Regime Type, Societal Conflict, and Restriction of Foreign Investment

Transcription

1 Regime Type, Societal Conflict, and Restriction of Foreign Investment Jia Chen Department of Political Science University of Colorado, Boulder Abstract While the existing literature suggests political liberalization enhances the freedom of Foreign Direct Investment (FDI), the observation that some countries with more liberal political regimes have more restrictive policies toward direct investment than others remains unexplained. The existing theories on the relationship between political liberalization and FDI have not taken into account the domestic distributional consequence of cross-border flows of industrial capital. With a model of plural political competition and empirical evidences gathered from panel data, I argue the liberalizing effect of political liberty on foreign investment hinges on the local factor endowment. This theory explains why political liberalization could not always help lifting restrictions on FDI. Address: Campus Box 333, Boulder, CO jiac@colorado.edu

2 1 Introduction International flows of industrial capital to developing countries have experienced dramatic increase in the past few decades as more and more countries in the third world becomes democratic. Inspired by the conventional wisdom that economic and political liberalization go hand in hand, much of the recent scholarship explores the impact of domestic political change on the flows of Foreign Direct Investment (FDI). In particular, scholars have been interested in understanding the relationship between political liberalization and the change in the policy toward foreign investment. As recent works on the politics of foreign investment question the conventional belief which posits that the recipient nation as a whole uniformly benefits from the inflow of industrial capital, more and more attention has been drawn to the pattern of domestic distributive conflict accompanying cross-national flows of capital and their political implications. 1 Based on the observation that political liberalization in developing economies tends to grow the political influence of the labor, the biggest beneficiaries of direct investment inflows, recent studies have found that democratization has contributed to more liberal policies in the developing world towards inflows of direct investments. While political liberalization undoubtedly elevates the political influence of labor by expanding the voting rights, the domestic capital owners, in defense of their interests, could also make political efforts to counteract labor s power if the loss from investment liberalization outweighs the cost of lobbying. Moreover, the loss incurred on owners of capital by investment liberalization tends to increase in the local scarcity of capital as a factor of production. The distributive conflicts between labor and capital would thus be more intense in capital scarce or labor abundant economies. Instead of letting the labor s influence to dominate the policy outcome, political liberalization alters the rule of the game by which capital owners could still effective compete with the labor in shaping the policy. While political liberalization has been shown to lead to 1 While political liberalization has be shown in existing studies to have the effect of increasing the supply of FDI through consolidating the security of foreign assets and enhancing the efficiency in domestic economic governance, the domestic demand for foreign industrial capital has to be filtered through a highly political process of policy making before resonating with the supply of foreign investment.

3 more liberal policy towards investment inflows, the key underlying mechanism characterizing the competition between the labor and capitalist in the changing political context, particularly the strategic response from the domestic capital owners to the growth of labor s political influence, remains largely unexplored. With a factoral model of political competition on FDI policy and the empirical evidences garnered from panel data analysis, the rest of this paper shows that the interaction between the domestic political institution and the local factor endowment constitutes a key shaping factor of the domestic capital owners strategy in affecting the policy on foreign investment. The main argument of the paper is derived from a theoretical model that tracks down the impact of political liberalization on restrictions of foreign direct investment through characterizing the strategic competition between labor and capital in the context of developing countries. This theoretical approach is taken in an attempt to provide a fuller account for the variation in the level of restriction on direct investment in the developing world. Even after controlling for levels of political liberalization, it is still puzzling to see some of the developing countries being more friendly to FDI inflows than others. Argentina and Brazil display comparable average level of political liberty between 1995 and 2002 with Argentina scoring an average of 7.5 and Brazil scoring similarly 7.7 in Polity 2 score. Yet Argentina scored an average of 70 in Foreign Investment Freedom Index while Brazil scored only 50. Why is Brazil less friendly to foreign investment than Argentina? Data shows that the capital stock per worker in Argentina is 1.6 times greater than that in Brazil, indicating the domestic capital owners in Brazil has more to lose from foreign capital inflows than in Argentina. Given the two states were at similar levels of political liberalization, liberalization of foreign investment would face greater opposition in the state with scarcity in capital, leading to more restrictive policy toward FDI. The comparison is preliminary but it illustrates the importance of incorporating the factor endowment into the picture in understanding the impact of political institutional change on investment policy. Both theoretical predictions and empirical evidences presented in this paper suggest that political change and factor endowment interact with each other in determining the equilibrium level of investment restrictions imposed by the government.

