## _wp05189

## Source details

**Canonical URL:** [_wp05189](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05189.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05189.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05189.pdf.json)

---

### Introduction and key findings
- China’s share of FDI in developing Asia rose from about 10 percent in the early 1980s to over 50 percent by the early 1990s.
- China was the world’s top destination for FDI in 2002.
- Dataset covers 14 Asian economies from 1984 to 2002.
- Main empirical findings:
  - The average diversion of FDI flows across countries is significant.
  - Crowding out by China was concentrated on two countries only: Singapore and Myanmar.
  - Other countries do not appear to have been affected.
  - There is no evidence that low-income economies suffered more than higher-income ones.
  - Low investment in education or in sciences is not associated with increased crowding out by China.

### Context and motivations
- Rationale for studying FDI:
  - FDI is viewed as less volatile than other flows, can boost technology transfers and growth, and can raise domestic investment.
- Source composition:
  - Hong Kong SAR, Taiwan Province of China, and Singapore account for over 50 percent of total FDI in China (noted as “over 50 percent” in the source).
- Alternative channels:
  - China’s development may increase foreign investment in suppliers of inputs (raw materials, components) to China.
- Prior literature referenced:
  - McKibbin and Woo (2003); Chantasasawat et al. (2004) (the latter find China’s foreign investment positively related to other economies’ inward direct investment; the paper critiques that assessment).

### Estimation methodology and challenges
- Reduced-form baseline specification used:
  - Equation (1): fdi_{j,t} = X_{j,t} β + ε_{j,t}
- Key econometric concerns:
  - Logarithmic specifications assume crowding out depends on rates of change rather than levels, which can be misleading.
  - Using a country’s share of regional FDI on the LHS and log China FDI on RHS mechanically produces a negative coefficient as China’s FDI grows.
  - Nominal FDI is non-stationary and needs scaling.
  - Scaling by GDP (China or LHS country) introduces strong implicit assumptions and potential endogeneity.
  - Common shocks or trends can bias estimates.
- Two preferred indicators of FDI to China for estimating diversion:
  1. FDI to China over combined GDP of other countries in the region (FDI_China,t / sum_{all countries\China} GDP_j,t)
     - Rationale: assumes diversion (in absolute amount) is proportional to economy size relative to the region.
     - Potential issue: estimated α may capture common shocks or trends; instruments used to address endogeneity.
  2. China’s share of total regional FDI (FDI_China / total regional FDI)
     - Rationale: if China’s FDI diverts flows from others, China’s share should enter with a negative sign; if not, it should be insignificant.
     - Advantage: eliminates some biases from common shocks or trends affecting both China and other nations.
     - Caveats: China’s share is theoretically endogenous because the denominator includes the LHS country’s FDI; addressed with instrumental variables.
- Country-specific diversion estimates also use FDI to China scaled by the GDP of the LHS country.

### Empirical specification and estimation details
- Main estimating equation:
  - Equation (5): fdi_{j,t} = α China_{t-1} + β X_{j,t-1} + μ_j + δ fdi_{j,t-1} + ε_{j,t}
  - Dimensions: j = 1, ..., 13 and t = 1984, ..., 2002.
- Dependent variable:
  - fdi_{j,t}: FDI to GDP ratio of country j at time t.
- Regressors (lagged one period):
  - Inflation
  - Budget balance (as share of GDP)
  - Official reserves (as share of GDP)
  - GDP growth (real)
  - GDP per capita as a share of the United States’ (proxy for relative wage levels)
  - Change in real exchange rate
  - Ratio of money over GDP
  - Corruption
  - Socioeconomic conditions
  - Government stability
  - Business environment
  - Country dummies
  - Bond yields in the G3 (weighted average; proxy for opportunity cost of investing in Asia)
  - Lagged FDI
- Human capital and technology variables (in separate specifications due to data limits):
  - Rates of secondary and tertiary education
  - Publications in scientific journals per capita
- Timing: all regressors lagged once to reflect implementation delays and to reduce endogeneity.

### Panel estimation issues and instruments
- Primary estimator: OLS with fixed country effects.
- Heteroscedasticity testing:
  - White test and Breusch-Pagan test performed; both always reject heteroscedasticity in the source’s analysis.
  - Robust OLS and GLS estimates were very similar to OLS; therefore OLS reported.
- Dynamic panel concerns:
  - Lagged FDI correlated with disturbance; judged unlikely to be significant given time dimension (1984–2002).
  - Arellano and Bond (1991) estimator used as robustness check.
- Endogeneity of China indicators:
  - Instruments: lagged values of corruption, business environment, socioeconomic conditions, and government stability in China.
  - Reported average correlation of instruments with China indicator: about 0.45 in absolute value.

### Data sources
- Main data sources:
  - IMF’s International Financial Statistics (IFS)
  - IMF’s World Economic Outlook (WEO)
  - World Bank’s World Development Indicators (WDI)
  - PRS Group indicators for corruption, investment conditions, government stability, and socioeconomic conditions
- Variable notes (selected):
  - Corruption: From PRS group. Ranges from 0 (most corrupt) to 6.
  - Government Stability: From PRS group. Ranges from 0 (least stable) to 12.
  - Scientific publications: Publications in scientific journals per capita, from WDI. (Does not include Taiwan Province of China. Data available from 1987 to 2000.)
  - Secondary and tertiary education: from WDI. (Data does not include Taiwan Province of China. Data covers 1991-2001, missing for some countries after 1997.)

