - CEE countries exhibited a more balanced distribution between tradable and nontradable sectors.
- Definitions used in the study:
  - Tradable sectors = manufacturing, agriculture, mining, retail, hotels and restaurants.
  - Nontradable sectors = construction, electricity, transport, communication, real estate, and financial intermediation.
- Data caveat:
  - Intercompany loans are recorded as FDI in some countries, which may exaggerate the size of FDI inflows.

### How sectoral composition of FDI relates to trade deficits — cross-country evidence
- Conceptual links:
  - FDI in tradable sector likely increases exports over time.
  - FDI in nontradable sector may fuel domestic demand booms and boost imports; only short-run import effects likely from tradable FDI.
  - Countries with FDI predominantly in nontradable sectors likely have higher trade deficits than those with FDI in tradable sectors.
- Empirical cross-section findings:
  - Positive correlation between stock of FDI in tradable sector (percent of GDP) and export-to-GDP ratio; highest export-to-GDP ratios observed in Slovak Republic, Czech Republic, and Hungary, which also have high tradable FDI stock.
  - Positive correlation between stock of FDI in nontradable sector and import-to-GDP ratio; Bulgaria and Estonia have the highest stock of nontradable FDI and high import-to-GDP ratios.
  - Link between nontradable FDI and credit growth is positive; a large share of nontradable FDI is often financial intermediaries.

### How sectoral composition of FDI relates to trade deficits — time-series evidence
- Country group patterns (2003–2007):
  - Non-EU Balkans: little-changed trade balance since 2003 (except Bosnia and Herzegovina), while share of tradable FDI is generally declining.
  - Baltics and New Member States: negative correlation between share of tradable FDI and trade account balance; in Bulgaria, Romania and Latvia sharp increase in trade deficits coincided with declining share of tradable FDI.
  - CEE: Czech Republic, Hungary, and Slovak Republic have high share of tradable FDI and improving trade balance; Poland and Slovenia show worsening trade balance as FDI increasingly flows to nontradable sectors.
- Medium-run inference:
  - More FDI in tradable sectors seems to improve the trade balance via export and import performance.

### Effects of tradable FDI on exports
- Mechanisms cited:
  - FDI augments domestic capital, transfers technology and new products, trains the local workforce, and upgrades technical and managerial skills—promoting host-country export performance.
- Empirical magnitudes:
  - Example from literature: Zhang (2005) finds one dollar of FDI stock raises exports by about 70 cents in disaggregate industry-level data for China.
  - For CESE countries, the estimated link between tradable FDI and exports is substantially higher than the Chinese study (different aggregation units).
  - Fitted relation reported: a one percentage point of GDP increase in tradable FDI leads to about three times as much increase in exports (upper-left panel of Figure 4): fitted line y = 3.4683x + 1.4941 with R² = 0.9508.
  - Using aggregate FDI (including nontradable FDI), the positive relation persists but is weaker (example cited: 1.8 dollar as opposed to 3.5).
- Panel econometrics (Appendix 1):
  - Panel estimates show a positive link between export performance and FDI in the tradable sector after controlling for real exchange rates and market size.
  - Export equation reported FDI_tradable/GDP coefficients: 0.00239**; 0.00388**; 0.00371**; 0.00428***.
  - FDI_tradable/FDI coefficients reported: 0.193***; 0.106*; 0.189***.
  - Lagged export term EX/Y(-1) coefficients reported: 0.548***; 0.570***; 0.583***; 0.866***; 0.872***; 0.866***; 0.962***; 0.804***.
  - log(REER) coefficients include negative and significant values in some specifications: -0.125**; -0.107**; -0.108*; -0.141*; -0.344**.