4 The rest of the paper is organized as follows. I first present an overview of existing literature on the political determinants of restrictions on foreign investment and highlight the inadequate understanding of the strategy of domestic capital owners in vying for the favorable policy. I then develop a simple theoretical model of plural political competition from which key hypotheses are generated regarding the equilibrium behavior and outcome. I then move on to test these hypotheses using a dataset composed of 82 developing countries across 9 years from 1995 to Distributive Politics and Investment Policy in Changing Political Context Recent research in international political economy has seen the emergence of a distinctive theoretical approach referred to as the Open Economy Politics by Lake (2009). Open Economy Politics studies the patterns of societal preferences and outcomes of economic policies in an open economy through characterizing of the domestic distributional consequences of international economic interchanges. It is well known that the cost and benefit of global economic cooperation are unevenly distributed among domestic groups and the ensuing conflicting preferences between the winners and losers structure the domestic cleavages over economic policy. The Open Economy Politics approach well applies to the study of foreign investment liberalization. In the Stolper-Samuelson framework, factor endowment remains a key concept in understanding the motivation of cross-border capital flows and the structure of the ensuing domestic distributional conflict. Although there is a lack of directly evidence linking factor endowment with the motives of cross-border flows of industrial capital, the disparity in specific economic, technological, and market conditions between capital-abundant and capital-scarce economies constitutes the driving forces of investment flows to the developing world. 2 More concretely, foreign investments motivated by high 2 The existing economics literature, summarized in Helpman (2006), has not been able to identify significant empirical association between factor endowment and destination of FDI.

5 capital returns in capital-scarce economies falls into the catergory of efficiencyseeking international capital. Efficiency gains from higher capital returns is a direct result of the distinctive market structure in capital-scarce economies. Factors such as cost structure, market concentration, and low regulatory policy makes inflows of capital more profitable than in advanced economies. This observation is confirmed theoretically and empirically in Markusen and Venables (2000), Konings (2001), and Yeyati, Panizza, and Stein (2007). The domestic distributional implication of industrial capital inflows is salient. Domestic labor benefits the most from the formation of new industrial capital which brings about more job opportunities and higher wages. Domestic capital owners, on the other hand, are harmed because the marginal return to capital decreases as the incoming foreign capital drives up the cost of labor and unfavorably alters the market structure by stiffening competition. Support for this factorial characterization of distributional implication of FDI can be found in Pandya (2010) where it is suggested that the individual preference of regarding FDI inflows is consistent with the factorial model of capital mobility. Pinto and Pinto (2008) also identified evidences of factorial divide in FDI preference in advanced industrial democracies in their study of the effect of partisanship of incumbent on FDI inflows. Although their theoretical model is built on top of a sectorally structured economy, the central argument and reasoning is elaborated in a characterization of the conflict of interests between domestic labor and capital. Pinto and Pinto argues that partisanship of the incumbent determines the political favorability of direct investment inflows. Pro-labor governments are more prone to liberalize FDI that complements labor input such that increased industrial capital inflows would benefit their core constituent - domestic labor - more. On the other hand, domestic firms who would not want to face the competition from foreign enterprise resulted from horizontal or vertical FDI will exert pressures on the government for restrictive policy towards foreign investment as Li (2006) and Garland and Biglaiser (2008) suggested. This finding, however, should be cautiously interpreted in the context of the theory developed here because these studies focuses on the observed flows of FDI, which is significantly shaped by national policy of foreign investment.

6 The competition between labor and capital over FDI policy is politically structured. In a capital-scarce economy ruled by a non-democratic regime, the domestic capital owners are more advantageous than the domestic labor in shaping the policy outcome, primarily ascribed to their economic advantages as well as their capability of overcoming collective action problems among the relatively small number of actors involved. While labor can be nominally organized in unions which seem to carry some political influence in some nondemocratic regimes, the real strength of labor is weak in a political system build on the concentration of political power. 3 As a result of the sheer disparity of power and influence between the two groups of actors, the political elites in non-democratic regimes systematically favor the interests of the capitalists by restricting the inflow of foreign capital. Political liberalization alters the landscape of the competition by distributing more de jure political power to labor. When the political franchise is expanded to include the majority of the citizens, the political incumbent will weigh the preference of the labor more in making investment policies given the affinity of labor s policy preference to that of the median voter. The political incumbent under the electoral pressure is more likely to accommodate the preferences of labor. Liberalization of investment very naturally follows the empowerment of labor in the process of political liberalization. This logic underlies the key argument in the most recent studies on democratization and liberalization of foreign investment represented by Pandya (2014). The empowerment of labor after political liberalization, however, does not direct eliminate the de facto power of the capitalists, which lies in their collective organizational capacity and economic advantage. While the political incumbent is prompted to take labor s preference over investment policy more seriously, the foundation of the de facto power of the domestic capital owners remains intact and there is no reason for them to simply given in in the contest. The immediate period following political liberalization could actually 3 The presence of unions and labor organization in authoritarian regimes actually undermines the influence of labor in that the leaderships of such organizations are usually allied with the ruling elites and are paid patronage to not further the rights and welfare of the working class. For a treatment on this problem in the Latin America context see Levitsky and Mainwaring (2006).

