The Journal of Developing Areas, Volume 48, Number 1, Winter 2014, pp (Article) DOI: /jda

Transcription

1 Th D l P l t l ff t f F r n D r t nv t nt n D v l p n ntr Feng Sun The Journal of Developing Areas, Volume 48, Number 1, Winter 2014, pp (Article) P bl h d b T nn t t n v r t ll f B n DOI: /jda For additional information about this article Access provided by Penn State Univ Libraries (21 Feb :09 GMT)

2 T h e J o u r n a l o f D e v e l o p i n g A r e a s Volume 48 No. 1 Winter 2014 THE DUAL POLITICAL EFFECTS OF FOREIGN DIRECT INVESTMENT IN DEVELOPING COUNTRIES Feng Sun Troy University, USA ABSTRACT This study revisits the political effect of Foreign Direct Investment (FDI) on the level of democracy in developing countries. The author finds that FDI has dual political effects based on the panel corrected standard error (PCSE) analysis using panel data covering 124 developing countries from 1970 to Although the political effect of aggregate FDI is negative, FDI from developed democracies exerts a significant positive effect on democracy. FDI in the primary sector plays a negative role in the political development in developing countries. The author also finds a regional difference in the political effect of FDI due to uneven distribution of disaggregated FDI. JEL Classifications: A13; F21; F59 Keywords: Dual Political Effects, Disaggregated FDI, Developing Countries Corresponding Author s Address: fsun57924@troy.edu INTRODUCTION In the last 30 years, the world has witnessed a simultaneous trend of democratization along with globalization in the developing world. By 1996, North-South FDI flows have swelled from 15 percent to over 40 percent of the world total. Meanwhile, since the late 1980s, 70 percent of developing countries have made substantial efforts to promote political freedoms (Rudra, 2005; Freedom House, 2010). An apparent link between FDI and the level of democracy has warranted growing attention of both the scholars and policy makers. However, conclusions of existing theoretical and empirical studies are mixed. Some suggest a positive effect of FDI on the level of democracy (e.g., Firebaugh, 1992; de Soysa, 2003); while others posit a negative or an insignificant political effect (e.g., Rudra, 2005; Li & Reuveny, 2003; Bornschier & Chase-Dunn, 1978; Gallagher, 2002). The purpose of this study is to shed some light on this puzzle by exploring the disaggregated and regional variations of the political effects of FDI. Scholars argue that FDI from developing economies and/or in the primary sector is different from the counterpart from the developed democracies and/or in the non-primary sector in terms of type of ownership, economic effect, level of influence on host society, requirement for liberal economic settings, and social spillovers (Rothgeb, 1984; Russett & Oneal, 2001; Alfaro, 2003; Schulz, 2009; Mihalache-O keef1& Li, 2011; United Nations Conference on Trade and Development (UNCTAD), 2006). In addition, the geographical distribution of FDI to developing countries remained highly uneven. FDI to Eastern and Central Europe and Latin America with relatively high level of democracy were mostly from

3 108 developed economies and manufacture-based. For the Asia, Africa, and the Middle East regions with low level of democracy, many countries received FDI from neighboring developing economies and had high percentage of FDI in the primary-sector (UNCTAD, 2006). Therefore, I propose that the effects of FDI on politics in developing countries may depend on the targeted industrial sector, regime type of the source country, and the region. Perhaps existing analyses have reached conflicting results because they did not take these factors into account. This study analyzes the political effects of FDI by sector, regime type of the source country, and region, using panel data for 124 developing countries from 1970 to 2005 with the Panel Corrected Standard Error (PCSE) analysis. The empirical results support the hypothetical theories, showing that the search for a universal effect of aggregate FDI on democratic development is misguided. The findings are threefold: first, although the overall political effect of FDI is significantly negative in the developing world, FDI from advanced democracies presents a consistent positive effect on democracy. Second, FDI in the primary sector exerts a negative political impact. And third, the political effects of FDI vary across regions. This article proceeds as follows: the next section provides a review of theoretical literature and existing empirical analyses. Section 3 discusses the political effects of disaggregated FDI by regime type of the source country and by sector. The data and model specification are presented in section 4. Section 5 reports the empirical findings and the final section concludes. EFFECTS OF FDI ON DEMOCRACY Available literature and empirical studies on the nexus between overall FDI and the level of democracy presents conflicting views and results (e.g., Li & Reuveny, 2003; Gallagher, 2002; Rudra, 2005). Opponents argue that Multinational Companies (MNC) are attracted to countries with authoritarian regimes. They accuse FDI of crowding out national enterprises, destroying economies, degrading the environment, and deteriorating accountability, human rights, and civil society by fostering authoritarianism to ensure a stable, wellcontrolled, and relatively low-risk investment environment in host developing countries (Bornschier & Chase-Dunn, 1978; O Donnell, 1988; Gallagher, 2002; He, 2012). Proponents contend that in order to operate efficiently and effectively, MNCs emphasize basic requirements such as a stable political environment (Jensen, 2006; Li, 2006), an enforceable system of property rights, impartial courts, and predictable rules, all of which are common features of democracy (Biglaiser & Staats, 2010; Harms & Ursprung, 2002; Li & Resnick, 2003). Second, by promoting economic growth, FDI boosts the middle class, reduces income inequality, and improves education, eventually fostering democratization (Lipset, 1959; Bussmann, de Soysa, & Oneal, 2005; Firebaugh, 1992; Dutta & Osei-Yeboah, 2010). Third, MNCs and their affiliates act as an effective proxy for democracies to spread liberal values and ideas like transparency, accountability, labor standards, and commercial probity, etc, to developing countries. Russett and Oneal (2001, p.198) suggest that authoritarian countries with free capital inflows from democracies may be encouraged to democratize through this type of communication.