### Determinants of sectoral (tradable) FDI — host country factors and estimation strategy
- Objective: explain share of tradable FDI to total FDI and identify host-country policies that tilt FDI toward tradable (export-oriented) sectors.
- Initial independent variables used in panel regressions:
  - (log of) GDP (market size)
  - Income per capita
  - Wage
  - Education (tertiary enrollment rate)
  - Availability of infrastructure (EBRD index)
  - Trade integration (exports + imports as share of GDP)
  - Quality of bureaucracy and rule of law (ICRG indexes)
  - Distance from Western Europe (distance in kilometers from Dusseldorf to capital city)
- Additional controls for transition initial conditions:
  - Share of industry in 1989
  - Dummy for early transition (equals one if started transition in 1991 or earlier, zero otherwise)
- Policy variables included:
  - Restrictions on capital inflows (Schindler (2009) index)
  - Privatization revenue as share of GDP (EBRD)
  - Overall fiscal balance to GDP
  - Inflation rate (proxy for monetary stability)
- Estimation methods:
  - RE, FE, and multiple Sys-GMM specifications. System-GMM uses lagged dependent variable and secondary education as endogenous IVs; exogenous IVs include FDI stock in the region, political risk, quality of bureaucracy, rule of law, inflation, and trade integration.

### Key empirical findings (panel regression results)
- Positive and significant determinants of a larger share of tradable FDI:
  - Market size: ln(GDP) coefficients reported across specifications: 0.105***; 1.304*; 0.0873***; 0.0873***; 0.134***; 0.287**; 0.0997***; 0.0755***; 0.0616*.
  - Infrastructure: coefficients reported: 0.0820***; 0.0608***; 0.169***; 0.169***; 0.211***; 0.194***; 0.181***; 0.249***; 0.202***.
  - Trade integration: coefficients reported: 0.00204***; 0.000979; 0.00199***; 0.00199***; 0.00276***; 0.00457***; 0.00150*; 0.00295***; 0.00182***.
  - Education_tertiary: coefficients reported: 0.00165**; 0.00124; 0.00354**; 0.00354**; 0.00372*; 0.00407**; 0.00205; 0.00731**; 0.00236.
- Negative and significant determinant:
  - Distance to Dusseldorf: coefficients reported: -0.192*; -0.192*; -0.401**; -0.397**; -0.0921; -0.401*; -0.0957.
- Insignificant or non-supportive determinants for sectoral composition:
  - Wage (low labor cost): not significant; interpretation: foreign investors value labor quality (education) over low wage per se.
  - Institutional quality (quality of bureaucracy, rule of law): regression results do not support a role for institutional quality in determining the sectoral composition of FDI (although better institutions likely increase aggregate FDI).
  - Initial conditions (share of industry in 1989, early-transition dummy): coefficients insignificant or with unexpected signs.
  - Policy variables (capital controls on inflows, privatization revenues, fiscal balance, inflation): statistically insignificant for the sectoral distribution of FDI.
    - Capital controls on inflows do not appear to deter tradable-sector FDI in the sample.
    - Privatization revenue is statistically insignificant, including when splitting sample into Western Balkans and other countries.
- Additional reported correlations and fitted lines (figures):
  - Export/GDP vs Tradable FDI/GDP: y = 3.269x + 20.088, R² = 0.5736.
  - Import/GDP vs Nontradable FDI/GDP: y = 0.8709x + 53.258, R² = 0.1766.
  - Change in Private Credit/GDP vs Nontradable FDI/GDP: y = 0.4027x - 1.7307, R² = 0.6103.
  - CAB/GDP vs Nontradable FDI/GDP: y = -0.4222x - 4.8353, R² = 0.1922.
  - Market Size (log(GDP)) vs Share of Tradable FDI: y = 7.6039x + 16.206, R² = 0.4397.
  - Infrastructure (EBRD index) vs Share of Tradable FDI: y = 13.076x + 2.3915, R² = 0.2852.

