Democracy, Dictatorship and Technological Change

Transcription

1 Democracy, Dictatorship and Technological Change Carl Henrik Knutsen, Department of Political Science, University of Oslo 1 Introduction Does democracy enhance a country's rate of technological change, and thus technology-induced economic growth? 1 The literature on democracy's economic effects, both theoretical and empirical, has primarily focused on how democracy affects growth through physical capital accumulation (see e.g. Przeworski and Limongi, 1993; Tavares and Wacziarg, 2001) and human capital accumulation (see e.g. Baum and Lake 2003; Doucouliagos and Ulubasoglu 2008), but there are some exceptions: Using data from 1950 to 1990, Przeworski et al. (2000) find that democracies have a higher Total Factor Productivity (TFP) growth, though only among rich countries. Pinto and Timmons (2005) also investigated this relationship, but relied on (even) more problematic proxies of technological change than TFP such as foreign direct investment and trade. In the economic growth literature, however, technological change is generally considered the central determinant of long-run growth (see, e.g. Romer, 1990; Aghion and Howitt, 1992; Grossman and Helpman, 1991; Helpman, 2004). Therefore, an effect from democracy on technological change is potentially crucial for economic development. Historically, economists have emphasised input accumulation as the determinants of income and growth (see, e.g. Helpman, 2004, 9). Nonetheless, Abramowitz (1956), Solow (1957) and Denison (1962) found early evidence to indicate that technological change, as measured by TFP growth, contributed substantially to economic growth in the US, whereas Denison (1968) found the same pattern for European countries. Some empirical estimates indicate that differences in technological efficiency explain about 90% of global cross-country variation in income (Klenow and Rodriguez- 1 This chapter draws heavily on Chapter 5, Section 3 in Knutsen (2011b). 1

2 Clare, 1997), while other estimates indicate a larger role for human capital (Mankiw, Romer and Weil, 1992), although separating growth due to human and physical capital accumulation and technology is not straightforward. 2 Much of the literature on economic development in poorer countries has focused on capital accumulation (Easterly 2001), but the impact of idea flows on economic development can be immense. Ideas are non-rivalrous, and can therefore in principle be used to enhance efficiency in multiple locations at the same time (Romer, 1993). For this reason, technological change not only contributes to growth in rich countries at the "technological frontier", but also in developing countries (e.g. Easterly, 2001; Helpman, 2004) since poorer countries can adopt technological (and organisational) improvements developed elsewhere. Understanding why some countries are better at adopting techniques and diffusing them throughout their economies is therefore crucial for understanding differences in income level and growth rates. Technological change, as new growth economists (e.g. Romer, 1990) and evolutionary economists (e.g. Verspagen, 2005; Nelson, 2005) have recognised, is endogenous to institutional factors, among others. Growth theorists have mostly focused on economic institutional factors such as business structures patent rights and economic policies. However, economic institutions and policies are again endogenous to deeper political structures such as regime type (e.g. North, 1990; Rodrik, 2000). This paper argues that democracy enhances the dissemination of ideas and technologies into and within an economy, thereby affecting technological change. More specifically, this chapter discusses how self-interested dictators may restrict civil liberties for political survival purposes. In an imperfect world where dictators cannot fine-tune policies to block all politically dangerous information and allow all economically productive information, such restrictive policies 2 For example, new technologies often accompany new investments in machinery (Nelson, 2005), and human capital may enhance the spread of more efficient technologies (Kremer, 1993; Nelson and Phelps, 1966). 2