7 provide more opportunities for political corruption as the evidence in Sung (2004) suggested. Even in cases where it may become more costly and difficult for the capital to influence the policy outcome after political liberalization due to diminishing political opportunities for rend seeking, domestic capital owners could yet counteract the growing influence of labor through lobbying and other economic effort executable in the new political environment. Liberal political change could drastically reshape the distribution of political power among domestic groups but would hardly eliminate certain groups of actors completely from the competition. Instead, it makes the political competition more competitive and strategic. In that vein, what remains unexplored in the existing literature is the strategic response from the domestic capital owner as the former vested interest group to the changing political landscape. In particular, it is imperative to understand how the capitalist could use their economic advantages strategically to counter the de jure power of domestic labor. This particular side of the story of investment liberalization that characterize the strategic choice of the vested interest in the changing political context is largely missing in the existing research. Moreover, the strategic response of the vested interest to political liberalization interacts with the government in an intricate way that brings about a steady policy that stabilize the behavior of both actors. Thus it is critical to rigorously characterize the strategic pattern of interactions in order to fully understand the underlying political process of investment liberalization. To highlight the theoretical approach of this paper, two pieces of existing research on policies toward FDI worth mentioning. Taking a similar theoretical perspective, Pinto and Pinto (2008) studied the impact of the ideological bias of the incumbent government and the sectoral allocation of FDI inflows. With an extensive economic model, they formally theorize the political dynamics and economic consequences of the conflict of interests between labor and capital over investment inflows. The theoretical framework of this paper differs from that of Pinto and Pinto s in that the strategic political actions of the societal actors are not explicitly incorporated in their argument. Admittedly

8 the strategic response of the vested interest is of less significance in scenarios of ideological heterogeneity of governments, it constitutes a key aspect in understanding the strategic political dynamics of economic liberalization in times of political change. Relatedly, the recent research by Pandya (2014) empirical evaluated the relationship between democratization and FDI liberalization. Pandya s paper shares a common focus with the theory developed here on the connection between the distributive effect of foreign investment and political liberalization. The key argument developed here, however, is derived from a niche but important extension of the perspective of Pandya s paper: what factor affects the way in which the domestic capital owners respond to the reshaped political landscape, and how does such response influence the trajectory of economic liberalization following democratization? The analysis in the next section takes on these questions formally through developing a parsimonious theoretical model of political competition over FDI policies. 3 A Model of Plural Political Competition and FDI Restrictions To focus on the political dynamics of the policy making process, the theoretical model presented here is developed with parsimonious assumptions regarding the economic mechanisms of foreign investment. The model is written in a way to capture firstly the impact of political institutions on the process and outcome of the societal contest between pro-liberalization and anti-liberalization forces in the society, and secondly the strategic response of the organized interest groups to transformation of political structure and redistribution of political power and influences. In particular, the model theorizes the linkage between economic motivation structured by the endowment of production factors and the strategic incentive in the political competition. There are three actors in the model: the government, domestic capital owner, and domestic labor. The government is assumed to be self-interested in that it maximizes the political rent to be extracted. The domestic capi-

9 tal owner and labor are engaged in a political competition with one another, attempting to influence the government s policy on industrial capital inflows. The government is facing conflicting preferences of capital and labor and will set an optimal policy that maximizes government s utility function given particular political settings. First of all, let the capital-labor ratio, or capital stock per labor, on the world market be r w [0, 1] and the domestic capital-labor ratio is r d [0, 1]. Country A is abundantly endowed with labor which indicates r d < 1. Assume the domestic capital-labor ratio is lower than the world average, that is r d < r w. The gap between domestic and world capital-labor ratios is given by λ = r w r d. The marginal return to the input of production factors is reflected by the capital-labor ratio: the more abundant one production factor gets, the smaller marginal returns the owner receives. The domestic return for capital as production factor is given by r 1 d is given by r d. and the domestic return for labor Let R [0, 1] denote the level of restriction on capital inflow imposed by the government. R = 0 indicates the government imposes no restriction on capital inflow and R = 1 means the government prohibits any foreign capital inflow. The freedom of cross-border capital flow is consequently given by β = 1 R. The restriction on capital inflow determines to what extent the gap will be filled by inflows of foreign capital. rd P is the domestic capital-labor ratio in presence of government restrictions on FDI: r P d r d + β(r w r d ) = r d + (1 R)λ (3.1) Let the total population of the country be an unity. Capital is impersonated: one unit of capital is owned by one unit of the population. The mass of domestic capital factor owners is given by (r 1 d +1) 1 and the mass of domestic labor is (r d + 1) 1. Note (r d + 1) 1 + (r 1 d + 1) 1 = 1. The aggregated level of welfare of domestic capital owners, w K, and domestic labor, w L, is thus given by: w K = r 1 d (r 1 d + 1) 1,