### Imports, credit, and nontradable FDI (empirical results)
- Import equation (Appendix 2) highlights:
  - IM/Y(-1) coefficients: 0.502***; 0.555***; 0.503***; 1.001***; 0.983***; 0.973***; 0.953***; 0.895***.
  - log(REER) coefficients vary across specs: 0.156**; 0.120*; 0.175***; 0.0570; -0.188**; -0.0742; -0.155; -0.137.
  - FDI_nontradable/GDP shows mixed results with one specification reporting -0.00152** and other specifications small or insignificant (0.00124; 0.0000185; 0.000324).
  - FDI/GDP coefficient reported as -0.00140*** in one specification; -0.000768 in another.
- Correlations link higher nontradable FDI shares with larger import-to-GDP ratios and with credit growth, suggesting nontradable FDI can fuel domestic-demand-driven imports and private credit expansion.

### Descriptive statistics and data definitions
- Selected summary statistics (Obs, Mean, Std. Dev., Min, Max) include:
  - log(FDI_tradable): 126, 0.9, 1.6, -2.4, 5.0
  - log(FDI_non-tradable): 126, 1.0, 1.5, -2.9, 3.8
  - log(REER): 189, 4.6, 0.1, 4.0, 5.0
  - log(income per capita): 186, 2.7, 0.8, 0.7, 4.1
  - export/GDP: 186, 0.50, 0.20, 0.20, 0.90
  - import/GDP: 186, -0.60, 0.2, -1.0, -0.2
  - Wage: 132, 25.7, 15.0, 2.0, 77.0
  - Distance from Dusseldorf: 162, 1240, 313, 55, 91673
  - Infrastructure: 153, 2.7, 0.6, 1.3, 3.7
  - Education_tertiary: 142, 49.9, 17.4, 16.1, 85.5
  - Trade openness: 189, 105.3, 35.6, 0.0, 173.8
  - Restrictions on capital inflows: 158, 0.6, 0.3, 0.0, 1.0
  - Quality of bureaucracy: 153, 2.2, 0.9, 1.0, 4.0
- Data definitions and sources (selection):
  - FDI_tradable and FDI_nontradable: computed as total FDI times sectoral shares. Sources: WEO, WIIW Database on Foreign Direct Investment.
  - REER, EX/Y, IM/Y, GDP, trade integration: WEO.
  - Infrastructure: EBRD transition index.
  - Education_tertiary: WDI.
  - Distance to Dusseldorf: CIA factbook.
  - Restrictions on capital flows: Schindler (2009) index.
  - Privatization revenue: EBRD.

### Implications for external balances and policy recommendations
- Composition of FDI matters: excessive FDI in nontradable sectors can exacerbate external imbalances by fueling domestic demand and imports rather than supply.
- From 2003 onwards many countries saw FDI tilt toward nontradable sectors, contributing to large current account deficits that amplified vulnerability during the global financial crisis.
- Channels:
  - FDI in tradable sectors improves external balance mainly through the export channel; effect on imports is ambiguous because imports include intermediate inputs for export production.
- Policy recommendations to tilt FDI toward tradable sectors:
  - Short run: progress toward greater trade integration.
  - Medium to long run: address bottlenecks in infrastructure and upgrade human capital to create a level playing field favoring tradable-sector FDI.
- Additional observations:
  - Countries physically close to Western Europe have an inherent advantage as export platforms due to lower transportation/transaction costs.
  - Better institutional quality may not change sectoral composition but is likely to raise aggregate FDI.

*Source: _wp11123 - References (extracted chapter content).*

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

### _wp11123 - References .............................................................................................................

### Introduction
- Research questions:
  - Does the composition of FDI matter for current account deficits?
  - Can policies influence the sectoral composition of FDI?
- Scope:
  - Analysis of effects of sectoral distribution of FDI on trade balance via exports and imports in fifteen CESE countries in the run-up to the global financial crisis.
  - Empirical examination of determinants of FDI in the tradable sector across CESE countries.
  - Policy recommendations for host countries to affect sectoral allocation of FDI from the viewpoint of external stability and competitiveness.