### Estimation results — overview
- Sample:
  - Covers all Asian economies with population greater than 4 million except Japan: Bangladesh, China, India, Indonesia, Korea, Malaysia, Myanmar, Papua New Guinea, Philippines, Singapore, Sri Lanka, Taiwan Province of China, Thailand, and Vietnam.
  - Annual observations for 1984-2002.
- Primary question: whether China’s ability to attract FDI diverted flows away from other countries in the region.

A. Average diversion
- IV estimates on the coefficients on two indicators of flows to China are negative and significant at the 5 percent level.
- A 10 percent increase in China’s FDI market share appears to have lowered annual flows to other nations by about 0.4 percent of GDP on average.
- China’s market share rose from an average of 26 percent in the period 1984-91 to one of 56 percent in 1992-2002; China’s negative impact on flows to other countries was around 1.3 percent of GDP a year on average.
- Specification using FDI to China over total GDP of other countries suggests the increase in average flows to China before and after 1991 reduced flows to other nations by about 2.1 percent of GDP a year on average.

B. Diversion by country
- IV estimates indicate crowding out concentrated on Singapore and Myanmar; other countries not significantly affected.
- Results robust across three specifications of FDI to China:
  - China’s share of total FDI to the region.
  - FDI to China over total GDP of other countries.
  - FDI to China over the GDP of the LHS country.
- Estimated reductions associated with the increase of average flows to China before and after 1991 (similar across specifications):
  - Singapore: estimated reduction of flows is 2.8, 1.9 to 2.3, and 2.0 percent of GDP a year using respectively (i) China’s share of total FDI to the region, (ii) FDI to China over total GDP of other countries, and (iii) FDI to China over the GDP of the LHS country.
  - Myanmar: estimated reduction of flows is 3.9, 4.2, and 4.5 percent of GDP a year for the same three specifications, respectively.
- Additional notes:
  - China’s impact on Indonesia appears significant at the 10 percent level when using FDI to China over the GDP of the LHS country as regressor, but not significant with the other two indicators; impact not significant when allowing for four levels of diversion.
  - Interpreting coefficient magnitudes:
    - A 10 percent increase in China’s FDI market share appears to have lowered annual flows to Singapore and Myanmar by about 0.9 and 1.3 percent of GDP, respectively (per one reported coefficient interpretation).
    - FDI to China over the GDP of the LHS country: an increase in flows to China by 1 billion dollars reduces flows to Singapore by 57 million dollars and to Myanmar by 8 million dollars (as reported in the source).

C. Impact of GDP per Capita, Sciences and Education on diversion
- Interaction results:
  - Low-income economies do not appear to have suffered more from diversion than higher income ones.
  - Low levels of secondary or tertiary education or fewer scientific publications per person are not associated with increased crowding out by China.
- Robustness:
  - OLS and Arellano-Bond estimates of the impact of scientific publications on diversion are significant for one indicator of flows to China, but IV estimate is not significant and no estimate is significant with the other indicator of flows to China.
  - Significant coefficients on scientific publications become insignificant when dropping Singapore from the sample (Singapore has the highest number of scientific publications per capita).

D. Other determinants of FDI
- Economic fundamentals associated with increased FDI:
  - Healthy government balances.
  - An appreciating real exchange rate.
  - Low inflation.
  - Low interest rates in the G3 have a positive impact on flows to the region (G3 bond yield negative coefficient implies lower G3 yields increase regional FDI).
- Openness (sum of imports and exports over GDP) appears associated with lower levels of FDI; possible explanation: openness correlates with sophistication of financial system, which may be negatively correlated with FDI.
- Scientific publications may increase FDI per OLS and Arellano-Bond but result driven by Singapore only; excluding Singapore removes significance.

### Selected quantitative results (high-level table highlights)
- Lagged FDI coefficients (representative significance):
  - OLS 0.309*** (0.000); A-Bond 0.276*** (0.000); IV 0.296*** (0.000) in average-impact regressions.
- China indicator (effect on other economies’ FDI):
  - China’s share of total FDI to the region: OLS -0.025** (0.013); A-Bond -0.024** (0.028); IV -0.042** (0.017).
  - FDI to China / total GDP of other economies: IV -1.156** (0.022) in one specification.
- Country fixed effects (selected, China’s share regressions, OLS):
  - Singapore 0.136*** (0.000)
  - Myanmar 0.091*** (0.000)
  - Vietnam 0.084*** (0.000)
- Table-specific notable coefficients:
  - Myanmar (China’s share regressions): OLS -0.092*** (0.009), IV -0.132*** (0.005); FDI to China / total GDP of other economies: OLS -1.581** (0.029), IV -2.319*** (0.004).
  - Singapore (China’s share regressions): OLS -0.098*** (0.001), IV -0.094*** (0.006).