3 reduce technological change. The hypothesis that democracies have a higher technology-induced growth is tested on data from more than 140 countries, with some time series going back to the 19th century, and finds robust empirical support. 2 Why democracy may affect technological change 2.1 Some arguments Several arguments point to an effect of regime type on technological change. One classic argument in the regime type growth literature is that dictatorial governments, which are less constrained by their citizens' preferences than more accountable democratic governments, can force through unpopular, but growth-enhancing policies (see, e.g. Przeworski and Limongi, 1993). This argument has mainly been used for explaining high savings rates in some dictatorships. Still, the argument may also apply to large-scale, costly investments in R&D projects. The Soviet Union's history, as well as Nazi Germany's, indicates that dictatorial governments may amass enormous resources for large-scale R&D projects. The Soviet Union invested heavily in specific areas of science, such as mathematics and the natural sciences, and also invested large amounts of resources in projects of military and prestige value (Boesman, 1993; Balzer, 1993) such as its space program. These projects reduce present consumption for many citizens, though may increase long-run technological efficiency, at least in some sectors. However, there are likely to be several negative effects of dictatorship on technological change. First, democracies enhance property rights protection (see Knutsen, 2011a). As noted by North (1990), this is one mechanism through which dictatorship may reduce its investment in general, and its investment in innovative activities in particular. As highlighted by the new growth literature, property rights, including intellectual property rights, may have a positive effect on investment in 3

4 innovative activities (see, e.g. Romer, 1990). 3 Private actors may be wary of investing in activities which only yield potential future benefits if they fear the government will expropriate them in the future; dictatorial governments may have a hard time providing credible promises it will not expropriate (North and Weingast, 1989), and investments in innovative activities will suffer as a result. Secondly, the ability of various actors to efficiently utilise new technologies depends on the actors' and the overall economy's "absorptive capacity" (see, e.g. Abramowitz, 1986; Cohen and Levinthal, 1990). Absorptive capacity depends on a pre-existing R&D infrastructure, but crucially as well on the human capital level within a country. 4 Policies that enhance human capital thrive in democracies (see, e.g. Lake and Baum, 2001; Baum and Lake, 2003; Knutsen, 2011b). As a result, democracies may have a better "infrastructure" for adopting, adapting and efficiently utilising new technologies than dictatorships. A third possible, but undocumented, positive effect of democracy on technological change is related to democracy's impact on individual psychological characteristics. 5 There may be negative effects on human cognitive capabilities from living in a conformist, hierarchical society, in which debate and free thinking is discouraged. This hypothesis is best described by Mill (1974), who considered the stifling of debate and restrictions on freedom of speech to have far-reaching consequences: "the price paid for such intellectual pacification is the sacrifice of the entire moral courage of human mind" (Mill, 1974, 74). Restrictions on freedom of speech lead to an environment in which conform behaviour dominates, and in which new thoughts and alternative ways of doing things and experimentation suffer, thus reducing technological dynamism. 3 However, as one of Archibuggi and Filippeti's chapters in this volume notes, the impact of intellectual property rights on innovation and technology diffusion may be overrated by many analysts. 4 See Hübler's chapter in this volume for a review. 5 Psychological-experimental studies could investigate this mechanism, for example by studying the creativity of individuals in different environments. 4

5 2.2 Regime types, civil liberties, information flows and technological change As seen above, most arguments point to a positive effect on technological change from having a democracy. There is an additional argument which points towards a democratic technological advantage. This argument is discussed below and presented formally in Knutsen (20116). To sum up, the main conclusion is that dictators have incentives for restricting information flows, thereby crippling the diffusion of technology and economic growth. Dictators restrict information flows by curbing civil liberties, with the reason being that it increases his probability of staying in power. There are different reasons for why dictators may want rapid technological change in their country's economy. First, dictators can use their political power to acquire a substantial portion of their country s economic output (e.g. North, 1990; Olson, 1993; Robinson, 1998; Knutsen, 2011b), and appropriable resources will likely increase in absolute terms when the economy grows. These resources may subsequently be used for future personal consumption, which dictators, as most people, likely value. Second, extra resources due to a more efficient economy may be used for paying off necessary supporters in order to retain the dictator's political position. Third, technological change and increased production are likely to enable a well-funded, modern and strong military. This may help ward off external security threats in the form of foreign armies. Consequently, rapid technological change, which again increases GDP, has effects that a dictator and his supporters will benefit from. Nevertheless, self-interested dictators may want to mitigate economic growth, particularly in relation to restricting information flows that enhance technological change. Because of preferences for private consumption or political survival, dictators may have incentives to take actions that have negative consequences for their national economies (e.g. Wintrobe, 1990; Olson, 1993; Robinson, 2001; Bueno de Mesquita et al., 2003; Acemoglu and Robinson, 2006a). In Olson's (1993) model, dictators, especially those with a short time horizon, expropriate property to maximise personal 5