### Capital inflows in the CESE countries — key observations
- Capital inflows into CESE were already high in 2003 and increased further thereafter, driven by the prospect of EU accession, ample liquidity, and strong world growth.
- FDI was generally the largest component of capital inflows in the region.
- Country- and subregion-specific patterns:
  - Bulgaria and Romania (EU Balkans) recorded the largest inflows of FDI relative to GDP.
  - The Baltics (Estonia, Latvia, and Lithuania) saw momentum upon EU accession in 2004.
  - Albania, Bosnia & Herzegovina, Croatia, Macedonia, and Serbia (Non-EU Balkans) experienced an increasing trend since 2005, mainly due to large-scale privatization.
  - CEE countries (Czech Republic, Slovak Republic, Hungary, Poland, and Slovenia) had a more moderate increase in FDI after 2003.
- Sectoral composition in 2007 varied markedly across CESE:
  - SEE countries: FDI dominated by nontradable sectors with exceptions (Macedonia and Romania).
  - Estonia and Latvia showed a similar nontradable-dominated pattern.
  - CEE countries exhibited a more balanced distribution between tradable and nontradable sectors.
- Definitions used in the study:
  - Tradable sectors = manufacturing, agriculture, mining, retail, hotels and restaurants.
  - Nontradable sectors = construction, electricity, transport, communication, real estate, and financial intermediation.
- Data caveat:
  - Intercompany loans are recorded as FDI in some countries, which may exaggerate the size of FDI inflows.

### How sectoral composition of FDI relates to trade deficits — cross-country evidence
- Conceptual links:
  - FDI in tradable sector likely increases exports over time.
  - FDI in nontradable sector may fuel domestic demand booms and boost imports; only short-run import effects likely from tradable FDI.
  - Countries with FDI predominantly in nontradable sectors likely have higher trade deficits than those with FDI in tradable sectors.
- Empirical cross-section findings:
  - Positive correlation between stock of FDI in tradable sector (percent of GDP) and export-to-GDP ratio; highest export-to-GDP ratios observed in Slovak Republic, Czech Republic, and Hungary, which also have high tradable FDI stock.
  - Positive correlation between stock of FDI in nontradable sector and import-to-GDP ratio; Bulgaria and Estonia have the highest stock of nontradable FDI and high import-to-GDP ratios.
  - Link between nontradable FDI and credit growth is positive; a large share of nontradable FDI is often financial intermediaries.

### How sectoral composition of FDI relates to trade deficits — time-series evidence
- Countries grouped into three broad categories based on time-series patterns of tradable FDI share and trade account balance:
  - Non-EU Balkans (Figure 5A): Generally little-changed trade balance since 2003 (except Bosnia and Herzegovina), while share of tradable FDI is generally declining.
  - Baltics and New Member States (Figure 5B): Negative correlation between share of tradable FDI and trade account balance; in Bulgaria, Romania and Latvia sharp increase in trade deficits coincided with declining share of tradable FDI.
  - CEE (Figure 5C): Czech Republic, Hungary, and Slovak Republic have high share of tradable FDI and improving trade balance; Poland and Slovenia show worsening trade balance as FDI increasingly flows to nontradable sectors.
- Medium-run inference:
  - More FDI in tradable sectors seems to improve the trade balance via export and import performance.

### Effects of tradable FDI on exports
- Widely held view: FDI can promote host-country export performance by augmenting domestic capital, transferring technology and new products, training the local workforce, and upgrading technical and managerial skills.
- Empirical context:
  - Example from other literature: Zhang (2005) finds that one dollar of FDI stock raises exports by about 70 cents in disaggregate industry-level data for China.
  - For CESE countries, the estimated link between tradable FDI and exports is substantially higher than those found in the Chinese study (not directly comparable due to different aggregation units).
  - Cross-country correlation coefficient suggests that one dollar of FDI in the tradable sector leads to an increase in exports by about [text truncated in source].