### Conclusion and policy implications
- Main conclusions:
  - Using data from 14 Asian economies from 1984 to 2002, the paper finds little evidence that China’s success in attracting FDI has been at the expense of other countries in the region, with the notable exceptions of Singapore and Myanmar.
  - Low-wage economies do not appear to have been particularly affected by China’s emergence.
  - Low levels of education or scientific development are not associated with increased crowding out by China.
  - Some economic fundamentals help explain allocation of FDI flows among Asian countries.
- Implications for policy (inferred from determinants):
  - Countries might attract more FDI by improving fundamentals shown to be associated with higher FDI: healthier government balances, stable and appreciating real exchange rates, low inflation, and benefiting from lower international interest rates.
- Suggested extensions (as stated in the source):
  - Assess whether the increase in FDI to China had an impact on particular sectors.
  - Estimate how FDI to China enhances regional trade as investment in China spurs demand for capital goods from other nations.

*Source: _wp05189 (excerpts) — IMF working paper content unit provided in the input.*

### 1. FDI to China, 1982-2002..............................................................................................

### 1. FDI to China, 1982-2002

### Introduction and key findings
- China’s share of FDI in developing Asia rose from about 10 percent in the early 1980s to over 50 percent by the early 1990s.
- China was the world’s top destination for FDI in 2002.
- Dataset covers 14 Asian economies from 1984 to 2002.
- Main empirical findings:
  - The average diversion of FDI flows across countries is significant.
  - Crowding out by China was concentrated on two countries only: Singapore and Myanmar.
  - Other countries do not appear to have been affected.
  - There is no evidence that low-income economies suffered more than higher-income ones.
  - Low investment in education or in sciences is not associated with increased crowding out by China.

### Context and motivations
- FDI considered valuable because it is less volatile than other flows, can boost technology transfers and growth, and can raise domestic investment.
- Large share of FDI to China comes from overseas Chinese (Hong Kong SAR, Taiwan Province of China, and Singapore account for over 50 percent of total FDI in China).
- Possible channels besides diversion: China’s development may increase foreign investment in suppliers of inputs (raw materials, components) to China.
- Previous literature: McKibbin and Woo (2003); Chantasasawat et al. (2004) — the latter find China’s foreign investment positively related to other economies’ inward direct investment but the paper critiques their assessment.

### Figures cited
- Figure 1: FDI to China, 1982-2002 (in billions of U.S. dollars).
- Figure 2: FDI to China as a Percent of Emerging Asia’s Total, 1982-2002.

---

### Estimation methodology and challenges
- No generally accepted theory of FDI; many studies adopt a reduced-form specification:
  - Equation (1): fdi_{j,t} = X_{j,t} β + ε_{j,t} (notation as in source).
- Key econometric concerns when estimating crowding out:
  - Logarithmic specifications (log FDI to China) assume crowding out depends on the rate of change of FDI to China rather than the level — this is problematic (example contrasts 500 million to 1 billion vs. 10 billion to 20 billion).
  - Using country’s share of regional FDI on LHS and log China FDI on RHS mechanically produces a negative coefficient as China’s FDI grows; this is not conclusive evidence of diversion.
  - Nominal FDI is non-stationary and needs scaling.
  - FDI-to-GDP ratio for China introduces the assumption that crowding out is inversely related to size of Chinese economy.
  - Scaling by GDP of the LHS country implies China diverts the same dollar amount from all countries regardless of size; endogeneity of denominators and common shocks can bias estimates upward.
- Two suggested indicators of FDI to China for estimation:
  1. FDI to China over the combined GDP of other countries in the region:
     - Rationale: assumes diversion (in absolute amount) is proportional to the size of an economy relative to the region.
     - Derivation summarized from equations (2)–(4) in the source; leads to regressing (FDI_j,t / GDP_j,t) on (FDI_China,t / sum_{all countries\China} GDP_j,t) to estimate crowding out parameter α.
     - Potential issue: estimated α might capture common shocks or trends on FDI or GDP on both sides; instruments used to address this.
  2. FDI to China over total FDI to the region (China’s share of total regional FDI):
     - Rationale: if China’s FDI diverts flows from others, China’s share should enter with a negative sign; if not, it should be insignificant.
     - Advantage: eliminates some biases from common shocks or trends affecting both China and other nations.
     - Caveats: assumes diversion decreases when total regional flows increase; China’s share is theoretically endogenous (denominator includes LHS country’s FDI) — addressed with instrumental variables.
- For country-specific diversion estimates, also use FDI to China scaled by the GDP of the LHS country.

