Democracy and Economic Development: A Simultaneous Equation Analysis

Democracy and Economic Development: A Simultaneous Equation Analysis Uk Heo Department of Political Science University of Wisconsin-Milwaukee P.O.Box 413 Milwaukee, WI 53201 heouk@uwm.edu Abstract Despite a plethora of studies on the relationship between democracy and economic development, a dominant theoretical framework has yet to emerge. Since Seymour Martin Lipset s (1959) seminal work, the notion that economic development leads to democracy has been widely accepted in the literature. There is also a debate on the effects of regime types on economic performance. Some scholars argue that democracy promotes economic growth while others contend that democracy hampers economic growth. In a recent study, Gerring et al. (2005) developed a new concept, democracy stock, to incorporate institutional maturity. This concept is theoretically and empirically useful because institutional maturity matters. To address the potential issue of endogeneity between democracy and economic development, I investigate the relationship using a simultaneous equation method and data from 1950 to 2000. The empirical results show that democracy and economic development help each other. *I would like to thank John Bohte, Marc Ethridge, Aie-Rie Glasure, Jonathan Krieckhaus, and Terence Roehrig for their helpful comments and suggestions. I would also like to thank John Gerring for kindly sharing his data with me. All the remaining errors are mine. Earlier versions of this paper were presented at the University of North Texas, Texas Tech University, and the annual meeting of the Southwest Political Science Association in Houston, Texas, March 31- April3, 2010. (Under review at Comparative Politics; Nominated for the best paper award at the annual meeting of the Southwestern Political Science Association)

INTRODUCTION Despite a plethora of studies on the relationship between democracy and economic development (level of wealth measured by per capita GDP), there is yet to be a dominant theoretical framework or set of empirical findings. Since Seymour Martin Lipset s (1959) seminal work, the notion that economic development leads to democracy, the so called modernization perspective, has been widely tested and accepted in the literature with an abundance of empirical evidence (e.g. Barro 1990; Bollen 1979, 1993; Burkhart and Lewis-Beck 1994; Cutright 1963; Dahl 1989; Feng 1997; Londregan and Poole 1994; Przeworski 1991). In fact, Brunk, Caldeira, and Lewis-Beck (1987) argue that economic development accounts for the transition to democracy more than the combined effects of any other variables (e.g. education, industrialization, foreign influence) included in quantitative studies. Burkhart and Lewis-Beck (1994) go even further arguing that Lipset s notion of economic development as a prerequisite of democracy is one of the few iron laws that exist in political science. The question remains unsettled, however. Neubauer (1967) and Jackman (1973), for example, contend that the relationship between economic development and democracy is curvilinear rather than linear. According to them, economic development does lead to increases in the level of democracy up to a certain point, but the effects disappear after a while because of the decline of marginal effects over time. Other researchers find no empirical evidence to support the idea that economic development leads to democracy (Arat 1988; Gonick and Rosh 1988). Przeworski et al. (2000: 137) writes, [w]e failed to detect any thresholds of development that would make the emergence of democracy predictable. In sum, modernization theory appears to have little, if any, explanatory power. There is also a vast volume of studies exploring how regime type affects economic 1

growth (rate of wealth accumulation measured by per capita GDP growth rates). Some studies report positive effects of democracy on economic growth while others suggest a negative relationship or no relationship between the two variables. For instance, a group of scholars (e.g. De Haan and Sierman 1995; Feng 1997; Leblang 1994; Olson 1993; Quinn and Woolley 2001) assert that democracy is better for economic performance because of the way it protects individual property rights and promotes political stability. In contrast, another group of scholars (Chirot 1977; Rao 1985) argue that democracy hampers economic growth in less developed countries or newly democratized countries because economic growth is often incompatible with other goals associated with democracy, such as socioeconomic equality. Gerring et al. (2005) argue that the institutional maturity of a system has a crucial effect on economic performance. To assess the institutional maturity of democracies, Gerring et al. developed a new measure, democracy stock. The scores on this measure are obtained by summing the Polity 2 variable score since 1900 in the Polity IV data set. The Polity 2 variable score provides a single regime score between -10 and +10 with full autocracy at -10 and full democracy at +10. If a country maintains a democratic system, its democracy stock score will continue to rise. On the other hand, if democracy does not survive, its democracy stock score will decline as the Polity 2 variable score will be negative. Considering that institutions mature over time, Gerring et al. (2005) assume that institutional maturity will be higher in countries where democracy stock scores are high. Their findings demonstrate that more mature democracies enjoy better economic performance. This is an important finding making a significant contribution to the literature. Thus, there are two schools of thought regarding the relationship between democracy and economic performance. One school of thought maintains if economic development leads to 2