### Implications for policy and further analysis (as discussed)
- Sectoral composition matters for external vulnerability: policies that influence allocation of FDI toward tradable sectors can support external stability and competitiveness.
- Next analytical steps in the paper (as indicated in source):
  - Empirically examine determinants of FDI in the tradable sector.
  - Derive policy recommendations for host countries to affect sectoral allocation of FDI.
  - Suggest directions for future research.

*Source: _wp11123 - References .............................................................................................................*

### 3.5 in the CESE region. (See the upper panel chart on the next page). A one percentage point

### 3.5 in the CESE region. (See the upper panel chart on the next page). A one percentage point

### FDI in tradable sector and exports: empirical relationship
- A one percentage point of GDP increase in tradable FDI leads to about three times as much increase in exports (upper-left panel of Figure 4): reflected in the fitted line y =  3.4683x + 1.4941 with R² = 0.9508.
- Using aggregate FDI (including nontradable FDI), the positive relation between FDI and exports persists but is weaker (e.g., 1.8 dollar as opposed to 3.5).
- Mechanism: self-reinforcing effect where countries with a profitable exporting sector are more likely to attract more FDI in the tradable sector.
- Appendix 1: panel data econometrics show a positive link between export performance and FDI in the tradable sector after controlling for real exchange rates and market size.

### Cross-country patterns (2003–2007)
- General increase in export propensity across the region between 2003 and 2007, with large cross-country variation in export-to-GDP ratios.
- Top three exporters in 2007 are CEE countries that embarked on transition early.
- Export and tradable FDI highlights:
  - Exports of CEE countries (except Poland) account for about 70 percent of GDP.
  - FDI stock in the tradable sector in these CEE countries accounts for over 15 percent of GDP.
  - Countries with little or no increase in export-to-GDP ratio: Albania, Croatia, Latvia, Lithuania, Romania, Serbia — these have lower tradable FDI shares.
  - Macedonia and Bosnia & Herzegovina (SEE) saw significant improvement in export performance and also a high share of tradable FDI.

### Determinants of sectoral (tradable) FDI — host country factors
- Objective: explain share of tradable FDI to total FDI and identify host-country policies that tilt FDI toward tradable (export-oriented) sectors.
- Initial independent variables used in panel regressions (following Campos and Kinoshita (2003)):
  - (log of) GDP (market size)
  - Income per capita
  - Wage
  - Education (tertiary education enrollment rate)
  - Availability of infrastructure (composite index from EBRD)
  - Trade integration (exports + imports as share of GDP)
  - Quality of bureaucracy and rule of law (ICRG indexes)
  - Distance from Western Europe (distance in kilometers from Dusseldorf to capital city)
- Additional controls for transition initial conditions:
  - Share of industry in 1989
  - Dummy for early transition (equals one if started transition in 1991 or earlier, zero otherwise)
- Policy variables included:
  - Restrictions on capital inflows (Schindler (2009) index)
  - Privatization revenue as share of GDP (EBRD)
  - Overall fiscal balance to GDP
  - Inflation rate (as proxy for monetary stability)

### Key empirical findings (panel regression results)
- Positive and significant determinants of a larger share of tradable FDI:
  - Larger market size (GDP)
  - Sufficient infrastructure (positive and significant coefficient throughout regressions)
  - Greater trade openness
  - Highly educated labor force (tertiary enrollment)
- Negative and significant determinant:
  - Distance to Dusseldorf (countries closer to Germany receive more tradable FDI)
- Insignificant or non-supportive determinants:
  - Wage (low labor cost): foreign investors value labor quality (education) over low wage per se.
  - Institutional quality (quality of bureaucracy, rule of law): regression results do not support a role for institutional quality in determining the sectoral composition of FDI (although better institutions likely increase aggregate FDI).
  - Initial conditions (share of industry in 1989, early-transition dummy): coefficients insignificant or with unexpected signs.
  - Policy variables (capital controls on inflows, privatization revenues, fiscal balance, inflation): statistically insignificant for the sectoral distribution of FDI.
    - Capital controls on inflows do not appear to deter tradable-sector FDI in the sample.
    - Privatization revenue is statistically insignificant, including when splitting sample into Western Balkans and other countries.
- Interpretation: results consistent with hypothesis of vertical FDI (outsourcing by Western manufacturers), where proximity to Western manufacturers, infrastructure, market size, trade openness, and labor quality matter most.