10 and w L = r d (r d + 1) 1. The total welfare of capital and labor owner taking into account the effect of government restriction, R, on capital inflow is thus: w K = (r P d ) 1 (r 1 d + 1) 1 = (r d + βλ) 1 (r 1 d + 1) 1, and w L = r P d (r d + 1) 1 = (r d + βλ)(r d + 1) 1. As has been discussed in the previous section, the government s utility function has two sources, namely popular political support and political contribution. The utility function of the government is given by: u G = w K (R) + α w L (R) + c(r), (3.2) where α represents the coverage of political franchise and c represent the political contribution offered by domestic capital factor owners. w K (R)+α w L (R) is the utility gain from popular political support and c(r) is the utility gain from capital owners political contribution. c (R) > 0 and c (R) > 0, such that greater political contribution requires higher level of restrictions and the marginal gain in political contribution increases in the level of restrictions. Substantively, c (R) > 0 also suggests from the perspective of capital owners that the marginal return of political contribution decreases in the present level of restrictions. It takes less political contribution to induce greater restriction when the restriction is low. At this point, it is assumed that the coverage of political franchise does not affect the channeling of capital owners interests into government s concern for political support, because capital owners are better positioned than labor to act collectively to influence the government regardless of regime type. The value of α only affects the extent to which labor s interest, w L (R), is taken seriously by the administration. The model, particular reflected by the parameter α, is being structured such a way to allows the continuity in the different forms of political regimes in the analysis. The parameter α aligns democratic

11 and non-democratic regimes along a continuum which allows the succeeding analysis to draw parsimonious implications on the equilibrium outcome with regard to the differences in political structures. The utility function of a domestic capital owner is the aggregated welfare gain minus political contributions made to the government, which is given by: u K = w K (R) c(r). (3.3) Given the utility function of government and domestic capital owners, it fits into a typical principal-agent model framework. The government, as the agent, determines the value of R that maximizes u G. Given the maximization mechanism of the government, capital owners determine a level of contribution c as a function of R which maximizes u K. From solving the principal-agent model for the optimal contract arrangement we can derive the equilibrium policy R E from which the effect of capital-labor ratio and coverage of political franchise in shaping the equilibrium can be elaborated. 3.1 Equilibrium and Comparative Statics Solving the principal-agent model for the optimal contract arrangement, which is a Nash Equilibrium of the game played by government and domestic capital owners, I established the following propositions. Proofs of proposition and corollaries can be found in Appendix A. Proposition 3.1. In the unique equilibrium, the equilibrium level of government restrictions on capital inflow, R E, is provided by: R E 2rd r d α = 1 λ α (3.4) R E is the equilibrium level of restrictions on capital inflow from which both government and capital owners simultaneously fulfill their maximization objectives. From the function of R E we can observe that the equilibrium level of restriction is a function of the domestic capital-labor ratio r d, the magnitude of the difference between domestic and international capital labor ratio, λ, and

12 the type of political regime indicated by the coverage of political franchise, α. With the equilibrium level of restrictions specified, I am now able to study the comparative statics focusing on the cross-border gap of capital-labor ratio, λ, and regime type, α. Corollary 3.1. The first derivative of R E with respect to λ is positive: R E λ > 0, given r d [0, 1], α [0, 1]. A positive first derivative of R E with respective to λ means that as the value of λ increases, the equilibrium level of government restriction increases. A higher value of λ, which indicates a greater gap between the domestic and world capital-labor ratio, would result in higher levels of policy restrictions on capital inflow. The logic underlying this comparative statics can be interpreted intuitively: a larger gap between domestic and world capital-labor ratio means domestic capital owners are earning a lot more relatively than their foreign counterparts, which gives them a strong incentive to lobby the government to establish restrictions on capital inflow. The price of such protection for domestic capital owner, c, will also be high under this circumstance. The huge benefits from maintaining tight investment restriction, however, offsets the high cost. The hypothesis derived from Corollary 3.1 is stated as follows. Hypothesis 1. The level of the capital-labor ratio is negatively correlated with the level of restrictions on capital inflow imposed by the government. Corollary 3.2. The sign of the first derivative of R E with respect to α depends on the value of λ : R E α = 2rd r d α r d 2λα 3/2 2λα > 0, if λ < 0 0, if λ 0 According to this corollary, the equilibrium level of restrictions indicated by R E will decrease as the coverage of the franchise, α increases only when the domestic capital-labor ratio is lower than the world average(λ < 0). The