democracy while the other perspective believes democracy/regime types influence economic growth. These two schools have not communicated with each other on the grounds that the former school of thought focused on economic development level of prosperity commonly measured by GDP per capita--while the latter perspective often employed economic growth quantitative expansion in their analysis using the growth rates of GDP or GDP per capita. Gerring, et al s (2005) suggestion of using democracy stock to account for institutional maturity introduces a new dimension in studying the relationship between democracy and economic performance on the grounds that democracy stock includes both the survival and maturing of democracy. Thus, this paper will employ the democracy stock concept and data from 1950-2000 to investigate whether economic development leads democracy to survive and mature. If economic development helps democracies to survive and mature, it may also be the case that a maturing democracy reinforces the relationship by facilitating further economic development. The contribution of this paper is three fold. First, using the theories developed in two schools of thoughts in the relationship between democracy and economic performance, I attempt to link the two perspectives. Second, potential endogeneity between democracy and economic performance has been frequently raised although most studies did not empirically consider in their analysis. Using a simultaneous equation analysis method I address the endogeneity issue in the democracy-economy nexus. Third and finally, following Gerring et al (2005) I take into account institutional maturity in the analysis of the relationship between democracy and economic performance by employing Gerring et al s democracy stock. The findings of this study will provide a better understanding of the long studied relationship between democracy and economic development. 3

THE THEORETICAL RELATIONSHIP BETWEEN DEMOCRACY AND ECONOMIC PERFORMANCE The Effects of Economic Development on Democracy There are a number of theoretical channels through which economic development can be linked to democracy. First, economic development leads to democracy through improvements in human capital. The rationale behind this argument is that economic development improves literacy and education levels due to reinvestment in human capital. Citizens who are better educated should understand their political rights and national governance better than those who lack education. Under these circumstances, citizens demand a political system that provides the best environment for them to exercise their political rights, which is a democracy (Burkhart and Lewis-Beck 1994). Since economic development also increases the size of the middle class whose political attitude is presumed to be conducive to democracy, the demand for democracy goes up as the economy develops further (Haggard and Kaufman 1995: 25; Huntington 1991; Lipset 1959; Muller 1997; 133-4). In other words, education resulting from economic development contributes to political and social modernization. Another theoretical connection between economic development and democracy is that economic development leads to democracy through industrialization (Bollen 1979; Dahl 1989; Huntington 1991; Lipset 1959; O Donnell and Schmitter 1991). Industrialization leads to urbanization which also improves the quality of human capital through enhanced education, increases the number of workers, and allows people to access information on democratic political systems and human rights in other countries. These changes in society along with newly acquired information lead to a pluralistic society and undermine the legitimacy of the nondemocratic government system. As the political environment gradually changes in favor of political liberalization, elites desire to emulate democratic nations as they seek to survive in the 4

changing political environment. For instance, economic development in South Korea brought in industrialization, urbanization, and a rise in education levels. These changes led to an increased social demand for democracy. In response to the changed political environment, authoritarian leader Chun Doo-whan allowed a direct, popular, presidential election in 1987, which expedited the transition to democracy in South Korea (Heo and Roehrig 2010; Roehrig 2002, 2004). Norberg (2003: 38) nicely summarizes the theoretical link discussed above: The accelerating spread of information and ideas throughout the world, coupled with rising education standards and growing prosperity, is prompting demands for genuine political rights. Many quantitative studies in political science and sociology have provided substantial evidence on this argument (e.g. Almond 1991; Burkhart and Lewis-Beck 1994; Bollen and Jackman 1985; Brunk, Caldeira, and Lewis-Beck 1987; Feng 1997; Helliwelli 1994; Jackman 1973; Lipset 1959; Przeworski and Limongi 1997). In contrast, some scholars (e.g. Arat 1988; Gonick and Rosh 1988; Przeworski et al. 2000; Vanhanen 1990) find weak evidence or no evidence at all that economic development influences democracy. The general argument is that economic development does not necessarily help democratization because there are many factors that affect the transition to democracy. Nations go through various experiences of political growth because of their unique political cultures, environments, and histories. For instance, countries like India and the Philippines went through democratization before economic development; in contrast, economic development preceded the transition to democracy in South Korea and Taiwan (see Haggard and Kaufman 1995; Roehrig 2002, 2004). O Donnell and Schimitter (1986) assert that there are too many factors that influence the transition to democracy, making it virtually impossible to generalize the causal direction between 5

economic development and democracy. Among the many factors they identify are choices made by authoritarian leaders and democratic activists, the domestic and international political environment, unclear motives and interests of relevant parties, the talents of individual politicians, and the level of economic development. According to Przeworski et al. (2000: 137), wealthy countries tend to be democratic not because democracies emerge as a consequence of economic development under dictatorships but because, however they emerge, democracies are much more likely to survive in affluent societies. The causal relationship between democracy and economic development, accepted for years following Lipset s 1959 contribution, is increasingly unclear. Haggard and Kaufman (1995: 28) contend that the relationship between level of growth and regime type seems indeterminate among middle-income countries. In fact, democratization occurs after poor economic performance in some developing countries (e.g. Argentina, Bolivia, Brazil, the Philippines and Uruguay). Since good economic performance allows authoritarian leaders to justify their leadership, economic development strengthens the dictatorship and delays transitions to democracy in some cases (e.g. Singapore, South Korea, and Taiwan). Moreover, political leaders ability to effectively mobilize political support or opposition also affects transitions to democracy because economic-social structure and economic policy and performance affect various groups differently (Haggard and Kaufman 1995; Jesse et al. 2002; Roehrig 2002, 2004). This is the reason that Samuel Huntington (1991: 38) asserts no single factor is necessary or sufficient to account for transitions to democracy in general. Heo and Tan (2001), on the other hand, argue that the direction of the causal impact between economic growth and democracy may vary because of differences in political and economic environments. In their Granger causal analysis, they found that improvement in the 6