---

### Empirical specification and estimation details
- Main estimating equation (as in source):
  - Equation (5): fdi_{j,t} = α China_{t-1} + β X_{j,t-1} + μ_j + δ fdi_{j,t-1} + ε_{j,t}
  - Notation in source: ,,1,,jtjtjttjjt fdifdiXChinaδβα με − =++ ++  (presentation preserved as in source).
  - Dimensions reported in source: (j=1, ...,13 and t=1984,...,2002).
- Dependent variable:
  - fdi_{j,t}: FDI to GDP ratio of country j at time t.
- Regressors (X_{j,t}) include lagged determinants suggested in literature:
  - Inflation
  - Budget balance (as share of GDP)
  - Official reserves (as share of GDP)
  - GDP growth (real)
  - GDP per capita as a share of the United States’ (proxy for relative wage levels)
  - Change in real exchange rate
  - Ratio of money over GDP
  - Corruption
  - Socioeconomic conditions
  - Government stability
  - Business environment
  - Country dummies
  - Bond yields in the G3 (weighted average; proxy for opportunity cost of investing in Asia)
  - Lagged FDI (to capture serial correlation)
- Human capital and technology variables (in a separate model due to data limits):
  - Rates of secondary and tertiary education
  - Publications in scientific journals per capita
- Timing of regressors:
  - All regressors are lagged once to reflect delays between investment decisions and implementation and to reduce endogeneity.

### Panel estimation issues and instruments
- Primary estimation: OLS with fixed country effects.
- Tests for heteroscedasticity:
  - White test and Breusch-Pagan test performed; both always reject heteroscedasticity in the source’s analysis.
  - Robust OLS and GLS estimates were very similar to OLS; therefore OLS reported.
- Dynamic panel concern:
  - Lagged FDI correlated with disturbance; bias deemed unlikely to be significant given time dimension (1984–2002), but Arellano and Bond (1991) dynamic panel estimator used as a robustness check.
- Endogeneity of China indicators:
  - Instruments used: lagged values of corruption, business environment, socioeconomic conditions, and government stability in China.
  - Rationale: these are exogenous to developments in LHS countries and correlated with the indicator of FDI flows to China.
  - Reported average correlation of instruments with China indicator: about 0.45 in absolute value.

### Data sources
- Main data sources listed:
  - IMF’s International Financial Statistics (IFS)
  - IMF’s World Economic Outlook (WEO)
  - World Bank’s World Development Indicators (WDI)
  - PRS Group indicators for corruption, investment conditions, government stability, and socioeconomic conditions.

*Source: _wp05189 - 1. FDI to China, 1982-2002 (excerpts) from the provided IMF PDF content.*

### appendix presents the data in greater detail.

### _wp05189 - appendix presents the data in greater detail.

### III. ESTIMATION RESULTS — Overview
- Study covers all Asian economies with a population greater than 4 million, except Japan: Bangladesh, China, India, Indonesia, Korea, Malaysia, Myanmar, Papua New Guinea, Philippines, Singapore, Sri Lanka, Taiwan Province of China, Thailand, and Vietnam.
- Sample includes annual observations for the period 1984-2002.
- Main question: whether China’s ability to attract FDI diverted flows away from other countries in the region.

### A. Average Diversion
- IV estimates on the coefficients on two indicators of flows to China are negative and significant at the 5 percent level of confidence.
- A 10 percent increase in China’s FDI market share appears to have lowered annual flows to other nations by about 0.4 percent of GDP on average.
- China’s market share rose from an average of 26 percent in the period 1984-91 to one of 56 percent in 1992-2002; China’s negative impact on flows to other countries was around 1.3 percent of GDP a year on average.
- Specification using FDI to China over total GDP of other countries suggests the increase in average flows to China before and after 1991 reduced flows to other nations by about 2.1 percent of GDP a year on average.

### B. Diversion by Country
- IV estimates suggest crowding out by China was concentrated on two countries only — Singapore and Myanmar — while other countries have not been significantly affected.
- Results are robust to three specifications of FDI flows to China: China’s share of total FDI to the region; FDI to China over total GDP of other countries; FDI to China over the GDP of the LHS country.
- To conserve degrees of freedom, equation (5) estimated with only three levels of diversion: one for Singapore, one for Myanmar, and average diversion across other nations. Results confirm China’s emergence does not seem to have significantly reduced FDI to countries other than Singapore and Myanmar.
- Estimated reductions associated with the increase of average flows to China before and after 1991 (broadly similar across three specifications):
  - Singapore: estimated reduction of flows is 2.8, 1.9 to 2.3, and 2.0 percent of GDP a year using respectively (i) China’s share of total FDI to the region, (ii) FDI to China over total GDP of other countries, and (iii) FDI to China over the GDP of the LHS country.
  - Myanmar: estimated reduction of flows is 3.9, 4.2, and 4.5 percent of GDP a year for the same three specifications, respectively.
- Additional notes/robustness:
  - China’s impact on Indonesia appears significant at the 10 percent level when using FDI to China over the GDP of the LHS country as regressor, but not significant with the other two indicators; impact also not significant when allowing for four levels of diversion (Indonesia, Singapore, Myanmar, and average other nations).
  - Interpretation of coefficients: the coefficient on China’s share of total GDP to the region suggests a 10 percent increase in China’s FDI market share appears to have lowered annual flows to Singapore and Myanmar by about 0.9 and 1.3 percent of GDP, respectively.
  - Coefficient on FDI to China over the GDP of the LHS country suggests an increase in flows to China by 1 billion dollars reduces flows to Singapore by 57 million dollars and to Myanmar by 8 million.
  - Coefficient on FDI to China over total GDP of other countries has a less intuitive economic interpretation.