13 logic underpinning this corollary is also straightforward. Given the economy is relatively scarce in capital, since a change in the coverage of political franchise primarily affects the extent to which the interests of domestic labor are taken seriously by the government, a higher coverage of political franchise will result in the government putting greater weight on the welfare of domestic labor while deciding the level of restriction on capital inflow. Foreign investment benefits domestic labor by enhancing the marginal return for labor as production factors and policies made under a democratic political regime that restricts such a trend will cause massive discontent from labor who constitutes majority of the constituency in labor-abundant economy, jeopardizing the political viability of the incumbent government. In a scenario where capital is relatively abundant in the economy, labor would turn to support restrictions on cross-national flows of capital because it would result in decreases in marginal returns to labor. This generalizes a scenario where political liberty may not induce liberalizations of foreign direct investment. Corollary 3.2 establishes another hypothesis formulated as follows. Hypothesis 2. When capital is the scarce factor, political freedom, measured by coverage of political franchise, is negatively correlated with the level of restrictions on capital inflow imposed by the government. Corollary 3.3. The first derivative of dre dα zero: 2 R E α λ > 0, given r d [0, 1], α [0, 1]. with respect to λ is greater than This corollary is the most important result generated by the model. It is known from the Corollary 3.2 that greater coverage of the domestic franchise results in a lower level of policy restrictions on capital inflow. The magnitude of α s effect on the equilibrium restriction level depends on the value of λ, namely the gap between the domestic and world capital-labor ratio. The greater the value of λ, the greater increase of R E will be as a result of a one unit increase in the coverage of political franchise. The practical interpretation of this corollary is that the welfare implication of capital inflows for labor is also dependent on a comparison of the domestic to world capital-labor ratio. The greater the gap between the domestic and world capital-labor ratio is, the

14 greater impact that capital inflow will cause to labor, and consequently, the greater the political bargaining power will be unleashed by political liberalization. If, however, the gap between the domestic and international capital-labor ratio is small, indicating a small ensuing utility gain by labors from capital inflows, the effect of political liberalization on lowering the level of restriction will be quite limited. A small material basis of the economic motivation of political action diminishes the significance of political liberalization on the liberalization of international investment. Hypothesis 3 generalizes Corollary 3.3 as follows. Hypothesis 3. The effect of a change of regime type on liberalizing FDI is dependent on the factor endowment of the host country. The lower the capitallabor ratio, the greater the effect of regime type on liberalizing FDI will be. 4 Empirical Evidence 4.1 Data, Variables, and Hypotheses The equilibrium analysis of the political competition model presented above enables me to generate hypotheses for empirical tests. Constrained by the availability of data on restriction on foreign investment on years before 1995, I use a panel dataset consisting of data for 82 developing countries from 1995 to 2003 covering six variables. The following linear specification is used as the basic model for analyses in this section. Foreign Investment Freedom ij =β 0 + β 1 Regime Type ij + β 2 K-L Ratio ij + β 3 Regim Type ij K-L Ratio ij + γ Control Variables ij (4.1) The dependent variable, Foreign Investment Freedom, measures the level of government restriction on direct investment. Data on Foreign Investment Freedom and Business Freedom are collected from the annual report of Index of Economic Freedom from 1995 to 2003 published by Heritage Foundation and

15 Wall Street Journal (2009). Polity 2 score from the Polity IV project (Marshall and Jaggers, 2007) is used to document differences in political regime type. Data on capital-labor ratio, GDP, and Population was collected from the third version of the Extended Penn World Tables (Foley, 2008) which covers 120 countries from 1963 to [Table 1] The Regime Type variable is measured by Polity 2 score in the Polity IV dataset (Marshall and Jaggers, 2007). The Polity IV dataset assesses a country s level of democracy and level of autocracy on two separate 0-10 interval variables that reflect non-overlapping types of information about a political system. Regime Type variable in the model that I constructed represents the coverage of political franchise which characterizes the identity of the median voter in the political system. Capital-labor ratio data is collected from EPWT 3.0 dataset. Capital-labor ratio measures the ratio of the number of participants in the labor force to the total stock of physical capital which reflects the country s relative endowment in capital. A low capital-labor ratio reflects that the country is relatively abundant with labor while a high capital-labor ratio reflects relative abundance in capital. The expected sign of coefficients in (4.1) is summarized in Table 1. The level of restriction on foreign investment is measured by the Index of Investment Freedom in the annual report of Economic Freedom published by Heritage Foundation and Wall Street Journal (2009). The investment freedom index is a combination of indices capturing eight different aspects of government restrictions on foreign business and investment. These aspects are listed as follows. Restrictions on foreign ownership of business; Restrictions on industries and companies open to foreign investors; Restrictions on performance requirements on foreign companies;