level of democracy preceded economic growth in 10 of 42 developing countries while economic growth resulted in higher levels of democracy in 11 out of 42 cases. Only three countries showed the feedback relationship between the level of democracy and economic growth whereas 8 countries showed no relation between the two variables. Thus, they conclude that it is difficult to generalize the causal direction between the level of democracy and economic development in developing countries. The Effects of Democracy on Economic Performance According to Sirowy and Inkeles (1990; see also Feng 1997), there are three schools of thought on the effects of democracy on economic performance: 1) the conflict perspective; 2) the compatibility perspective; and 3) the skeptical perspective. According to the conflict perspective, democracy hampers economic growth in less developed or newly democratized countries because a democratic political system is costly and ineffective (Crozier, Huntington, and Watanuki 1975; Chua 2003). The conflict perspective posits that democracy is not compatible with economic growth for three reasons: 1) the role of interest groups; 2) myopic policy decisions for electoral success; and 3) labor unions. First, according to this perspective, interest groups in democracies influence policy making by lobbying or making demands for the interests of specific groups at the expense of national growth (Krueger 1974; Olson 1982). As a result, government spending increases to meet the demands, which results in tax increases. Under these circumstances, keeping social spending in check is difficult (Haggard 1990: 262) and thereby private investment decreases which slows economic growth (Kurzman, Werum, and Burkhart 2002; Przeworski et al. 2000: 142-143). This is the reason Rao (1985: 74-5) contends, The resources necessary for 7

investment cannot be accumulated by democratic means. 1 This effect is particularly harmful to newly democratized countries because the transition to democracy leads dispersed interest groups to organize to aggregate and articulate their interests (e.g. enhancement in social welfare programs). As a result, the new government faces greater demands and pressures. However, political institutions in the newly democratized countries tend to be immature and unprepared to deal with the increased demands (Huntington 1968; Rao 1985; Przeworski and Limongi 1993; Barro 1996; Kaplan 2001; Zakaria 2003; Chua 2003). 2 Second, Comeau (2003) argues that elected officials in democracy in general make decisions based on electoral success. Since elections are held every few years, policy decisions made by elected officials tend to be myopic and ineffective in the long run. Third, according to Gupta et al. (1998), labor unions in democracies constantly demand higher wages and better benefits, As a result, business profits become smaller, international competitiveness declines, and eventually, economic growth slows down. Thus, democracy is not conducive to long-term development (see Barro 1996) In summary, the conflict perspective insists that democracy forces officials to attend to conflicting and incompatible policy goals economic growth and socioeconomic equality at the same time. Authoritarian regimes are better for economic growth in less developed countries because basic civil and political rights need to be suppressed in order to have effective government involvement in developing the economy, nearly always a necessity in these countries (see Comeau 2003; Huntington 1987; McGuire and Olson 1996.). For this reason, 1 Kaplan (2001: 62) claims, If a society is not in reasonable health, democracy can be not only risky but disastrous. Zakaria (2003) also argues that democracy is attributable to corruption and lawlessness in Africa. The reason is that when institutional maturity is lacking, elected leaders in newly democratized countries in Africa distort the system and abuse their power for personal gain. 2 Olson (1982) argued that the interest group problems that produce economic decline may take many years, perhaps many decades, to emerge. 8

Amsden (1989) claims that East Asia under authoritarian regimes was able to achieve impressive economic development by avoiding the negative effects of lobbying by special interest groups (see also Heo and Kim 2000; Jesse et al. 2002). According to the compatibility perspective (Barro 1990; Feng 1997; Gerring et al. 2005; Olson 1993; Przeworski 1991; Quinn and Woolley 2001), democratic governments are better for economic performance because they attempt to allocate resources effectively due to their concern over electoral outcomes. If political leaders and government officials pursue self-interest at the national expense, voters and social groups can vote them out of office (see Krieckhaus 2006). As a result, democracies tend to privatize state-owned companies more than authoritarian regimes because the private sector has demonstrated better productivity than the public sector (see Biglaiser and Danis 2002). In other words, politicians in democratic systems are accountable and realize the ramifications of poor economic performance. Thus, they focus on economic development in their policy-making decisions (see Przeworski et al. 2000: 142-145). According to Olson (1993), democracy helps economic growth because it has the institutions that are most conducive to economic growth. Sirowy and Inkeles (1990: 133-4) also contend that democratic processes and the exercise of civil liberties and political rights lead social conditions to be friendlier to economic growth than other types of government. The reason is that the political pluralism allowed in democracies provides outlets for the public s energies and creates environments that promote entrepreneurial risk and economic growth. Thus, as Rodrik (2000) noted, democratic institutions provide an inclusive conflict management mechanism for social groups. Thus, democracies are more stable than other types of governments and tend to have better communication between the government and the public than autocracies (Feng 1997; Kurzman, Werum, and Burkhart 2002; Quinn and Woolley 2001). 9