4 109 Li and Reuveny (2003), de Soysa (2003), and Rudra (2005) represent the few statistical studies on the political effects of overall FDI. Based on the sophisticated empirical analysis of 127 countries from 1970 to 1996, Li and Reuveny (2003) show that FDI inflows significantly and positively affect the level of democracy, though the effect weakening over time. Rudra analyzes 59 developing countries from 1972 to 1997 to explore the link between trade, capital flows and democratization. Nevertheless, FDI is ultimately dropped out of the model due to the high correlation with trade and an insignificant test result. de Soysa (2003) examines roughly 100 countries from 1970 to He finds that FDI stocks as a percentage of GDP has a small but positive long-run effect on democracy. The reverse effect of democracy on FDI stocks is significant, but with a negative coefficient. POLITICAL EFFECTS OF DISAGGREGATED FDI FDI from Developed Democracies and Developing Countries Although FDI from developing economies constitutes about 20 percent of the world total, the uneven distribution and the intraregional nature of this type of FDI make it very influential in some developing economies. For example, in some developing countries such as China, Thailand, Tanzania, and Paraguay, over 40 percent of the total incoming FDI comes from other developing economies (UNCTAD, 2006). The previously discussed literature doesn t distinguish the political effects of FDI based on regime type of the source country. Nevertheless, some positive effects only belong to FDI from developed democracies. For example, FDI from developed democracies not only has a crucial effect in financial liberalization but also plays an important role in diffusing democratic ideas and values in developing countries (de Soysa, 2003; Li & Reuveny, 2003; Rudra, 2005). It is unlikely that MNCs from developing autocracies would disseminate democratic values and promote political liberalization in host countries. Also, compared to their developed counterparts, FDI from developing countries possess two distinct features. First, a relatively high degree of state ownership can be observed. It is believed that state ownership raises the risk of a FDI transaction being undertaken for other than purely economic motives, especially for acquisitions related to energy, infrastructure services, extractive resources, or other industries with a security dimension (UNCTAD, 2006). A significant bulk of FDI from developing countries has been reported in the primary sector. In some specific industries, such as petroleum refining, MNCs from developing countries have a particularly strong presence, for example Petrobras from Brazil, CNPC and CNOOC from China, and Gazprom and Lukoil from Russia (UNCTAD, 2006). Second, MNCs from developing countries generally apply a different strategy of building affiliates abroad than their developed counterparts. MNCs from developed countries are most likely to utilize firm-specific advantages based on ownership of assets, such as technologies, brands, and other intellectual property, while developing country MNCs rely more on other firm-specific advantages, derived from production process capabilities, networks and relationships, and organizational structure (UNCTAD, 2006, p.24). Apparently, if organizational structure, networks, and relationships are the basis of foreign facilitates invested by developing

5 110 countries, corporate governance and transparency are questionable, whether the investors are considered private or state-owned. FDI in the Primary and Non-Primary Sector Scholars find that FDI in the primary sector such as mining tends to have a negative effect on growth, while investment in the manufacturing sector has a positive one (Alfaro, 2003; Schulz, 2009; Rothgeb, 1984; UNCTAD, 2001). The political effect of FDI in the primary sector, which is directly associated with its economic effect, also seems negative. Most developing countries abundant in natural resources frequently lack the capital or expertise to extract these resources and have to partner with foreign investors to manage them (Schulz, 2009). Resource-seeking FDI is often associated with authoritarian regimes. A growing literature on the political resource curse argues that authoritarian regimes take advantage of the large revenues from taxes, licenses, and profit-sharing arrangements generated by foreign companies through the exploitation of natural resources to buttress their elite political positions (Jensen & Johnson, 2011; Schulz, 2009; Ross, 2001; Smith, 2004). FDI in the primary sector has little impact on society in developing countries. Most resource-seeking MNCs operate in geographically remote regions, take few inputs from local suppliers, and do not sell their products in domestic markets (Alfaro, 2003; Schulz, 2009). A World Bank report (UNCTAD, 2001, p.138) points out that the scope for linkages between foreign affiliates and local suppliers is often limited in the primary sector, while the non-primary sectors have a broad variation of linkage intensive activities. MNCs in manufacturing and service sectors are deeply integrated into the host country economy through a network of relationships with local suppliers, customers, and labor (Schulz, 2009; Mihalache-O keef1& Li, 2011). By contrast, MNCs focusing on natural resources are isolated from the rest of the host country economy. As a result, this kind of FDI has few interactions with the society of the host developing countries. Based on the above discussion, it is problematic to examine aggregate FDI while ignoring the possible different effects of different types of FDI, because the results could be deceptive. For example, China is always considered an exception in terms of the political effects of FDI. In past decades, China was generally viewed as a major destination for foreign capital. However, the growing influx of FDI has not encouraged corresponding political changes. Nevertheless, if the composition of FDI inflows to China is considered, only about 20 percent of FDI inflows come from developed democratic countries (UNCTAD, 2006). DATA AND MODEL SPECIFICATION The statistical model covers the period from 1970 to 2005, for a sample of 124 developing countries and emerging market economies. The sample selection is based on data availability. FDI data have only been available since 1970 for most developing countries. In addition, the sample excludes countries with either a population of less than 1 million or a ratio of FDI to GDP greater than 200 percent because very small countries often have idiosyncratic investment profiles and countries with a high ratio of FDI to GDP usually serve as conduits for FDI transshipment (Schulz, 2009, p.14). The sample