16 Restrictions on foreign ownership of land; Equal treatment under the law for both foreign and domestic companies; Restrictions on the repatriation of earnings; Restrictions on capital transactions; Availability of local financing for foreign companies. These eight categories provide specific measurements of policy restrictions on the operation of foreign industrial capital. In particular, most of the eight indices capture the discriminatory treatment of foreign enterprises on the policy dimension, which very well matches the theorization of the concept in my theoretical model. The index provides a full characterization of the restriction and obstacles targeting foreign industrial capital including ownership restrictions, domestic and international finance, and restricted entry to specific sectors. Besides regime type and factor endowment, there are other variables which could also intervene in determining the equilibrium level of government restrictions on capital inflow. These variables are likely to associate with both foreign investment freedom and one or more of the independent variables. I control for three main control variables and the hypotheses regarding these variables are discussed below Size of Economy The size of a country s economy is believed to be positively correlated with more liberal policy toward foreign investment. Chan and Mason (1992) find that direct investment from advanced economies is concentrated in developing countries with relatively larger economies. Nunnenkamp and Spatz (2004) find that variables measuring the market size of the host country have a significant correlation with government s policy regarding foreign capital inflow. Kobrin (2005) argues that developing countries with larger scale of economy are more confident of their ability to maintain a positive benefit-cost ratio for FDI through negotiation with foreign investors. The size of economy is

17 measured by real GDP collected from the ETWP 3.0 dataset. Hypothesis 4. The recipient country with a larger economy adopts more liberal policies towards FDI Economic Freedom I create a variable to measure domestic economic freedom using two indexes, Business Freedom and Freedom from Corruption, from the 2009 Report on Economic Freedom (Heritage Foundation and Wall Street Journal, 2009). This variable is intended to measure the overall business freedom in the domestic economy on a universalistic basis. The difference between this variable and investment freedom is theoretically and empirically relevant. While it is reasonable to expect correlation between foreign investment freedom and domestic business freedom, economic policies can be biased toward either domestic or foreign businesses. One country can have a fairly liberal policy toward domestic business and more restrictive constraints on foreign companies as a result of strong preference for protectionism among the domestic audience, as commonly observed in Latin America. The gap between domestic economic regulations and constraints on foreign capital merits attentions in studying the effect of political change on liberalizing foreign investment inflows. Hypothesis 5. A higher level of domestic business freedom in a host country results in a more liberal policy towards FDI Natural Resource Abundance Foreign investors in different industries have different motivations in their incentives and bring about different distributional consequences to the recipient country (Pinto and Pinto, 2008; Schulz, 2009). In contrast to other types of FDI recognized for their merit of promoting advances of technology in host country, foreign investments in resource-oriented sectors convey little benefit in technology transfer and oftentimes result in damages to the environment. Busse (2004) suggests resource-oriented investment is more likely to flow to countries ruled by autocratic regimes. Li (2006) extends the claim to the policy dimension arguing that greater political liberty is expected to result in fewer

18 investment incentives and more policy restrictions in resource-rich economies. This intervening effect of resource endowment is examined in Hypothesis 6. Hypothesis 6. There is an intervening effect between regime type and resource abundance on FDI restrictions. As the level of resource abundance increases, the liberalizing effect of political liberty on FDI restrictions decreases. 4.2 Regression Results and Discussion The dataset used for the empirical analysis consists of 82 countries cross nine years ( ). A common way to handle panel data is pooling the observations and estimating the coefficients with OLS estimator. The major problem with this method is that the standard errors of the OLS estimator are affected by the presence of heteroskedasticity. Beck and Katz (1995) recommended Panel Corrected Standard Errors (PCSE) of the OLS estimator as a remedy. But such an approach leaves the heterogeneous unit-fixed effect, which is a primary concern in panel data, intact. Fixed or random effect panel regression model is believed to be better at addressing the problem of group-specific heteroskedasticity. I use these methods to examine the data and obtained preliminary evidences supporting my hypotheses. I also conduct robustness checks with a dynamic model specification and an alternative measurement of investment freedom Pooled OLS Regression with Panel Corrected Standard Errors (PCSE) The first set of estimations is done by running a pooled OLS regression with Panel Corrected Standard Errors (PCSE) to remedy heteroskedasticity. The result is presented in Table 2. The key hypothesis on the level of investment freedom, Hypothesis 3, is supported by the results. In Models 1 and 2, the coefficient of the interaction term Regime Type K-L ratio is negative and statistically significant, confirming that the negative effect that the level of the capital-labor ratio has on the effect of political liberalization on freedom