### C. Impact of GDP per Capita, Sciences and Education on Diversion
- Investigated whether low GDP per capita as a share of China’s, rates of secondary and tertiary education, and publications in scientific journals per capita are associated with increased diversion by adding interaction terms (indicator of FDI to China multiplied by corresponding variable).
- Results (Tables 4 to 7):
  - Low-income economies do not appear to have suffered more from diversion than higher income ones.
  - Low levels of secondary or tertiary education or fewer scientific publications per person are not associated with increased crowding out by China.
  - OLS and Arellano-Bond estimates of the impact of scientific publications on diversion are significant for one indicator of flows to China, but:
    - IV estimate is not significant.
    - No estimate significant with the other indicator of flows to China.
    - Significant coefficients become insignificant when dropping Singapore from the sample (Singapore has the highest number of scientific publications per capita).

### D. Other Determinants of FDI
- Economic fundamentals associated with increased FDI:
  - Healthy government balances.
  - An appreciating real exchange rate.
  - Low inflation.
  - Low interest rates in the G3 have a positive impact on flows to the region.
- Openness (sum of imports and exports over GDP) appears associated with lower levels of FDI; possible explanation: openness correlates with sophistication of financial system, which may be negatively correlated with FDI (as argued by Hausmann and Fernandez-Arias (2000)).
- Real exchange rate appreciation increases FDI flows; possible interpretation: appreciation may signal further appreciation that would bring capital gains to foreign investors.
- OLS and Arellano-Bond (but not IV) estimates suggest scientific publications may increase FDI; result driven by Singapore only — excluding Singapore, variable no longer significant.

### IV. CONCLUSION
- Using data from 14 Asian economies from 1984 to 2002, the paper finds little evidence that China’s success in attracting FDI has been at the expense of other countries in the region, with the notable exceptions of Singapore and Myanmar.
- Low-wage economies do not appear to have been particularly affected by China’s emergence.
- Low levels of education or scientific development are not associated with increased crowding out by China.
- Some economic fundamentals help explain the allocation of FDI flows among Asian countries, suggesting countries might attract more FDI by improving these fundamentals.
- Suggested extensions: assess whether the increase in FDI to China had an impact on particular sectors, and estimate how FDI to China enhances regional trade as investment in China spurs demand for capital goods from other nations.

### APPENDIX I. Data Sources — Variable definitions and sources
- FDI: Mostly from IFS. Missing values replaced by Lane-Milesi-Ferretti database when available. 1999-2002 figures for Taiwan Province of China come from the WEO database.
- Corruption: From PRS group. Ranges from 0 (most corrupt) to 6.
- Government Stability: From PRS group. Ranges from 0 (least stable) to 12.
- Socioeconomic conditions: From PRS group. Reflects unemployment, consumer confidence, poverty. Ranges from 0 (worst conditions) to 12.
- Business environment: From PRS group. Reflects contract viability, expropriation risk, profits repatriation and payment delays. Ranges from 0 (worst conditions) to 12.
- Openness: Computed as the sum of imports and exports over GDP. Computed using data from WEO.
- GDP per capita as a share of China’s: Computed using data from IFS.
- RER change: Computed as the rate of change of real exchange rate. Real exchange rate calculated as the ratio of the product of nominal exchange rate by U.S. consumer price index (CPI) to the country’s CPI. Data from WEO.
- Reserves: Reserves as share of GDP, using data from WEO.
- Growth: Real GDP growth, computed using WEO’s GDP in constant prices.
- Inflation: Computed as the rate of change of the CPI.
- Government balance: Computed using WEO data as central government balance over GDP. For Thailand, general government balance used to make up for lack of data.
- Money: Computed as money supply over GDP using WEO data.
- G3 bond yield: GDP weighted average of bond yields for the U.S., Japan and Germany. Bond data comes from IFS.
- Scientific publications: Publications in scientific journals per capita, from WDI. (Does not include Taiwan Province of China. Data available from 1987 to 2000.)
- Secondary and tertiary education: Secondary and tertiary school enrollment rates, from WDI. (Data does not include Taiwan Province of China. Data covers the period 1991-2001, but is missing for some countries after 1997.)