### Implications for external balances and policy recommendations
- Composition of FDI matters: excessive FDI in nontradable sectors can exacerbate external imbalances by fueling domestic demand and imports rather than supply.
- From 2003 onwards many countries saw FDI tilt toward nontradable sectors, contributing to large current account deficits that amplified vulnerability during the global financial crisis.
- Channel: FDI in tradable sectors improves external balance mainly through the export channel (effect on imports is ambiguous because imports include intermediate inputs for export production).
- Policy recommendations to tilt FDI toward tradable sectors:
  - Short run: progress toward greater trade integration.
  - Medium to long run: address bottlenecks in infrastructure and upgrade human capital to create a level playing field favoring tradable-sector FDI.
- Additional observations:
  - Countries physically close to Western Europe have an inherent advantage as export platforms due to lower transportation/transaction costs.
  - Better institutional quality may not change sectoral composition but is likely to raise aggregate FDI.

_Italic: Source: IMF WEO Database; WIIW Database on Foreign Direct Investment; extracted from the supplied chapter content._

### References

### _wp11123 - References

### Key literature cited
- Works on FDI determinants, financial flows, and transition economies, including:
  - Abiad, A., D. Leigh, and A. Mody (2007), “International Finance and Income Convergence: Europe is Different,” IMF Working Paper 07/64.
  - Alfaro, L., A. Chanda, S. Kalemli-Ozcan, and S. Sayek (2004), “FDI and Economic Growth: the Role of Local Financial Markets,” Journal of International Economics, vol. 64(1), pages 89-112.
  - Bakker, B. and A. M. Gulde (2010), “The Credit Boom in the EU New Member States: Bad Luck or Bad Policies?” IMF Working Paper 10/130.
  - Goldberg, L. (2004), “Financial-Sector FDI and Host Countries: New and Old Lessons,” Federal Reserve Bank of New York Staff Report No. 183.
  - Mody, A. (2004), “Is FDI Integrating the World Economy?” The World Economy, Vol. 27, No. 8, pp. 1195-1222.
  - Ostry, J.D., A. Ghosh, K. Habermeier, M. Chamon, M. S. Qureshi, and D. B. S. Reinhardt (2010), Capital Inflows: The Role of Controls. IMF Staff Position Note SPN/10/04 (February 19).
- Data sources and institutional reports cited include IMF WEO Database, WIIW Database on Foreign Direct Investment, EBRD, World Bank, NBER, and various IMF Working Papers and Staff Papers.