6 111 countries could be classified into three types: those that are considered democratic but the quality of democracy remains tenuous (e.g., India, Venezuela, and Peru), countries that maintain strong authoritarian regimes (e.g., Egypt, China, and Singapore), and postcommunist countries that have experienced or are still in the process of economic and political transition (e.g., Poland, Czech Republic, and Russia). The unit of analysis is country year. The independent variable FDI is measured by the annual value of net inflows of FDI or FDI stocks as a percentage of a country s GDP, respectively. The aggregate FDI data are obtained from the UNCTAD website. The World Development Indicator (WDI) dataset provides the GDP data from 1970 to The values of FDI from democratic countries or in the primary sector are calculated by multiplying total FDI with the percentage of FDI from developed democracies or in the primary sector. The percentage data are collected from different sources. Detailed information on FDI by sector and by regime type of the source country is available in the ten-volume series of the World Investment Directory (WID), published between 1993 and 2008 by UNCTAD. The Organization for Economic Cooperation and Development (OECD) also provides FDI data by sector and by regime type of the source country for OECD countries, which complements the UNCTAD disaggregated FDI data. The FDI dataset has missing values, though it is the best option for studying the effect of disaggregated FDI over a reasonably long period for a large and regionally diverse sample of developing economies. The dependent variable the level of democracy is measured by Freedom House s political rights and civil liberty indices. Some previous studies adopt the Polity IV data as the measurement of democracy (Li & Reuveny, 2003; Rudra, 2005). However, Polity IV dataset concentrates more on political institutional criteria such as regulation, competitiveness, openness of executive recruitment, and executive constraints. In addition to the institutional aspects, the Freedom House indices include political rights and civil liberties indicators, such as the right to compete for public office, the freedom to develop views, personal autonomy apart from state, and enforcement of the rule of law. In this sense, the Freedom House data could capture the variations of democracy more accurately than the Polity data. For example, in some developing democracies, civil rights are not respected. In pseudo-democracies, such as Nigeria, Indonesia, and Ukraine, rights of contestation are protected and institutional settings are maintained, but civil rights and liberties are precarious (Diamond, 2002). In such cases, the Polity data may overestimate the democratic quality. However, there are some cases in which the Polity data may also underestimate the level of democracy. For example, the Polity score of China has remained at the same score, -7, since 1976, although communist ideology in that country has been declining dramatically over the past three decades. Civil liberties and human rights are beginning to be recognized and respected in a modest but steadily increasing way. The original political rights and civil liberties indices are each ranked on a scale from 1 to 7, with 1 representing the highest level of freedom and 7 the lowest. A revised and combined new index is adopted by adding these two indices together and reversing the combined scale to improve readability. The new Freedom Index rates from 2 to 14, 2 representing the lowest level of democracy and 14 the highest. Most developing countries fall into this range with values between 2 to 10.

7 112 The Polity IV data will also be included in the statistical analysis to make a comparison with the Freedom House data. A new variable Polity Score is calculated as the difference between democratic and autocratic indexes, generating a new index ranging between -10 (an extremely autocracy) and +10 (a fully democracy). Some important determinants of the level of democracy are adopted here as control variables. The domestic economic control variables include GDP per capita, real economic growth rates, the inflation rates, and urbanization rates. GDP per capita denotes the logged yearly GDP per capita in constant dollars. Growth rate refers to the real economic growth of a country. Inflation is measured as the yearly percentage of change in the GDP deflator. This variable, as some studies show, is often used as a proxy for a macroeconomic crisis (Gasiorowski, 1995). Urbanization is measured as the percentage of urban population in the total population. According to Acemoglu and Robinson (2001), urbanization makes it more difficult for the elite group to maintain strong societal controls. The data for these variables are taken from WDI Database (2010). Although some variables such as colonial legacy, institutional structure, and democratic tradition may also affect the level of democracy, they are relatively constant in the period analyzed, implying that democracy exhibits inertia (Li & Reuveny, 2003). To control for this effect of inertia, the lagged variable of democracy is introduced as an independent variable. It is possible that there is a contagion effect from democratic neighbors, especially in regions such as Central and Eastern Europe. A regional democracy variable measured as the percentage of democratic regimes in the region to which the country belongs was introduced to take any regional democratic effects into account. Although most empirical studies on the effect of globalization measure its degree by both capital and trade flows (Bussmann et al., 2005; Li & Reuveny, 2003; Rudra, 2005), these two forces overlap. Rudra (2005) dropped FDI out of the model due to the high correlation between the two variables. de Soysa finds that from 1970 to 1999, the correlation between trade and FDI is In our sample, for the period 1970 to 2005, the correlation between these two variables declines to This controversial variable was also included in this study. Table 1 shows summary statistics for the main variables. TABLE 1. SUMMARY STATISTICS OF THE MAIN VARIABLES Variable Observations Mean Standard Min Max Deviation Freedom Index Polity Score Inflow/GDP Stock/GDP Logged GDP/capita Growth rate Inflation Urbanization Trade The Time Series and Cross Sectional (TSCS) statistical models are employed to test the association between FDI and the level of democracy across countries and over time. Equation (1) is designed to test the hypotheses on the political effect of FDI. One