De Haan and Sierman (1995) as well as Leblang (1994) assert that democracy indirectly helps economic growth because it protects individual property through institutional checks and balances which prevent systemic abuse and/or predatory behavior that happens in authoritarian governments; this in turn boosts private investment and indirectly helps economic growth (see also Olson 1993). Pastor and Sung (1995: 225) also wrote, a potentially positive relationship between democracy and investment may exist precisely because more open political systems allow policymakers to better read and respond to distributive pressures and thereby reduce social conflict; we also suggest that democracy is likely to allow private capitalists (as well as other citizens) institutional access to policymakers. By contrast, the unpredictability of autocracy and the lack of systematic checks on autocratic government lead entrepreneurs to hesitate when considering investment decisions which may result in low domestic private investment in non-democracies. This may be the reason that Jones and Olken (2004) find the end of autocracy resulted from the death of the authoritarian leader tends to bring about better economic performance. Olson (1993: 572) nicely summarizes the argument as follows, (a)n economy will be able to reap all potential gains from investment and from long-term transactions only if it has a government that is believed to be both strong enough to last and inhibited from violating individual rights to property and rights to contract enforcement... Interestingly, the conditions that are needed to have the individual rights needed for maximum economic development are exactly the same conditions that are needed to have a lasting democracy. 10

Another reason for concluding that democracy is a better system than other types of government is that democratic governments often adopt a capitalist economic system to ensure market liberalism (Lindblom 1977). This enhances market competition, improves individual property rights, thereby leading to higher productivity and greater domestic and foreign investment respectively (see Biglaiser et al. 2007). In contrast, authoritarian leaders tend to have unlimited power which often results in cronyism and corruption (Bueno de Mesquita et al. 2001). According to Heo and Tan (2001: 463), individuals in an economy want the maximum confidence that any property they accumulate will be respected, and only democratic societies provide the confidence. For this reason, democracy attracts foreign investment better than nondemocracies (Biglaiser et al. 2008). Finally, the skeptical view notices variations in economic performance among nations with the same type of government concluding the lack of a systematic relationship between democracy and economic growth. The argument here is that economic policies and institutional structures matter more than regime types. According to Pye (1966), there is no universal relationship between democracy and economic performance because economies can develop in both democratic and authoritarian frameworks. Proponents of this perspective (Clague et al. 1996; Helliwell 1994; Przeworski 2004; Przeworski et al., 2000; Gasiorowski and Poptani 2006) argue that having a democratic government alone means very little for economic growth because so many factors can influence economic performance. 3 They contend that economic policy decisions, the way that policies are implemented, institutional maturity, and the coordination of government organizations play important roles in economic performance, and that these factors 3 Sala-i-Martin (1997), in his cross-national study of economic development, included 67 independent variables. Furthermore, various growth models include different combinations of control variables due to the lack of consensus on model specification. 11

are more important than the presence or absence of democracy (see Jesse et al. 2002). Moreover, Przeworski and Limongi (1993), in a review of empirical studies investigating the relationship between regime types and economic growth, found inconclusive evidence to support the claim that the economies of democracies tend to grow faster than those of nondemocracies. Helliwell (1994) and Przeworski et al. (2002) also report an insignificant relationship between democracy and economic growth. In more recent studies, Krieckhaus (2004, 2006) found inconsistent effects of democracy on economic growth over various time periods and different regions. 4 Thus, coherent theoretical perspectives and existing empirical findings support three different conclusions regarding the relationship between democracy and economic development. There are reasons to conclude that democratic government facilitates economic growth, that it undermines economic growth, and that the effect of democracy on economic growth is dwarfed by other factors. Economic development may lead to democracy or it may help authoritarian regimes stay in power. The present article is an effort to resolve the inconsistencies in the literature. RESEARCH DESIGN Theory and Hypotheses The disagreement among researchers regarding the relationship between democracy and economic performance may be a function of the fact that, as Gerring, et al (2005) noted, democracies may go through different economic development experiences because nations vary 4 Krieckhaus (2004) reports that democracy hampers economic growth in Latin America and Asia, but it promotes growth in Africa. In another study (2006), he finds that democracy had detrimental effects on economic growth in the 1960s, but facilitated growth in the 1980s. No significant impact was found in the 1970s. 12