### Empirical findings from figures and regressions
- FDI patterns and composition (figures):
  - Figure 1: FDI Inflow in Emerging Economies, 2000-08 (Percent of GDP) — regional groups defined in footnote 1.
  - Figures 2–6: Country-level distributions of FDI across tradable vs nontradable sectors and correlations with export/GDP, import/GDP, change in private credit/GDP, CAB/GDP, education, market size (log(GDP)), trade openness, distance (log distance to Dusseldorf), EBRD infrastructure index, and wage (US=100).
- Determinants of FDI in tradable sectors (Table 1: Dependent variable = FDI_tradable/FDI):
  - ln(GDP) reported coefficients across specifications: 0.105***; 1.304*; 0.0873***; 0.0873***; 0.134***; 0.287**; 0.0997***; 0.0755***; 0.0616*.
  - infrastructure coefficients: 0.0820***; 0.0608***; 0.169***; 0.169***; 0.211***; 0.194***; 0.181***; 0.249***; 0.202***.
  - Trade integration coefficients: 0.00204***; 0.000979; 0.00199***; 0.00199***; 0.00276***; 0.00457***; 0.00150*; 0.00295***; 0.00182***.
  - Education_tertiary coefficients: 0.00165**; 0.00124; 0.00354**; 0.00354**; 0.00372*; 0.00407**; 0.00205; 0.00731**; 0.00236.
  - Distance to Dusseldorf coefficients: -0.192*; -0.192*; -0.401**; -0.397**; -0.0921; -0.401*; -0.0957.
  - Quality of bureaucracy shows mixed signs with some significant negative coefficients: -0.140***; -0.140***; -0.170***; -0.256**; -0.126** in select specifications.
  - Estimation methods: RE, FE, and multiple Sys-GMM specifications. Number of observations ranges (e.g., Observations: 89, 77, 76, 73); Number of id: 13, 11, 11, 11.
  - Notes: ***, **, and * indicate 1%, 5%, and 10% significance levels, respectively. System-GMM uses lagged dependent variable and secondary education as endogenous IVs; exogenous IVs include FDI stock in the region, political risk, quality of bureaucracy, rule of law, inflation, and trade integration.
- Export equation (Appendix 1: Dependent variable = EX/Y):
  - Lagged dependent variable EX/Y(-1) coefficients: 0.548***; 0.570***; 0.583***; 0.866***; 0.872***; 0.866***; 0.962***; 0.804***.
  - log(REER) coefficients include negative and significant values in some specs: -0.125**; -0.107**; -0.108*; -0.141*; -0.344**.
  - FDI_tradable interactions: FDI_tradable/GDP coefficients: 0.00239**; 0.00388**; 0.00371**; 0.00428***. FDI_tradable/FDI coefficients: 0.193***; 0.106*; 0.189***.
  - FDI/GDP coefficient in one specification: 0.00214**; another: 0.000722 (not significant).
  - Estimation methods include Fixed Effects and Sys-GMM. Observations and R-squared: e.g., Observations 96, 115, 96; R-squared up to 0.746; Number of id varies (15, 17, 15, 13).
- Import equation (Appendix 2: Dependent variable = IM/Y):
  - IM/Y(-1) coefficients: 0.502***; 0.555***; 0.503***; 1.001***; 0.983***; 0.973***; 0.953***; 0.895***.
  - log(REER) coefficients vary: 0.156**; 0.120*; 0.175***; 0.0570; -0.188**; -0.0742; -0.155; -0.137.
  - FDI_nontradable/GDP shows: -0.00152** in one specification; other specifications report small or insignificant values (0.00124; 0.0000185; 0.000324).
  - FDI/GDP coefficient: -0.00140*** in one specification; -0.000768 in another.
  - Estimation methods include Fixed Effects and Sys-GMM. Observations and R-squared: e.g., Observations 88, 107, 88; R-squared up to 0.770; Number of id varies (15, 17, 15, 12).
- Correlations and scatter relationships (Figure 4 and Figure 6):
  - Several fitted lines and R² reported in figures, e.g.:
    - Export/GDP vs Tradable FDI/GDP: y = 3.269x + 20.088, R² = 0.5736.
    - Other investment/GDP vs Nontradable FDI/GDP: y = 0.1596x + 5.9155, R² = 0.0308.
    - Import/GDP vs Nontradable FDI/GDP: y = 0.8709x + 53.258, R² = 0.1766.
    - Change in Private Credit/GDP vs Nontradable FDI/GDP: y = 0.4027x - 1.7307, R² = 0.6103.
    - CAB/GDP vs Nontradable FDI/GDP: y = -0.4222x - 4.8353, R² = 0.1922.
    - Change in Private Credit/GDP vs Nontradable FDI/GDP: y = 0.6466x + 15.134, R² = 0.1343.
    - Education (Tertiary) vs Share of Tradable FDI: y = 0.063x + 37.455, R² = 0.0066.
    - Market Size (log(GDP)) vs Share of Tradable FDI: y = 7.6039x + 16.206, R² = 0.4397.
    - Trade Openness vs Share of Tradable FDI: y = 0.1195x + 26.833, R² = 0.0855.
    - Distance (log distance to Dusseldorf) vs Share of Tradable FDI: y = -20.701x + 188.54, R² = 0.1894.
    - Infrastructure (EBRD index) vs Share of Tradable FDI: y = 13.076x + 2.3915, R² = 0.2852.
    - Wage (US=100) vs Share of Tradable FDI: y = -0.3256x + 49.131, R² = 0.0392.