8 113 important issue of model specification for the TSCS model is endogeneity. The relationship between FDI and the level of democracy may be reciprocal. Therefore, following many previous studies (e.g., Li & Reuveny, 2003; Rudra, 2005; Oneal & Russett, 1999), each of the independent variables is lagged to take into account the possibility of simultaneity bias. Though this is a partial solution to deal with the endogeneity problem, lagging the independent variables sometimes ensures the direction of causality running from the independent variables to the dependent variable. Democracy it = a + b 1 FDI it-1 + b 2 Trade it-1 + b 3 Democracy it-1 + b 4 Inflation it-1 + b 5 GDP per capita it-1 + b 6 GDP growth rate it-1 + b 7 Urbanization it-1 + b 8 Regional democratic effect it-1 + ε. (1) Panel heteroskedasticity and spatial contemporaneous autocorrelation are recognized potential problems in TSCS statistical models (Beck & Katz, 1995). Many panel studies involve cross-sectional units (states) of varying size and situation, implying that error terms of the model may not have the same variances. The large number of cases increases the probability of heteroskedasticity. Plus, in the panel data, the shock in one state could also affect the neighboring countries at the same time points or different time points. Instead of independence, the error terms in a TSCS model are serially correlated. These two features violate basic assumptions of Ordinary Least Squares (OLS) analysis, resulting in misleading and inefficient OLS estimates of the standard errors of the TSCS model (Cheng, 2003; Verbeek, 2004). Table 2 presents the results of Wooldridge test for within panel autocorrelation and test of heteroskedasticity. Both show that there is a problem with autocorrelation and heteroskedasticity. Thus, a Panel Corrected Standard Error procedure is adopted here to address these two problems. According to a general rule of thumb of statistics, a serious multicollinearity problem occurs if the Variance Inflation Factor (VIF) is greater than 10 (Belsley, Kuh, & Welsch, 1980). The test (Table 2) detects no serious multicollinearity problem in this study. While applying PCSE on an unbalanced dataset, one has to specify how missing observations are treated when computing the covariance matrix of the disturbances: either casewise, using only observations common to all panels and completely excluding timeperiods with at least one missing observation, or pairwise, using observations common to the two panels used to calculate the covariance. According to Amable, Gatti, and Schumacher (2006), for highly unbalanced panels with many missing observations, the latter (pairwise) may be advisable. This study follows their method. RESULTS Table 3 shows the results of the PCSE analysis for the models using the Freedom Index as the dependent variable. Overall FDI exhibits a weak but significant negative effect on the level of democracy in developing countries. However, once regime type of the source countries is taken into account, the direction of the effects changes. Inflows of FDI from developed democratic countries exert a positive and significant effect on the level of democracy in developing countries. The standardized regression coefficients for the baseline equation show that a standard deviation increase of the ratio of FDI inflows from advanced democratic countries to GDP raises the democratic score by 0.2 points. Given

9 114 that the mean value for democracy and the ratio of FDI inflows from developed democratic countries to GDP among developing countries are only 7.08 and 1.44 in the time-series data, respectively, a 0.2 point increase is relatively significant. TABLE 2. TESTS ON AUTOCORRELATION, HETEROSKEDASTICITY, AND MULTICOLLINEARITY Wooldridge Test for Autocorrelation in Panel Data H0: no first-order autocorrelation F (1, 120) = Prob>F = Likelihood-Ratio Test for Heteroskedasticity LR Chi 2 (122) = (Assumption: nested in hetero) Prob> Chi 2 = Variance Inflation Factors Test Variable VIF FDI 1.27 Logged GDP per capita 2.96 Trade 1.29 GDP growth rate 1.12 Inflation rate 1.04 Urbanization 2.81 Regional democratic influence 1.59 Prior democracy 1.45 The political effect of FDI stocks from developed democratic economies is expected to be positive and significant; however, the sign of the coefficient is positive but not significant. This is possibly because of the uneven distribution of the FDI stocks data from advanced democratic countries. FDI inflows data from advanced democratic countries are available for almost all of the countries, but the data for FDI stocks are mostly missing for the countries in Latin America and Central and Eastern Europe. These two regions are expected to yield strong effects of FDI on the level of democracy due to the large amount of FDI inflows and the substantial variations of democracy over time. With regard to FDI in different sectors, although FDI inflows in the primary sector have positive effects, FDI stocks in the primary sector exert negative influence on democratic development, as expected. A standard deviation increase of the ratio of FDI stocks in the primary sector to GDP would lead to a decrease of the democratic score by 0.14 points. FDI stocks in the primary sector would exhibit a stronger negative effect if the data were not mostly missing for African countries, popularly perceived as states with low democratic scores and large FDI in the industries focusing on natural resources exploitation (UNCTAD, 1996). In addition, data in the primary sector from the series of WIDs not only include extractive FDI in mining, quarrying, and the petroleum industry, which, according to the theory, exerts a negative impact on democratic development in host countries; but also refers to FDI in agriculture, forestry, and fishing, which may have different effects that offset the negative effects of FDI (Alfaro, 2003; Rothgeb, 1984).