in terms of their institutional maturity. Older democracies, thanks to greater institutional maturity, are likely to enjoy the benefits of democracy while newly democratized countries or less developed countries may not because of the lack of stabilized institutions or resources. In contrast, Olson (1982) asserts that mature democracies may suffer from slow growth because of powerful interest groups influencing policy decisions. It is also possible that there is no significant relationship between democracy and the economy (both economic growth and economic development) because there could be so many factors that influence the relationship between the two variables. Previous studies have also noted the possibility of simultaneity/endogeneity in the relationship between democracy and economic performance (e.g. Burkhart and Lewis-Beck 1994, Feng 1997; Helliwell 1994; Heo and Tan 2001; Krieckhaus 2004; Przeworksi et al. 2000) although Helliwell (1994) as well as Burkhart and Lewis-Beck (1994) did not find the endogeniety problem. It is difficult to test if democracy helps economic development because most democracies are relatively wealthy. Moreover, the level of democracy does not change once it reaches a certain level although the economy may continue to grow. However, we can assume as a country maintains democratic government, their institutions become more mature rather than they become more democratic. It only makes sense to test if democracy performs better than other types of government in economic growth. Gerring et al. (2005), found the positive effects of democracy stock on economic growth. The implication of this finding is that mature democracies enjoy higher rates of growth, suggesting that their per capita GDP (economic development) is likely to increase more in mature democracies than immature democracies. In other words, the maturity of democracy is likely to have a positive effect on economic development as well as economic growth on the 13

grounds that continuing economic growth will increase per capita GDP. They argue that endogeneity between democracy stock and economic growth is not a concern because economic growth has no impact on accumulated democratic experience and institutional maturity which they take into account in their measure of democracy, democracy stock. 5 The rate of annual growth may not have a meaningful effect on democracy stock as long as the economy is growing because democracy will mature as it survives and lasts longer. However, if the economy declines and democracy does not survive, the system cannot mature. Accumulated democracy experience measured by democracy stock depends on democracy survival and maturing on the grounds that democracy level score will plummet as the regime goes back to authoritarianism. In other words, democracy stock score will decrease if democracy does not survive. It will only increase as democracy survives and lasts. According to Przeworski et al. (2000: 142-145), democracies are likely to survive and mature when their economies develop because democracies are more sensitive to economic performance than nondemocracies. They (2000: 109) report that, When they [democracies] face a decline in income, they die at the rate of 0.0512, so that about one in twenty of them dies, but when incomes are growing they die at the rate of 0.0152, one in sixty-six. When a longer-term economic performance--three or more consecutive years of growth--is considered, the survival rate of democracy significantly increases (only 1 in 135 dies). In another work, Przeworski (2004) argued that education, income distribution, political institutions, and the relations of political forces all have some impact on the survival of democracy, but per capita income has far greater 5 Gerring et al. (2005) argue that endogeneity between democracy and growth should not be a concern because some previous studies found no relationship between democracy and growth (Helliwell 1994; Londregan and Poole 1996). Moreover, it seems even less likely that a country s growth performance in time T would have any effect whatsoever on its democracy stock at T-1 (stock being a measure that extends back over many decades) (Gerring et al. 2005: 345). 14

effect. Przeworski and Limong (1997) also noted that continuing economic development plays a key role in the survival of democracies in less developed countries. 6 In other words, when democracy is combined with good economic performance, these governments are likely to survive and mature, which may provide the empirical evidence of Dahl s (1990, 2001) argument that democracies tend to mature as they exist longer. This means a potential feedback relationship between democracy stock and economic development. Based on the discussion above, the relationships between democracy stock and economic development can be hypothesized as follows: Hypothesis 1 Economic development helps democracy survive and mature Hypothesis 2 The maturity of democracy helps economic development. To test these hypotheses, I develop two empirical models based on extant literature controlling for relevant variables. The models are 7 : 6 Democracies in Latin American during the 1980s experienced economic downturns after transitions, which led t a debt crisis across the region. Many worried about whether the transitions would hold; in the end, democracy survived. Thus, not all democracies enjoy good economic performance after transitions, nor it is necessary to have economic development for democracies to survive and mature (see Roehrig 2002). 7 According to Krieckhaus (2006), recent studies (e.g. Bleaney and Nishiyama 2002) on economic growth are increasingly excluding investment from the growth equation because investment is endogenous to economic growth. Moreover, democracy may promote economic growth by inducing greater investment. In other words, investment is an intervening variable rather than an independent variable (Krieckhaus 2006: 326). Although the dependent variable in this equation is economic development, not growth, the same line of argument is applied. 15