*Source: _wp05189 - appendix presents the data in greater detail.*

### Appendix II. Tables

### Appendix II. Tables

### Impact of China on FDI Flows to Asia: Average Impact (Table 1)
- Lagged FDI:
  - OLS: 0.309*** (0.000)
  - A-Bond: 0.276*** (0.000)
  - IV: 0.296*** (0.000)
  - Corresponding FDI-to-China / total GDP of other economies: OLS 0.320*** (0.000), A-Bond 0.289*** (0.000), IV 0.299*** (0.000)
- China indicator (effect on other economies’ FDI):
  - China’s share of total FDI to the region: OLS -0.025** (0.013), A-Bond -0.024** (0.028), IV -0.042** (0.017)
  - FDI to China / total GDP of other economies: OLS -0.355 (0.130), A-Bond -0.284 (0.288), IV -1.156** (0.022)
- Other significant controls (selected):
  - Openness: OLS -0.015* (0.065), A-Bond -0.020** (0.021), IV -0.012 (0.139)
  - RER change: OLS 0.028** (0.010), A-Bond 0.029*** (0.008), IV 0.028** (0.010)
  - Inflation: OLS -0.059* (0.070), A-Bond -0.044 (0.207), IV -0.062* (0.061)
  - Government balance: OLS 0.163*** (0.005), A-Bond 0.133** (0.037), IV 0.174*** (0.003)
  - G3 bond yield: OLS -0.293*** (0.003), A-Bond -0.361*** (0.001), IV -0.364*** (0.002)
- Country fixed effects (selected, China’s share regressions):
  - Bangladesh 0.043*** (0.000)
  - India 0.047*** (0.001)
  - Indonesia 0.048*** (0.001)
  - Korea 0.051*** (0.003)
  - Malaysia 0.082*** (0.000)
  - Myanmar 0.091*** (0.000)
  - Singapore 0.136*** (0.000)
  - Thailand 0.056*** (0.000)
  - Vietnam 0.084*** (0.000)
- Observations: 212 (OLS), 198 (A-Bond), 212 (IV)

### Impact of China on FDI Flows to Asia: Impact by Economy (Table 2)
- Lagged FDI (coefficients across columns):
  - 0.310*** (0.000), IV 0.281*** (0.000), other columns 0.312*** (0.000), 0.247*** (0.002), 0.295*** (0.000), 0.249*** (0.002)
- Country-specific China impacts (selected coefficients and p-statistics):
  - Myanmar: China’s impact on China’s share regressions:
    - OLS -0.092*** (0.009), IV -0.132*** (0.005)
    - FDI to China / total GDP of other economies: OLS -1.581** (0.029), IV -2.319*** (0.004)
  - Singapore: China’s impact:
    - China’s share: OLS -0.098*** (0.001), IV -0.094*** (0.006)
    - FDI to China / total GDP of other economies: OLS -0.718 (0.186), IV -1.068* (0.095)
  - Indonesia: FDI to China / GDP of LHS economy: IV -0.086* (0.060)
  - Malaysia: FDI to China / total GDP of other economies: IV -1.285* (0.081)
- Robust controls (selected):
  - RER change positive and significant: OLS 0.030** (0.011), IV 0.030*** (0.008)
  - Government balance positive and significant: OLS 0.165*** (0.007), IV 0.174*** (0.004)
  - G3 bond yield negative and significant: OLS -0.369*** (0.001), IV -0.250*** (0.009)
- Country fixed effects (coefficients for China’s share regressions, OLS and IV shown):
  - Bangladesh 0.050*** (0.002) / 0.039** (0.029)
  - India 0.049** (0.010) / 0.047** (0.022)
  - Indonesia 0.054*** (0.002) / 0.054*** (0.004)
  - Korea 0.062*** (0.004) / 0.071*** (0.004)
  - Malaysia 0.092*** (0.000) / 0.111*** (0.000)
  - Myanmar 0.138*** (0.000) / 0.143*** (0.000)
  - Singapore 0.199*** (0.000) / 0.208*** (0.000)
  - Vietnam 0.085* (0.090) / 0.101*** (0.000)
- Observations: 212 (all columns)

### Impact on Myanmar, Singapore, and Other Economies (Table 3)
- Lagged FDI:
  - OLS 0.319*** (0.000), A-Bond 0.296*** (0.000), IV 0.304*** (0.000)
  - FDI to China / total GDP of other economies: OLS 0.326*** (0.000), A-Bond 0.295*** (0.000), IV 0.311*** (0.000)
- China’s impact (selected):
  - Singapore: China’s share regressions: OLS -0.094*** (0.001), A-Bond -0.098*** (0.000), IV -0.097*** (0.004)
    - FDI to China / total GDP of other economies: OLS -0.738 (0.151), A-Bond -0.765 (0.160), IV -1.261** (0.042)
  - Myanmar: China’s share regressions: OLS -0.100*** (0.003), A-Bond -0.088** (0.012), IV -0.130*** (0.005)
    - FDI to China / total GDP of other economies: OLS -1.682** (0.013), A-Bond -1.435* (0.051), IV -2.282*** (0.004)
  - “Other economies” coefficient: China’s share OLS -0.011 (0.266); FDI to China / total GDP OLS -0.246 (0.319)
- Controls with consistent significance:
  - Openness: negative and significant across estimators (e.g., OLS -0.022*** (0.009))
  - RER change: positive and significant (e.g., OLS 0.027** (0.016))
  - Government balance: positive and significant (e.g., OLS 0.161*** (0.004))
  - G3 bond yield: negative and significant (e.g., OLS -0.332*** (0.001))
- Country fixed effects (China’s share regressions, OLS):
  - Singapore 0.182*** (0.000)
  - Myanmar 0.136*** (0.000)
  - Thailand 0.056*** (0.000)
  - Vietnam 0.085*** (0.000)
- Observations: 212 (OLS), 198 (A-Bond), 212 (IV)