6 consumption, thereby reducing the incentives for citizens to work or invest. In Robinson's (2001) and Acemoglu and Robinson's (2006a) models, public investment and economic development more generally strengthen opposition groups and reduce leaders survival probability, all of which reduce a dictator's expected future consumption. Bueno de Mesquita et al.'s (2003) model assumes that leaders are interested in surviving in office, and shows how dictators, particularly those with small winning coalitions and large selectorates, under-invest in growth-conducive public goods. For these leaders, it is rational to expropriate or heavily tax and redistribute resources as private goods to their core supporters. In Wintrobe (1990), certain power-motivated dictators overinvest in a repressive capacity. Investment in a repressive apparatus takes public resources away from more productive projects (see also Acemoglu and Robinson, 2006b). For various reasons, self-interested dictators may have incentives to conduct policies that hurt the overall economy. The particular mechanism presented here is related to how policies which affect civil liberties, and thus information flows and technological change, may have important negative impacts on the survival probability of dictators and their regimes. One way to model the relationship between a dictator's survival probability and information flow would be to assume an opposition consisting of several individuals, all desiring to overthrow the dictator, so the probability of the opposition overthrowing the dictator depends on their coordination. If one individual attempts to overthrow the dictator, the probability of success is small. But, as collective action problems are solved and opposition members coordinate, the opposition's success-probability increases. The opposition's ability to coordinate depends on their ability to use communication tools, assemble without harassment or detention, gain access to the media and travel freely within the country. Therefore, restrictions on civil liberties, which reduce politically relevant information flows, reduce the opposition's ability to coordinate, thereby increasing the dictator's survival probability. Moreover, information from the outside world may carry "dangerous" ideological content, e.g. democratic norms and values, or they may undermine governmental propaganda by presenting alternative or even opposite viewpoints to the governments on everything from economic policies to the framing 6

7 of particular events such as environmental disasters. Because of this, there are many reasons for self-interested, political survival-oriented dictators to curb civil liberties and the information flow. In practice, dictatorial regimes are unable to implement policy such that only politically dangerous information is banned; economically relevant information, for example, ideas on economic policies, foreign production or organisational technologies may also be affected by such restrictive policies. Generally speaking, it is difficult for dictatorial governments to screen every act of communication, travel and meeting, so governments must establish some general rules. As a result, information activities are banned due to the uncertainty of their content. General restrictions on civil liberties will more than likely not only reduce political communication, but economically relevant communication as well that affects the speed and extent of technology diffusion. 6 A broad diversity of ideas generally improves economic efficiency: A dynamic economy "is the outcome of a constant interaction between variety and selection" (Verspagen, 2005, 496). Selection reduces variety since more efficient techniques are adopted through learning and drive out older, inefficient production methods. In order to maintain variety, an economy needs to have a continuous introduction of novel ideas. Civil liberties such as freedom of speech, press and travel allow for improved idea diffusion both into and within the economy. Civil liberties also allow for a comparison of different ideas, which allows for a selection of the most efficient. With power concentration and limits to freedom of speech and press, leaders may suppress efficiency-enhancing ideas because of a limited amount of knowledge or self-interest. A variety and selection of ideas are not only important in the marketplace, but also in terms of economic policy making. As noted by North (1990) and Greif (1993), economic efficiency improvements are strongly affected by changes in policies, institutions and organisations. North (2005) argues that the inherent uncertainty about policy- and organizational effects necessitates a 6 Generation of cutting-edge technology in increasing-returns-to-scale sectors is mostly relevant for large and rich developed countries. However, for most countries the global technological frontier is predominantly exogenous. Thus, diffusion of international technology is key for efficiency. However, the diffusion-innovation distinction is only a crude, but helpful analytical tool. In practice, it may be difficult to separate between innovation and diffusion. Much innovation is incremental, and what some consider innovation may involve the adaptation and slight modification of already existing ideas. 7