(Equation 1) Democracy Stock = Economic Development + Economic Openness + Years of Independence + English Legal Origin - Social Conflict (Equation 2) Economic Development = Democracy Stock + Population Growth + Education + Life Expectancy + Government Expenditures - Inflation Social Conflict Since these equations may have a potential feedback relationship, I conduct the Hausman specification test for simultaneity first. The test is based on the logic that, [w]hen simultaneity is present, one or more of the explanatory variables will be endogenous and therefore will be correlated with the disturbance term (Pindyck and Rubinfeld 1998: 353). The test was conducted using per capita GDP for economic development and democracy stock. The Granger causal analysis is also an option for testing exogeneity, but, I chose the Hauseman specification test because the data used in this study are time-series-cross-sectional while the Granger test is designed for time-series data. The test is performed with a two-stage procedure as follows. First, I regress one of the endogenous variables (EV1), economic development, on all the exogenous variables in the model (the reduced form specification of the empirical model) to obtain the residual (R1) and the fitted value of EV1, economic development. Then, I regress the other endogenous variable (EV2), democracy stock, on the expected value of EV1, economic development, and the residual (R1) obtained from the previous regression to perform a t-test on the coefficient of the residual 16

(R1). Under the null hypothesis of no simultaneity, t-statistic will be statistically insignificant meaning the coefficient is zero; otherwise simultaneity exists in the system of equations (Pindyck and Rubinfeld1998: 353-5). The test results indicate that simultaneity does exist in the relationship between democracy stock and economic development (t-statistic = 23.31). Endogenous Variables: Democracy Stock and Economic Development A number of indicators have been developed to measure democracy (e.g. Polity IV, Gastil s civil liberty measure, Coppedge and Reinicke s polyarchy scale, Arat s democracy level score, Bollen s democracy rating). According to Inkeles (1990: 5-6), The indicators most commonly selected to measure democratic systems generally form a notably coherent syndrome, achieving high reliability as measurement scales. As a result, these indicators show high levels of association with correlations ranging between 0.83 and 0.94 (Przeworski et al. 2000: 56). Among them, one of the most commonly used democracy measures in recent studies is the level of democracy by Polity IV (e.g. Gerring et al. 2005; Krieckhaus 2004, 2006; Rodrik and Wacziarg 2005; Tavares and Wacziarg 2001). The Polity IV data set constructed annual measures for both institutionalized democracy and autocracy as many polities exhibit mixed qualities of both of these distinct authority patters (Marshall and Jaggers 2009: 13). The measures are based on the selection process of the executive, the level of constraints on executive power, and the types of political competition. Scores range between -10 and + 10, with full autocracy at -10 and full democracy at +10. The Polity variable in the Polity IV data set is derived by subtracting the autocracy value from the democracy value to provide a single regime score. The Polity 2 variable is a modified version of the Polity variable smoothed for time-series analysis by adjusting transitions or missing cases to the score between -10 and +10 in 17

the context of adjacent periods. For instance, country X has a POLITY score of -7 in 1957, followed by three years of -88 (missing data) and, finally, a score of +5 in 1961. The change (+12) would be prorated over the intervening three years at a rate of per year, so that the converted scores would be as follows: 1957-7; 1958-4; 1959-1; 1960 +2; and 1961 +5 (Marshall and Jaggers 2009: 16). Gerring et al. (2005) add the Polity 2 variable score since 1900 to reflect the accumulated democracy experience. According to Gerring et al. (2005: 339), the Polity2 score is highly sensitive (it employs a twenty-one-point scale) and offers extensive country coverage (all sovereign polities except microstates) and good historical coverage. However, skeptics point out that the Polity2 score itself reveals only the level of democracy without reflecting the institutional maturity of each country. Moreover, there is a wide range of variation in democracies when juxtaposing economic performance and their institutional maturity (see Przeworski 2000). Mature democracies (e.g. USA, UK, France etc.) tend to enjoy the benefits noted by the compatibility perspective including protection of individual property rights, political stability, and good communication between the government and the public. On the other hand, the drawbacks of democracy pointed out by the conflict school are often found in newly democratized or less developed countries, including the inefficiency of government and corruption. In other words, the difference in political institutional maturity is likely to affect economic performance, although the democracy level scores may be the same. According to Gerring et al. (2005), institutional maturity with respect to the structure of the executive branch of the government (parliamentary/presidential), electoral laws, and protection of individual property rights requires time because institutions become mature as the experience of democracy dealing with these issues accumulates over time. Barro (1996) also 18

recognized the time factor in effectiveness of democratic governance and used two lag periods to test the democracy-growth nexus. Recognizing the significance of institutional maturity, Rodrik and Wacziarg (2005) separately enter into their empirical model the variable of established democracy defined as the period five years after the transition to democracy. To incorporate institutional maturity in democracies, Gerring et al. (2005) developed a measure by adding up each country s Polity2 score (-10 to 10; -10 is the highest level of autocracy; 10 is the highest level of democracy) from 1900 to the present year with a 1 percent annual depreciation rate which they call democracy stock. This measure provides a way to estimate democracy while accounting for the level of institutional maturity. In summary, the two measures of democracy--the level of democracy and democracy stock--provide different information. The level of democracy is sensitive to political development and regime change, but the democracy stock variable reflects the political institutional maturity of the system. Turning to the measurement of economic development, a number of indicators have been employed in studying economic development (e.g. per capita GDP, energy consumption etc.). According to Przeworski et al. (2000: 81), however, One might question that per capita income is an adequate indicator of the level of growth. (per capita) income may not correspond perfectly to whatever one means by growth. Yet in our view it is the best indicator better than energy consumption, literacy, industrialization, and other alternative measures. Thus, I employ per capita GDP for economic development in this study. The data is from Gerring et al. (2005) which is per capita GDP in natural logarithm based on the World Development Indicators measured in constant dollars and additional data for the 19050s from the Penn World Table (PWT) 6.1 data set (Chain index, constant dollars). 19