10 115 TABLE 3. EFFECTS OF FDI ON DEMOCRACY (FREEDOM INDEX) Variables FDI inflows/gdp t ** (0.01) FDI Stocks/GDP t *** (0.002) Inflows from 0.07** Democracies t-1 (0.03) Stocks from Democracies t-1 (0.007) Inflows in Primary t ** (0.015) Stocks in Primary t *** (0.008) Trade t * *** (0.001) (0.001) (0.002) (0.0023) (0.002) (0.004) Prior 0.384*** 0.4*** 0.42*** 0.42*** 0.4*** 0.4*** Democracy (0.007) (0.01) (0.01) (0.01) (0.111) (0.02) Inflation t *** *** ** ** ( ) ( ) (0.0002) (0.0002) (0.0002) (0.0002) Logged GDP 0.76*** 0.725*** 0.68*** 0.5*** 0.75*** 0.8*** Per capita t-1 (0.04) (0.045) (0.08) (0.02) (0.08) (0.13) GDP growth ** Rate t-1 (0.005) (0.006) (0.013) (0.019) (0.009) (0.015) Urbanization t *** *** * *** (0.002) (0.002) (0.004) (0.006) (0.003) (0.005) Democratic *** -0.13*** ** -0.02*** Regional Effect t-1 (0.002) (0.002) (0.0033) (0.003) (0.002) (0.003) Constant 2.86*** 2.96*** 3.41*** 4.13*** 2.27*** 2.84*** (0.2) (0.23) (0.56) (0.81) (0.55) (0.77) Observations R square Notes: Panel corrected standard errors in parentheses below the regression coefficients. Two-tailed tests: * sig. at.05; ** sig. at.01; ***sig. at.001.

11 116 TABLE 4. EFFECTS OF FDI ON DEMOCRACY (POLITY SCORE) Variables FDI inflows/gdp t (0.016) FDI Stocks/GDP t *** (0.003) Inflows from Democracies t-1 (0.07) Stocks from * Democracies t-1 (0.011) Inflows in Primary t *** (0.05) Stocks in Primary t *** (0.02) Trade t * *** (0.002) (0.002) (0.003) (0.005) (0.003) (0.004) Prior 1.64*** 1.62*** 1.63*** 1.6*** 1.62*** 1.57*** Democracy (0.03) (0.03) (0.05) (0.052) (0.053) (0.07) Inflation t *** * ( ) ( ) (0.0003) (0.0004) (0.0007) (0.003) Logged GDP -0.83*** -0.7*** -0.71*** *** Per capita t-1 (0.1) (0.09) (0.17) (0.15) (0.15) (0.19) GDP growth *** Rate t-1 (0.01) (0.01) (0.022) (0.03) (0.03) (0.03) Urbanization t *** 0.015*** 0.016* (0.005) (0.005) (0.007) (0.011) (0.007) (0.01) Democratic 0.045*** 0.04*** 0.033*** 0.04*** 0.03*** 0.055*** Regional Effect t-1 (0.003) (0.003) (0.004) (0.005) (0.005) (0.007) Constant -8.63*** -8.67*** -8.2*** -10.4*** -8.4*** -9.68*** (0.52) (0.52) (0.83) (0.65) (0.84) (1.19) Observations R square Notes: Panel corrected standard errors in parentheses below the regression coefficients. Two-tailed tests: * sig. at.05; ** sig. at.01; ***sig. at.001.

12 117 The availability of FDI disaggregated data, and its effect on the result, is a serious concern. One way to tell if missing data has biased the estimation is to examine the R square of the analysis. Table 3 shows that the R square of the models with fewer observations do not present a substantial difference compared to the overall models. According to Li and Reuvney (2003), this indicates that the results in the models with fewer observations are not artifacts due to large missing values. Regarding the control variable, trade has a significant effect on democracy in the models based on the overall FDI stocks data and the data of FDI inflows in the primary sector. In the other models, most of the signs of coefficients are positive. This finding is consistent with the hypothesis that international trade promotes democracy (Russett & Oneal, 2001). Prior democracy has a positive and significant effect on current democracy. It illustrates that democracy as a social phenomenon tends to have inertia. If a country strengthens dictatorship and authoritarianism in the present, one simply cannot expect to find a higher level of democracy in the near future. In the same vein, the regime transition toward democracy is a process, accumulating over time, which in turn leads to more democratic reforms (Li & Reuveny, 2003). This also implies that the democratization of developing countries is a long and tough evolution, because most of the developing countries have a legacy of authoritarianism. Although inflation is presumed to have a positive impact on democracy (Li & Reuveny, 2003; Rudra, 2005), the result was not as expected. In all the models, the signs of the coefficients are negative. In some models, the effects are also significant, which could be explained by Li and Reuveny s (2003) finding: the effect of inflation weakens over time, with the largest significant effect occurring in the 1980s. Economic growth, which is measured as the GDP growth rate, does not have any significant influence on democratic development. As in many studies, economic development is positive and statistically significant, being robust in all the models. This finding confirms the well-known Lipset (1959) hypothesis regarding the importance of economic development for democratization. The proposition that economic development promotes democracy not only applies well in the developed countries, but also fits well in the developing world. The result also shows that the influence of urbanization on the level of democracy is significantly negative. Although urbanization could be used as a proxy for industrialization level, for most developing countries, this variable only retains its literal meaning. Urbanization is faster than industrialization in most of the developing countries. If industries do not grow fast enough to absorb migration from rural areas, it will increase the burden of government and may force the government to adopt more repressive policies to control the dramatically increasing urban population, most of which is lowincome (Rudra, 2005). The negative coefficient for the democratic regional effect variable is surprising, because this variable tends to be positive and highly significant (Rudra, 2005). But as some scholars (e.g., Diamond, 2002; Rudra, 2005; Zakaria, 1997) have pointed out, democracies, especially those in the developing world, do not necessarily bring about improvements in basic rights and liberties (Rudra, 2005, p.720). Particularly for developing democracies swept up by the third wave while still experiencing economic liberalization, repression became a means to maintain control and provide stability