19 of foreign investment. Based on the model specification in Equation 4, the marginal effect of Regime Type on Investment Freedom is given by ME(Regim Type) = β 1 + β 3 K-L Ratio. [Table 2] Notably, estimates in Table 2 show the marginal effect of political liberty varies by the value of capital-labor ratio. The marginal effect of political liberty and the histogram of the density of K-L ratio is plotted in Figure 1. It is evident in the figure that the liberalizing effect of political liberty on foreign investment is diminishing as capital becomes more abundant. Such a result confirms Hypothesis 1 and 2 where the marginal effect of political liberty on investment freedom is predicted to be positive but decreasing in the local capital-labor ratio. [Figure 1] The results also show that domestic economic freedom has a positive and independent effect on investment freedom. The coefficient of the interaction term Economic Freedom Regime Type is negative, indicating that political liberalization will have smaller effect on investment freedom in countries with a higher level of domestic business freedom which is counter-intuitive. In line with the existing theories, the coefficient of resource abundance is negative and significant. But the positive and significant coefficient of the interaction term Regime Type Resource Abundance is contrary to the expectations of Hypothesis 6. The positive coefficient of Regime Type Resource Abundance indicates that for a country with a greater reservation of natural resources political liberalization is likely to result in a greater liberalizing effect on investment policy. The Size of Economy does not gain statistical significance in any of the models in Table 2. [Table 3]

20 4.2.2 Random Effect (FGLS) Models OLS estimator with PCSE is relatively weak in tackling the problem of unit heterogeneity in panel data. I analyze the panel dataset with Feasible Generalized Least Squares (FGLS) regression and a random unit effect specification. The results of the FGLS regression are exhibited in Table 2. The results generated from FGLS regressions are consistent with those generated in the pooled OLS model. The key hypothesis that K-L ratio has a negative intervening effect with Regime Type on freedom of foreign investment is again confirmed by a negative and statistically significant interaction term Regime Type K- L ratio. The marginal effect of Regime Type, as plotted in Figure 2, gains statistical significance except for the limited cases with very high capital-labor ratio. Also, similar to the results from pooled regressions the Economic Freedom variable is shown in the results to have a positive and significant effect on investment freedom and a negative intervening effect with Regim Type in shaping the level of investment freedom. The results also show a negative independent effect of Resource Abundance on investment freedom and a positive intervening effect of Resource Abundance on the relationship between Regime Type and investment freedom. The effect of Size of Economy on investment freedom is again shown to be insignificant. The Busch-Pagan LM test barely rejects the unit-homogeneity hypothesis (p =.06), suggesting the random effect model and pool OLS estimation should produce similar results. [Table 4] Fixed Effect Regressions Random effect models are known to produce more efficient estimates than fixed effect models, but the orthogonality condition must be checked upon to make sure the unit-specific effect is not correlated with any of the independent variables included in the model. A fixed effect model estimates the effect of variables of interests taking into account the structure of panel data by rearranging the dataset to enable estimation of unit or time specific effects. The strength of using fixed effect models in analyzing panel data resides in the multiple benefits of yielding year or unit specific intercepts. Fixed effect models,

21 particularly within effect models, not only address the problem of heterogeneity and serial correlation but also tackle the threats from omitted variable as the effect of all other unobserved variables, which vary across countries but not over time or vary over time but not across states, can be captured in time or group specific dummy variables. I use both within effect and between effect model in analyzing the data, the result of which is exhibited in Table 4. Some discrepancies emerge between the results in previous analyses and the fixed effect regression model. The key hypothesis about the intervening effect of K-L ratio is no longer supported in the fixed effect model. The coefficient of the interaction term Regime Type K-L ratio becomes insignificant in both within effect and between models. Coefficients of most independent variables in the model are not consistently significant in the within effect and between effect models except for economic freedom whose coefficient is positive and statistically significant in all four models presented. But the Hausman Specification test I run following the fixed effect regressions suggests both the fixed effect model and random effect more should produce consistent estimates, partly confirming the orthogonality condition of using the error-components model. Given that fixed effect models is well known to be less efficient than the random effect models, the results from random effect regressions and Pooled OLS estimations do constitute evidences supporting the key prediction of the theoretical model. 4.3 Robustness Check Using an alternative measure of restrictions on direct investment, I conduct a robustness check of the regression results presented above. The financial openness index compiled by Chinn and Ito (2008) measures the openness of investment based on the restrictions on capital account which I use to quantify the restrictions on foreign direct investment. The Chinn-Ito index provides a better year coverage which enables the robustness check to be done with observations across 33 years from 1971 to Note capital account restriction is used as a tool to mainly stabilize flows of portfolio investment. It nevertheless makes direct investment more difficult by raising the cost associated with general movements of capital in and out of the country. When the government