Pre-Determined Variables Economic Openness As a nation trades more goods and services, people in the nation become more exposed to different types of government including democracy. In addition, as a country goes through industrialization, it tends to trade more goods and services. Trade openness also shows how liberal a country s society and economy are (Reuveny and Li 2003). Thus, the more open the country, the more like it is to democratize and foster an environment that allows democracy to survive and mature. To measure economic openness, total trade as percentage of gross domestic product is used. Years of Independence According to Gerring et al. (2005), a country s colonial history affects the country s political development process because how long ago a country was colonized affects political environment for democracy to survive and mature as well as the transition to democracy. The number of years a country has been sovereign is employed to measure this variable. English Legal Origin Legal origins theory posits that the origin of a country s laws significantly affects its legal rules and regulations. According to Mahoney (2001: 504), English common law was developed because landed aristocrats and merchants wanted a system of law that would provide strong protections for property and contract rights, and limit the crown s ability to interfere in markets. In other words, countries with English legal origins have a better environment for democratization as well as the survival and maturing of democracy (La Porta, Lopez-de-Silanes, and Andrei Shleifer and Robert Vishny 1999; La Porta, Lopez-de-Silanes, and 20

Andrei Shleifer 2007). This is a dummy variable; countries with English legal origins were coded 1 while others were coded 0. Social Conflict Countries with civil and ethnic conflicts experience difficulties in transitioning to democracy (Vanhanen 1990; see also Roehrig 2002, 2004). Moreover, social conflicts provide unfriendly political environment for democracy to survive and mature. At the same time, social stability is crucial in economic performance, and social conflicts seriously jeopardize stability (see Feng 1997; Quinn and Woolley 2001). Thus, this variable enters both equations. I employ the measure created by Gerring et al. (2005) by adding the number of incidents of civil violence, civil war, ethnic violence, and ethnic war to form a composite index. Education Barro s (1990, 2001) endogenous growth model posits that the quality of human capital plays an important role in economic performance (see also Barro and Lee 2001). Due to the lack of annual data available for educational attainment for all the countries and time period under investigation, the natural logarithm of illiteracy rates used by Gerring et al. (2005) are employed as a proxy of education. Population Growth Fertility growth theories suggest the inclusion of population growth in the economic growth model based on the logic that increases in family size reduce parents average investment in each child, which lowers the quality of human capital (see Krieckhaus 2006). Since population is also an important factor in economic development due to its contribution to labor input for economic output, I include this variable in the development model. Growth rates of total population are used for this variable. 21

Life Expectancy Human capital theories suggest that life expectancy be included in the economic growth/development model. The rationale behind this is that the health of workers affects productivity since they can work longer hours. This is particularly important for developing countries since they tend to have labor-intensive industries (Krieckhaus 2006). Government Expenditures Governments play an important role in economic performance on the grounds that governments make economic policy decisions including economic and financial policies. To measure the role of government, government expenditures are commonly used and this is the reason that neoclassical production function theorists included government expenditures in their economic growth model (Ram 1986). Thus, I employ government spending as a share of GDP for this variable. Inflation Rises in consumer prices lower the value of currencies; this results in lower domestic savings and private investment, leading to slower economic development. To estimate inflation, the natural logarithm of the annual percent change in consumer prices is employed (Barro 2001). Data and Method All the data used in this study are from Gerring et al. (2005). There are 224 countries included in the data set for 1950-2000, but due to missing values, data for 136 countries are used for analysis. In terms of research design, according to Rodrick and Wacziarg (2005), cross-national 22

research designs have been commonly employed in studying the relationship between democracy and economic development. However, a pooled time-series cross-national design is better for understanding the relationship between democracy and growth around the world because it incorporates time-varying dynamics in each country as well as across nations. According to Stimson (1985: 915), pooled time-series-cross-national modeling is an effective approach in studying phenomena that change over time and space, i.e. where one can only be understood with explicit reference to the other. The reason is that in order to capture the feedback relationship between economic development and democracy, we must be able to analyze the relationship between the two variables within nations over time across the world. However, violations of the classical linear regression assumptions of homoskedasticity and the lack of serial correlation of error terms are likely in pooled time-series cross-national analyses because, according to Stimson (1985: 919), panel data intrinsically have autocorrelation and heteroskedasticity (Burkhart and Lewis-Beck 1994). Another methodological concern for estimation comes from the simultaneity/endogeneity bias. There are two equations in our empirical model with two endogenous variables democracy stock and economic development. Due to the nature of the simultaneous relationship between these variables as discussed earlier, Ordinary Least Squares (OLS) estimators will be biased because of the correlation between coefficients and the error terms. To address these issues--the violation of classical linear regression assumption and simultaneity--i employ the instrumental variable technique and two-stage least squares for panel-data models. This results in Generalized Least Squares (GLS) estimators that eliminate simultaneity bias and generate unbiased estimators. For the statistical analysis, Intercooled STATA 10 is used. 23