13 118 (Rudra, 2005). In this sense, although those countries are labeled as democracies based on their institutional settings, they actually failed to consolidate meaningful aspects of democracy: e.g., promoting the rule of law and civil liberties and respecting political rights. This result further supports the aforementioned argument that Polity IV data have limitations for capturing some variations of democracy in developing countries. Table 4 presents the results of PCSE analysis for the models using the Polity Score as the dependent variable. As expected, the results are very different from the findings of the models using the Freedom Index in that respect. FDI stock has a significant and weak influence on democracy while the political effect of FDI inflow is not significant. The table shows that FDI inflow from democracies has no significant political effect. However, FDI stock from democracies exerts a significant negative effect on the level of democracy. FDI inflow per se in the primary sector plays a negative role in democratization, while FDI stock in the primary sector exhibits a positive political effect. These models also present several interesting results supporting the arguments regarding the differences between the Freedom Index and the Polity Score. The former focuses more on the spiritual or ideological aspects of democracy such as political rights and civil society, while the latter concentrates on the institutional aspect of democracy. First, the political effect of the economic development variable, supposed to be positive according to conventional wisdom, is negative and significant. The logic of economic development benefiting democracy lies in the fact that a booming economy creates a large highly educated middle class, itself critical in building a vibrant civil society. Apparently, economic development directly promotes the spiritual or ideological aspects of democracy. Second, unlike the models with the Freedom Index, the models with the Polity Score show that the democratic regional effects are all positive and significant. This finding implies that in some developing countries, the democratic institutional structures are more susceptible to the international or regional influences. But the spiritual or ideological aspects of democracy are based more on internal conditions and intrinsic impetus. In addition, the coefficients of prior democracy in the models with the Polity Score are almost four times larger than those in the models with the Freedom Index. In order not to soak up the variations in the dependent variable that could be explained by other independent variables (Li & Reuveny, 2003), the variable prior democracy in the models with the Freedom Index is measured by the lagged Polity Score while the same variable in the models with the Polity Score is measured by the lagged Freedom Index. The finding shows that in the developing world, the spiritual or ideological aspects of democracy have more important and stronger effects on future democratization than does the institutional aspect of democracy. This study is not about the difference between the two indices. More important, the results of the comparison further justify the decision to use the Freedom House data in this study. The Polity Score may be effective in measuring the democratic level of developed countries because these mature democracies have balanced levels of democratic institutional settings and civil liberty and political rights development. But for developing countries, the Polity Score may be not flexible enough to capture important non-institutional changes. Another important finding of this study is the regional variation of the political effects of FDI (see Table 4 and 5). FDI in Central and Eastern Europe exerts a significant positive effect on democracy, in line with the widely accepted demonstration that FDI is

14 119 an engine in the transition from state socialism in this region (IMF, 1997; UNCTAD, 1998; Schimidt, 1995). For the countries in this region, the average value of the Freedom index is 7.68, which is 0.6 point higher than the average value of all developing countries in the sample. FDI flows to this region are mostly from developed democracies and in the non-primary sectors. Among the 20 countries with disaggregated FDI data available, 16 of them have over 60 percent of FDI from developed democracies and most possess a very low percentage of FDI in the primary sector. For example, Bulgaria, Czech Republic, Poland, Hungary, Georgia, Estonia, Latvia, Slovakia, and Slovenia all have only 1-2 percent of FDI in the extractive industries (UNCTAD, 2006). Nevertheless, when this regional model is compared with the overall model, the R square of the regional model of Central and Eastern Europe increases to 0.86 (the R square of the overall model is only 0.77). According to Li and Reuveny (2003), this indicates that the selection of this specific region itself also contributes to the variation of democracy. Most of the Central and East European countries are high-income economies that experienced a great democratic revolution in the late 1980s and early 1990s. Moreover, half of the countries in this region are either members or candidates of the European Union, which requires each member country to have a functioning market economy as well as a stable democracy. These requirements could be compelling countries in this region to democratize. Compared to Central and Eastern Europe, Latin America has little impact from international organizations with a demanding agenda of imposing democracy in the world. It is true that some Latin American countries experienced regime changes under the sweeping trend of the third wave in the late 1970s and 1980s, but the R square in the regional model is very close to the overall model. Thus, unlike Central and Eastern Europe, Latin America does not have any overwhelming effects other than those within the model. This study shows that FDI contributes positively to political liberalization in this region. This region exhibits a higher level of democracy than the rest of the developing world. The mean value of the Freedom Index is 9.54, almost 2.5 points higher than the average value of the developing sample countries. Among the 15 Latin American countries with available disaggregated FDI data, 11 have over 60 percent of FDI from democratic economies and with less than 30 percent of FDI in the primary sector. Table 4 and Table 5 show that FDI in the Asian-Pacific area exhibits a negative influence on democratization. The political development of Asia-Pacific as a whole started from a much lower base than the aforementioned two regions. The average value of the Freedom Index in this region is 7.56, slightly higher than the developing world average, but less than the average values for Latin America and Central and East Europe. Composition of FDI in Asia-Pacific is more heterogeneous than they are in the other two regions. The sample has 21 Asian and Pacific countries, among them 20 have available disaggregate FDI data. 11 countries with available data fall into the category with a high percentage of FDI from democracies and a low percentage of FDI in the primary sector, 9 of them with relatively high level of democracy. Another six countries in this region depend largely on FDI from the developing world, all with relatively low levels of democracy. Africa and the Middle East are characterized as having low levels of political and economic development. The mean value of the Freedom Index for the Middle East and North Africa is 5.8, 1.28 points lower than the average value of the developing world.