22 attempts to restrict direct investment inflows, raising the cost of direct capital inflows would be a direct strategy. To control for the systematic effect of world financial environment, I also add the average level of financial openness as a control variable. The model used in the robustness check adopts a distributedlag specification to counteract time-dependent effect. [Table 5] The regression output shows the coefficients of Financial openness t 1, Capital-Labor Ratio, World financial openness and World financial openness K-L Ratio are significant and consistent with the results from pooled regression with panel corrected standard errors. Regime Type K-L Ratio also a gain negative sign and significant in Model 3. The marginal effect is plotted in Figure 3 which suggests in line with the theory that the liberalizing effect decreases in K-L Ratio and becomes insignificant if capital is extremely abundant. As for the appropriateness of the model, the Hausman specification test favors fixed effect model to random effect model. The hypothesis that the difference in coefficients is not systematic is rejected in all three specifications. 5 Concluding Remarks This paper combines a model of plural political competition with econometric evidences in an attempt to identify the political and economic determinants of policy restrictions on the inflow of foreign industrial capital in the developing world. With a model of political competition between domestic capital owners and labor, I show formally that the structure of domestic political regime and factor endowment, are pivotal in determining the optimal level of restriction on foreign capital inflows. Domestic factor endowment, measured by the capital-labor ratio, characterizes the pattern of the distributional conflict over liberalization of foreign economic policy between capital and labor. The greater the difference between the domestic and world capital-labor ratio, the more that domestic capital owners will suffer from the liberalization of FDI. Democratization is characterized by the broadening of franchise and enhanced

23 political freedom which elevates the influence of domestic labor s interests in the government s decision-making process. To guard the established interests against the expansion of labor s power, domestic capital owners are willing to make high political contributions to lobby the government to maintain restrictions on capital inflow. Government has to find a point along its policy continuum that balances the gain from accepting contributions from capital owners and the necessary political support from labor that is critical for securing the political tenure. There is an interaction effect between the domestic factor endowment and political liberalization in determining the equilibrium level of foreign capital restrictions. The effect of political liberty on lowering the level of restriction on capital flow depends on the gap between the domestic and world capital-labor ratio. The greater the cross-border capital-labor ratio gap is, the greater welfare effect that capital inflow is for labor, and consequently, the greater the political motivation will be released from broadening franchise. If the capital-labor ratio gap is small then the effect of political liberty on removing restrictions will be insignificant. Econometric analyses of the propositions generated by the formal model lend supportive evidences for the key hypothesis regarding the effect of political liberty dependent on the domestic capital-labor ratio controlling for economic freedom, natural resource abundance, and the size of the economy. One important implication from the study connects the level of development with investment freedom. A practical interpretation of the meaning of the capital-labor ratio of a country is related to the level of development indicated by the capital accumulation. The most important finding in this study of the effect of political liberalization on lifting restrictions of foreign investment is dependent on the domestic capital-labor ratio is that the specific level of development at which a country democratized affects the accompanying effect on lifting foreign investment restrictions. If political liberty takes place at a low development level, political liberalization will unleash greater forces of liberalization as a result of the greater cleavage segregating capital and labor, which pushes foreign investment liberalization further than in countries with higher development levels indicating a less salient cleavage between capital and labor. It can also be expected that political liberalization in undeveloped countries

24 is likely to see significantly enhanced freedom of foreign investment while in more developed states the effect of political liberty on investment freedom is likely to be smaller, holding other conditions constant. Such an implication may also be illuminating for solving the contradictory empirical observations produced in existing studies of political determinants of FDI. If the intervening effect of development level, measured by the capital-labor ratio, and political change is taken into account, political liberty will only result in a significant increase in foreign investment inflows in relatively undeveloped economies.

25 A Proof of Propositions and Corollaries The government maximizes its utility with: whose first order condition provides: max [w K(R) + α w L (R) + c(r)], R w K (R) R + α w L(R) R + c(r) R = 0 (A.1) As the assumption of the welfare functions of the actors suggest, the first term in (A.1) is positive indicating capital owners prefers high levels of restriction, where the second term is negative indicating labor prefers low levels of restrictions. The third term is also positive, meaning the political contribution from the captial owners increases in the level of restrictions. The domestic capital owners maximize u K = w K (R) c(r): whose first order condition of (A.2) provides max R [w K(R) c(r)], (A.2) w K (R) R c(r) R = 0 (A.3) This may seem counter-intuitive that the capital owners maximize their gain with regard to R but not to c. This is because the capital owners anticipated the function of optimal R for the government as a function of c based on (A.1) and could just find the best R given the cost of inducing it implicitly defined by c(r). The optimal political contribution c(r) as a function of R will later be obtained explicitly. Intuitively, (A.3) simply requires from the perspective of the capital owners that the marginal gain equals the marginal cost from raising the restriction on capital inflows. Substituting (A.1), or c(r) R that = w K(R) R α w L(R) R in (A.3), it is obtained αλ 1 + r d 2 w K(R) R + α w L(R) R = 0 (A.4) 2λ (1 + r 1 d )(λ λ R + r d) = 0 2 (A.5)