FINDINGS In order to provide general information of the variables included in the analysis, I provide a summary of descriptive statistics is in Table 1. As you can see in Table 1, economic development variable (natural log of GDP per capita) shows relatively low variance/standard deviation, but democracy stock reveals a large variance with standard deviation of 246. The reason is that democracy stock is the sum of Polity 2 score (democracy level) since 1950. Before conducting an empirical analysis, I also tested for multicollinearity employing bivariate correlations of predetermined variables which are reported in Table 2. As Table 2 reveals, no variables show significant bivariate correlations except illiteracy rates and life expectancy (-0.78). However, both variables show statistically significant effects indicating that multicollinearity is not a serious problem. Moreover, according to Judge et al. (1985), multicollinearity becomes problematic only when bivariate correlations are 0.8 or above. 8 (Table 1 and 2 about here) The results of the 2SLS with instrumental variable technique using GLS random effects model for panel data are reported in Table 3. According to Table 3, economic development (per capita GDP) does have a positive and statistically significant effect on democracy stock. A one percent improvement in per capita GDP results in a 0.48 increase in democracy stock. 9 Considering that the mean of democracy stock variable is -48 while the maximum is 637 and the minimum is -604, the impact seems to be rather small. But, the positive and significant impact of economic development on democracy stock provides empirical evidence that as the level of economic prosperity improves, a transition to democracy is likely to occur and democracy is 8 Variance Inflation Factor (VIF) also indicates there is no serious multicollinearity issue. VIF test results are available from the author upon request. 9 For detailed information about the interpretation of log transformed variables in regression analysis, see Woolridge (2000). 24

likely to survive and mature. In other words, good economic performance is important for democratic institutions to survive and mature. Turning to the effects of other pre-determined variables, education and economic openness do not have a significant effect on democracy stock. This result implies that education and trade do not significantly help or deter democratic institutions to mature. This finding is surprising on the grounds that education and economic openness have been argued to play a significant role in democratization. This result might have occurred because we used to instrumental variable for the endogenous variables to avoid simultaneity bias. In contrast, years of independence and English legal origin show positive and significant effects on democracy stock as expected. In other words, the longer the country has been sovereign, the greater the likelihood of democracy survival and maturation. The same findings hold for countries with English legal origins. Social conflicts, as expected, have negative and significant effects on democracy stock. This result makes sense since social instability caused by conflicts is likely to make democracy difficult to survive and mature. Turning to the democracy stock equation, democracy stock has a positive and significant effect on economic development. Each increase in democracy stock score results in a 0.37 percent growth in per capita GDP. In other words, as democracies mature, the economy also develops. This result supports the finding of Gerring et al. (2005) and the argument of the compatibility perspective by finding that as democracies mature, the level of economic wealth also improves implying that maturity of democratic institutions enhances economic performance. Moreover, this result provides the empirical evidence of recursive relationship between economic development and democracy stock indicating that good economic performance helps 25

democracy to survive and mature, and as democratic institutions mature, the economy also performs better. In terms of the effects of pre-determined variables, the results are strong and consistent. Life expectancy as well as population growth shows positive and significant effects on economic development supporting the argument of human capital theories. Education also helps economic development as Barro (1990, 2001) argued. However, inflation, government expenditures, and social conflict do not have significant effects on economic development. CONCLUSION The debate on the relationship between economic development and democracy is one of the oldest subjects in theoretical and empirical studies in political science. Since Seymour Martin Lipset s (1959) seminal work on the effects of economic development on democracy, there have been many studies empirically testing the modernization perspective. In addition, there is a long tradition of debate on how democracy affects economic performance. The conflict school contends that democratic government systems are ineffective and therefore hamper economic development. In contrast, the compatibility perspective posits that since democracy protects individual property rights, democratic systems enjoy high levels of private investment and economic development. With the third and fourth wave of democratization (Huntington 1991; McFaul 2002), there is renewed interest in the relationship between democracy and economic development, particularly the effects of democracy on economic growth (e.g. Chua 2003; Gasiorowski and Poptani 2006; Gerring et al. 2005; Heo and Tan 2001; Krieckhaus 2004, 2006; Kurzman et al. 2002; McFaul 2002; Przeworski et al. 2000; Quinn and Woolley 2001; Rodrik Wacziarg 2005; 26