15 120 African countries in the sample have the lowest average Freedom Index: 5.66, 1.42 points lower than the developing world average. The analysis indicates that FDI has a negative effect on democratization both in Africa and the Middle East. The features of FDI flows to these two regions may provide an explanation for this result: most of them are from advanced democratic countries but with a very high focus on natural resource exploitation. Recently, there has been a dramatic increase of FDI inflows from developing countries such as China, Russia, and neighboring countries in Africa. TABLE 5. POLITICAL EFFECTS OF FDI INFLOWS/GDP ACROSS REGIONS Variables Eastern and Latin Asia and Africa Middle East Central Europe America Pacific and North Africa FDI inflows/gdp t *** 0.06*** -0.11*** -0.06*** 0.06 (0.017) (0.02) (0.04) (0.015) (0.04) Trade t *** 0.007*** *** (0.0023) (0.002) (0.002) (0.002) (0.003) Prior 0.43*** 0.36*** 0.366*** 0.4*** 0.29*** Democracy (0.012) (0.01) (0.013) (0.01) (0.015) Inflation t *** *** * ( ) ( ) ( ) ( ) (0.002) Logged GDP 1.2*** 0.86*** 0.56*** 0.29*** 0.62*** Per capita t-1 (0.09) (0.078) (0.1) (0.07) (0.08) GDP growth * *** Rate t-1 (0.01) (0.013) (0.018) (0.008) (0.012) Urbanization t *** (0.007) (0.004) (0.006) (0.004) (0.005) Democratic *** -0.06*** Regional (0.013) (0.003) (0.005) (0.008) (0.08) Effect t-1 Constant *** 6.8*** 4.9*** 1.27 (1.07) (0.56) (0.59) (0.33) (0.9) Observations R square Notes: Panel corrected standard errors in parentheses below the regression coefficients. Two-tailed tests: * sig. at.05; **sig. at.01; ***sig. at.001.

16 121 TABLE 6. POLITICAL EFFECTS OF FDI STOCKS/GDP ACROSS REGIONS Variables Eastern and Latin Asia and Africa Middle East Central Europe America Pacific and North Africa FDI stocks/gdp t *** 0.007* * *** -0.12*** (0.0045) (0.003) (0.006) (0.0035) (0.004) Trade t *** 0.01*** *** 0.02*** (0.0023) (0.002) (0.002) (0.002) (0.004) Prior 0.43*** 0.37*** 0.4*** 0.4*** 0.31*** Democracy (0.012) (0.013) (0.012) (0.012) (0.016) Inflation t *** *** * ( ) ( ) (0.0013) ( ) (0.002) Logged GDP 1.2*** 0.9*** 0.52*** 0.304*** 0.71*** Per capita t-1 (0.09) (0.08) (0.08) (0.07) (0.09) GDP growth *** Rate t-1 (0.012) (0.014) (0.02) (0.01) (0.015) Urbanization t *** ** (0.007) (0.005) (0.004) (0.005) (0.005) Democratic *** -0.06*** Regional (0.016) (0.005) (0.0064) (0.01) (0.1) Effect t-1 Constant *** 7.18*** 4.9*** 0.41 (1.36) (0.65) (0.5) (0.4) (1.17) Observations R square Notes: Panel corrected standard errors in parentheses below the regression coefficients. Two-tailed tests: * sig. at.05; **sig. at.01; ***sig. at.001. CONCLUSIONS In summary, this study demonstrates that different types of FDI can have different impacts on democracy. The reason for the unreliable empirical results in previous works is the dual political effects of FDI: some components of FDI are pro-democracy (e.g., FDI from developed democracies), hence contributing to democratic development, while other parts of FDI are pro-authoritarian (e.g., FDI in the primary sector), fostering autocratic governance. In other words, not all forms of foreign investment seem to be beneficial to political development in the developing world. Another important finding was regional differences of the political effects of FDI due to uneven distribution of FDI from advanced democratic countries and FDI in the primary sector. Previous studies often assumed that the political effects of FDI are homogenous across quite different regional and national contexts and tried to make a universal conclusion about the impact of FDI on democratic development in developing