### Role of GDP per Capita in Diversion (Table 4)
- Lagged FDI:
  - OLS 0.309*** (0.000), A-Bond 0.272*** (0.000), IV 0.316*** (0.000)
  - FDI to China / total GDP: OLS 0.319*** (0.000), A-Bond 0.291*** (0.000), IV 0.311*** (0.000)
- China indicator and interaction with GDP per capita:
  - China indicator: China’s share OLS -0.018 (0.139), A-Bond -0.010 (0.462), IV -0.021 (0.172)
  - China indicator * GDP per capita: OLS -0.001 (0.302), A-Bond -0.002* (0.095), IV -0.001 (0.314)
- Selected controls:
  - Openness: OLS -0.018** (0.037), A-Bond -0.027*** (0.005)
  - RER change: OLS 0.030*** (0.007), A-Bond 0.033*** (0.003), IV 0.031*** (0.005)
  - Inflation: OLS -0.057* (0.079), IV -0.062* (0.060)
  - Government balance: OLS 0.162*** (0.005), IV 0.156*** (0.007)
  - G3 bond yield: OLS -0.304*** (0.003), A-Bond -0.389*** (0.001), IV -0.234** (0.012)
- Country fixed effects (China’s share regressions, OLS):
  - Singapore 0.141*** (0.000)
  - Malaysia 0.084*** (0.000)
  - Myanmar 0.090*** (0.000)
  - Vietnam 0.084*** (0.000)
- Observations: 212 (OLS), 198 (A-Bond), 212 (IV)

### Role of Scientific Development in Diversion (Table 5 and Table 6)
- Table 5 (including Singapore):
  - Lagged FDI: 0.210** (0.015) to 0.264*** (0.002) across columns
  - China indicator * scientific publications:
    - OLS -0.583** (0.020), A-Bond -0.599** (0.017), IV -0.349 (0.285)
    - For FDI to China / total GDP: coefficients very large and negative (e.g., -5.363 (0.231), -5.554 (0.240), -7.524 (0.251)) but with non-significant p-statistics reported
  - Scientific publications (main effect): OLS 0.450*** (0.008), A-Bond 0.467*** (0.006)
  - GDP per capita relative to China and other controls:
    - GDP per capita relative to China’s: IV -0.002*** (0.004) in China’s share regressions and negative significance in some columns
    - RER change: OLS 0.033*** (0.008), A-Bond 0.033*** (0.008)
    - Inflation: negative and significant (e.g., OLS -0.091** (0.033))
    - Government balance: OLS 0.209*** (0.004)
  - Country fixed effects (China’s share OLS): Singapore 0.121** (0.032), Malaysia 0.076** (0.023), Myanmar 0.091*** (0.001), Vietnam 0.081*** (0.001)
  - Observations: 158 (OLS), 145 (A-Bond), 158 (IV)
- Table 6 (excluding Singapore):
  - Lagged FDI: OLS 0.307*** (0.000), A-Bond 0.278*** (0.001), IV 0.291*** (0.001)
  - China indicator * scientific publications: coefficients not significant (e.g., 0.217 (0.828), -0.251 (0.802), 1.028 (0.322))
  - Significant controls when excluding Singapore:
    - Growth negative and significant: -0.090** (0.031) etc.
    - Inflation strongly negative and significant: e.g., -0.112*** (0.003)
    - Government balance positive and significant: e.g., 0.206*** (0.002)
  - Observations: 143 (OLS), 131 (A-Bond), 143 (IV)

### Role of Education in Diversion (Table 7)
- Lagged FDI:
  - China’s share regressions: OLS 0.303*** (0.010), A-Bond 0.140 (0.324), IV 0.311*** (0.009)
  - FDI to China / total GDP: OLS 0.317*** (0.008), IV 0.306** (0.012)
- China indicator interactions with education:
  - China indicator * secondary education: coefficients small and not significant (e.g., 0.000 (0.836))
  - China indicator * tertiary education: coefficients small and not significant (e.g., -0.000 (0.794))
- Education variables (main effects and notable controls):
  - Secondary education: IV 0.001* (0.071) for China’s share; IV 0.001** (0.047) for FDI to China / total GDP (coefficients reported as 0.001)
  - Government stability: negative and significant in some specifications (e.g., OLS -0.003** (0.031))
  - Business environment: OLS 0.005** (0.048) in China’s share regressions; other specifications show significance at the 10 percent level
  - RER change: strongly positive and significant across specifications (e.g., 0.053*** (0.005); IV 0.061*** (0.001))
  - Government balance: positive and significant (e.g., OLS 0.272** (0.020), IV 0.265** (0.025))
- Country-specific China impact coefficients (selected, many non-significant):
  - Korea: negative in some IV/A-Bond columns with weak to moderate significance (e.g., -0.125* (p-statistic reported in a column))
  - Observations vary by specification: 107 (OLS columns), 76 (A-Bond columns), 107 (IV columns)