### Descriptive statistics (Appendix 3)
- Selected variable summary (Obs, Mean, Std. Dev., Min, Max):
  - log(FDI_tradable): 126, 0.9, 1.6, -2.4, 5.0
  - log(FDI_non-tradable): 126, 1.0, 1.5, -2.9, 3.8
  - log(REER): 189, 4.6, 0.1, 4.0, 5.0
  - log(income per capita): 186, 2.7, 0.8, 0.7, 4.1
  - export/GDP: 186, 0.50, 0.20, 0.20, 0.90
  - import/GDP: 186, -0.60, 0.2, -1.0, -0.2
  - Wage: 132, 25.7, 15.0, 2.0, 77.0
  - Distance from Dusseldorf: 162, 1240, 313, 55, 91673
  - Infrastructure: 153, 2.7, 0.6, 1.3, 3.7
  - Share of industry in 1989: 144, 0.5, 0.1, 0.4, 0.6
  - Education_tertiary: 142, 49.9, 17.4, 16.1, 85.5
  - Trade openness: 189, 105.3, 35.6, 0.0, 173.8
  - Restrictions on capital inflows: 158, 0.6, 0.3, 0.0, 1.0
  - Quality of bureaucracy: 153, 2.2, 0.9, 1.0, 4.0

### Data definitions and sources (Appendix 4)
- Variable definitions and sources (selection):
  - FDI_tradable: FDI in the tradable sectors (= total FDI* share of tradable FDI). Sources: WEO, WIIW Database on Foreign Direct Investment.
  - FDI_nontradable: FDI in the non-tradable sectors (=total FDI* share of nontradable FDI). Sources: WEO, WIIW Database on Foreign Direct Investment.
  - REER: Real effective exchange rate. Source: WEO.
  - EX/Y: Export to GDP ratio. Source: WEO.
  - IM/Y: Import to GDP ratio. Source: WEO.
  - EU income: EU-15 real GDP. Source: WEO.
  - Income, GDP, Income per capita: WEO definitions (GDP in constant/current prices, in US dollars; GDP per capita index (euro area = 100)).
  - Wage: Average gross wage in euro, index (US=100). Source: Haver Analytics.
  - Infrastructure: EBRD transition index of infrastructure reform. Source: EBRD.
  - Education_tertiary: Gross enrollment rate of the tertiary school. Source: WDI.
  - Quality of bureaucracy: ICRG measure of bureaucracy quality.
  - Distance to Dusseldorf: Great circle distance from Dusseldorf to the capital city (km). Source: CIA factbook.
  - Share of industry in 1989: Share of Industrial production in 1989. Source: De Melo and others (1997).
  - Dummy for early transition: =1 if year of introducing stabilization program is before 1991, = 0 otherwise. Source: Fisher and Sahay (2000).
  - Restrictions on capital flows: Index on capital controls on inflows. Source: Schindler (2009).
  - Privatization: Privatization revenue as a share of GDP. Source: EBRD.
  - Fiscal balance: General government balance as a share of GDP. Source: WEO.
  - Trade integration: =(Exports + Imports)/GDP. Source: WEO.
  - Additional note: Quality of bureaucracy described as "The extent to which the bureaucracy has the strength and expertise to govern without drastic changes in policy or interruptions in government services."

*Italic: Source document: _wp11123 - References (IMF PDF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11123.pdf_