17 122 countries. In contrast, this study suggested that inflows of FDI into different regions vary greatly in terms of their impacts on the domestic political development. This research holds important implications for the policy debate about the political effect of FDI in developing countries and for future research as well. International institutions like the World Bank, the International Monetary Fund (IMF), and the OECD all emphasize the role of FDI as a vital catalyst to political development efforts in the South (UNCTAD, 2006; Kebonang, 2006). However if only considering the result of the political effects of overall FDI, it casts doubt on the liberal consensus that integration into the global economy has a direct positive effect on the domestic democratization. For example, Li and Reuveny (2003) suggest a policy dilemma between economic efficiency gains and democratic decline due to increasing negative political effects of FDI. Some critics even call for harsh economic sanctions against repressive authoritarian regimes as effective means to promote political change and human rights (e.g., Gallagher, 2002). This study shows that FDI is not necessarily problematic in terms of its role in political development. There is a third option to break the policy dilemma. This finding is substantial for developed democracies. After World War II, the spread of democracy has been a desirable policy goal for advanced democratic countries and Washington-based institutions to reduce regional conflicts and secure peace in the world. Many policy makers in developed democracies have long assumed that capital flows and financial liberalization encourage democratic improvement in developing countries (Rudra, 2005; Armijo, 1999; Rodrik, 1997). Findings in this study should help to eliminate the doubts of policy makers about the negative political effect of FDI and encourage more investment from democratic countries and in non-primary sector to the developing would, especially to economies with authoritarian or communist regime type. The regional variation of the effect of FDI on democracy also suggests that more FDI in manufacture, service, and technology sectors from developed democracies are needed in Asia, Africa, and the Middle East. If economic efficiency maximization is the only policy objective, then it does not matter if FDI exerts different political effects for the policy makers in developing countries. However more and more developing countries have begun to recognize the importance of democratic governance to economic development. For those governments that want both economic competitiveness and democratic accountability, findings here are encouraging because they suggest that countries do not need to face the so-called policy dilemma since the direct investment from foreign investors does not necessarily contradict domestic political improvement in terms of democracy. This study provides a policy recommendation that maximizes both economic and political gains by attracting more FDI from developed democracies and reducing the concentration of FDI on natural resource exploitation. This analysis provides a preliminary assessment to address the dual political effects of disaggregated FDI in a large sample of developing countries. The findings and implications should be interpreted with the following limitations in mind. First is data availability. Although the dataset used is the best current option, only about half of the sample countries have available disaggregated FDI data, leading to a limited generalization. Moreover, with the increasing amount of the data, it is possible to categorize FDI into four sub-groups based on sector and regime type of the source country and cross examine the political effect of FDI in different scenarios. This could be

18 123 a direction for the further study. The second limitation is plausible indirect effects of FDI. Beyond the direct political effects, FDI could also influence the level of democracy through other socioeconomic factors. Thus, including the socioeconomic control variables could blur the real effect of FDI. Further research on this matter could complete the understanding of the political effects of FDI in the developing world. REFERENCES Acemoglu, D., & Robinson, J. A. (2001). A Theory of Political Transitions, American Economic Review, 91(2), Alfaro, L. (2003). Foreign Direct Investment and Growth: Does the Sector Matter? (Working Paper). Boston: Harvard Business School. Amable, B., Gatti, D., & Schumacher, J. (2006). Welfare State Retrenchment: The Partisan Effect Revisited, Oxford Review of Economic Policy, 22(3), Armijo, L. E. (1999). Mixed Blessing: Expectations about Foreign Capital Flows and Democracy in Emerging Markets, in L. E. Armijo, ed., Financial Globalization and Democracy in Emerging Markets, (London: Palgrave Macmillan), pp Beck, N., & Katz, J. N. (1995). What to Do (and Not to Do) with Time-Series Cross Section Data, American Political Science Review, 89(3), Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression Diagnostics: Identifying Influential Data and Sources of Collinearity, New York: Wiley. Biglaiser, G., & Staats, J. (2010). Do Political Institutions Affect Foreign Direct Investment? A Survey of U.S. Corporations in Latin America, Political Research Quarterly, 63(3), Bornschier, V., & Chase-Dunn, C. (1978). Cross-National Evidence of the Effects of Foreign Direct Investment and Aid on Economic Growth and Inequality: A Survey of Findings and a Reanalysis, American Journal of Sociology, 84(3), Bussmann, M., de Soysa, I., & Oneal, J. (2005). The Effect of Globalization on National Income Inequality, Comparative Sociology, 4(3-4), Cheng, H. (2003). Analysis of Panel Data, Cambridge: Cambridge University Press. de Soysa, I. (2003). Foreign Direct Investment, Democracy, and Development: Assessing Contours, Correlates, and Concomitants of Globalization, London: Routledge. Diamond, L. (2002). Thinking about Hybrid Regimes, Journal of Democracy, 13(2), Dutta, N., & Osei-Yeboah, K. (2010). A New Dimension to the Relationship between Foreign Direct Investment and Human Capital: The Role of Political and Civil Rights, Journal of International Development, online version. Firebaugh, G. (1992). Growth Effects of Foreign and Domestic Investment, American Journal of Sociology, 98(1), Freedom House (2010). Freedom in the World 2010: The Annual Survey of Political Rights and Civil Liberties, Washington D.C.: Freedom House. Gasiorowski, M. J. (1995). Economic Crisis and Political Regime Change: An Event History Analysis, American Political Science Review, 89(4),