*Appendix II. Tables — source PDF: _wp05189 - Appendix II. Tables*

### References

### _wp05189 - References

### China, East Asia, and FDI studies
- Ahearne, Alan, Fernald, John, Loungani, Prakash, and John Schindler, 2003, “China and Emerging Asia: Comrades or Competitors?”, Board of Governors of the Federal Reserve, International Finance Discussion Paper No. 789 (Washington: Board of Governors of the Federal Reserve).
- Borensztein, Eduardo, De Gregorio, Jose, and Jong-Wha Lee, 1998, “How Does Foreign Direct Investment Affect Economic Growth?”, Journal of International Economics, Vol. 45, pp. 115-35.
- Chantasasawat, Busakorn, K.C. Fung, Hitomi Iizaka, and Alan Siu, 2004, “Foreign Direct Investment in China and East Asia,” Stanford Center for International Development Working Paper No. 233 (Stanford: Stanford University).
- Eichengreen, Barry, Rhee, Yeongseop, and Hui Tong, 2004, “The Impact of China on the Exports of Other Asian Countries,” (mimeo; Berkeley: University of California at Berkeley).
- McKibbin, Warwick, and Wing Thye Woo, 2003, "The Consequences of China's WTO Accession on its Neighbours," Asian Economic Papers, Vol. 2 (2), pp. 1-38.
- Sun, Qian, Tong, Wilson, and Qiao Yu, 2002, “Determinants of Foreign Direct Investment across China,” Journal of International Money and Finance, Vol. 21, pp. 79-113.
- Tseng, Wanda, and Harm Zebregs, 2003, “Foreign Direct Investment in China: Some Lessons for Other Countries,” in China: Competing in the Global Economy, ed. by Wanda Tseng (Washington: International Monetary Fund).
- Wei, Shang-Jin, 1999, “Can China and India Double Their Inward Foreign Direct Investment?,” (unpublished; Cambridge: Harvard University).

### Capital flows, external wealth, and private capital
- Bosworth, Barry, and Susan Collins, 1999, “Capital Flows to Developing Economies: Implications for Saving and Investment,” Brookings Papers on Economic Activity, Vol. 1, pp. 143-69.
- Fernandez-Arias, Eduardo, 1996, “The New Wave of Private Capital Inflows: Push or Pull?,” Journal of Development Economics, Vol. 48, pp. 389-418.
- Gastanaga, Victor, Jeffrey Nugent, and Bistra Pashamova, 1998, “Host Country Reform and FDI inflows: How Much Difference do they Make?,” World Development, Vol. 26(7), pp. 1299-1314.
- Hausmann, Ricardo, and Eduardo Fernandez-Arias, 2000, “Foreign Direct Investment: Good Cholesterol?” Inter-American Development Bank Working Paper No. 417 (Washington: Inter-American Development Bank).
- Kamaly, Ahmed, 2003, “Behind the Surge of FDI to Developing Countries in the 1990s. An Empirical Investigation,” (Cairo: The American University in Cairo, Department of Economics), unpublished.
- Lane, Philip, and Gian Maria Milesi-Ferretti, 2001, “The External Wealth of Nations: Measures of Foreign Assets and Liabilities for Industrial and Developing Countries,” Journal of International Economics, Vol. 55, pp. 263-94.
- Lensik, Robert, and Howard White, 1998, “Does the Revival of International Private Capital Flows Mean the End of Aid? An Analysis of Developing Countries’ Access to Private Capital,” World Development, Vol. 26(7), pp. 1221-34.
- Wei, Shang-Jin, 2001, “Negative Alchemy? Corruption, Composition of Capital Flows, and Currency Crisis,” NBER Working Paper 8187 (Cambridge, Massachusetts: National Bureau of Economic Research).

### Human capital, corruption, and determinants of FDI
- Borensztein, Eduardo, De Gregorio, Jose, and Jong-Wha Lee, 1998, “How Does Foreign Direct Investment Affect Economic Growth?”, Journal of International Economics, Vol. 45, pp. 115-35.
- Noorbaksh, Farhad, and Alberto Paloni, 2001, “Human Capital and FDI Inflows to Developing Countries: New Empirical Evidence,” World Development, Vol. 29, pp. 1593-610.
- Gastanaga, Victor, Jeffrey Nugent, and Bistra Pashamova, 1998, “Host Country Reform and FDI inflows: How Much Difference do they Make?,” World Development, Vol. 26(7), pp. 1299-1314.
- Hausmann, Ricardo, and Eduardo Fernandez-Arias, 2000, “Foreign Direct Investment: Good Cholesterol?” Inter-American Development Bank Working Paper No. 417 (Washington: Inter-American Development Bank).
- Sun, Qian, Tong, Wilson, and Qiao Yu, 2002, “Determinants of Foreign Direct Investment across China,” Journal of International Money and Finance, Vol. 21, pp. 79-113.

### Methodology and panel data
- Arellano, Manuel, and Stephen Bond, 1991, “Some Test of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations,” Review of Economic Studies, Vol. 58, pp. 277-97.

### Miscellaneous speeches and commentary
- Lamy, Pascal, 2003, Speech given at the International Chamber of Commerce in Bangkok on 1 April 2003.
- Mohamad, Mahatir, 2003, Euromoney, 1 October 2003.
- Panitchpakdi, Supachai, 2000, Financial Times, 23 November 2000.
- Wei, Shang-Jin, 1999, “Can China and India Double Their Inward Foreign Direct Investment?,” (unpublished; Cambridge: Harvard University).

* _wp05189 - References_*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05189.pdf_