8 process of trial and error, with proper feedback from society. Open systems associated with democratic government and civil liberties are crucial for informationflows that allow such efficiency enhancing trial and error and feedback processes. Restrictions on freedom of speech and media hurt efficiency, as important problems are not reported and alternative viewpoints on economic policies and organisational issues are not forthcoming to the political elite or possibly even the bureaucracy. The "dictator's dilemma" is also a relevant insight (Mueller, 2003, ). Because of a fear of falling out with the dictator or others in the regime, individuals and organisations might not be forthcoming with their most accurate information, resulting in a reduction in the quality of information the regime draws upon when making policy decisions. Moreover, freedom of speech contributes to actors "assessing and disseminating ideas from abroad, discourages insular thinking and stimulates vigorous debate" (Halperin, Siegle and Weinstein, 2005, 13). As von Hayek (1944, 1945) argues, different actors possess only partial knowledge about economic and political processes. Open debate and free idea flows are crucial for efficient decision-making by firms, bureaucrats and politicians, as "the knowledge of the circumstances of which we must make use never exists in concentrated or integrated form, but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess" (von Hayek, 1945, 519). Additionally, North, Wallis and Weingast (2009) argue that open competition in both economic and political spheres are crucial for adaptive efficiency: "Open access and free flow of ideas generate a range of potential ways to understand and resolve new problems... The free and open expression of ideas means that many ideas will be heard. In their quest to maintain or to regain power, competing parties will draw on this competition for solution" (North, Wallis and Weingast, 2009, ). A dictator interested in political survival will crack down on all information flows which are politically dangerous for him, but that he can identify as irrelevant for economic efficiency. Furthermore, the dictator will likely be open to information identified as harmless to his survival probability, but which improves economic efficiency, at least if this increases the resources 8

9 available for personal consumption. This interesting trade-off relates to policies with effects on both political survival and economic efficiency, and types of information that the dictator cannot assign as either being of purely "political" or "economic" relevance. On the one hand, the dictator wants to allow such information because it increases efficiency and private consumption growth, while on the other, he wants to restrict it since it puts his political survival at risk. The optimal policy from a dictator's point of view involves a balancing act between these two concerns: the dictator will balance an increase in marginal utility from consumption against the expected marginal utility decrease from a reduced survival probability when setting policy. If more, or perhaps all, such "mixed" or "indeterminable" information flows are allowed in democracies, democracies will experience more rapid technological change than dictatorships. A country where very little information is allowed, North Korea being a suitable empirical example (see, e.g. Kihl and Kim 2006), will have very low long-run growth rates according to this argument. Good examples of the above argument are related to modern communications technologies. Hachigian recognises the dictator's conundrum: "The Internet presents a dilemma to leaders of authoritarian states and illiberal democracies. It promises enticing commercial advantages, such as transaction cost reductions, e-commerce possibilities, and foreign trade facilitation. Yet, by giving citizens access to outside information and platforms for discussion and organization, the Internet can also help politically empower populations and potentially threaten regimes" (Hachigian, 2002, 41). Cell phone technology also presents both political problems and potential economic gains to dictators. Bans on cell phones have been imposed in Cuba and Turkmenistan, as phones are dangerous coordination devices for political opposition groups. However, restrictions on phone communication also cripple the efficiency of businesses, while the freedom to travel may also be 9

10 restricted because of political reasons. International travel and even travel into Pyongyang is restricted for North Koreans. Although strict regulation of national and international travel may enhance political survival, this severely affects North Koreans' ability to learn new and productive foreign technologies. A more curious example comes from the Soviet Union, where the Politbureau banned foreign economics journals seen to be spreading potentially dangerous ideas in opposition to the governing ideology (Greenspan, 2007). However, econometric journals were allowed, probably because they were seen as being less politically dangerous. In more general terms, historical experiences, especially from the 1970s onward, demonstrated that central planning was statically and dynamically inefficient. A more open political system than that of the Soviets would perhaps have experienced intense debates within the elites and population on how to pursue efficiencyenhancing policy reform. Ironically, when Gorbachev at last opened up the political and economic sphere, it arguably contributed to the regime's downfall some years later, which illustrates a key point: allowing information flows in a dictatorship is dangerous for the regime s survival. When combined with some other assumptions, the above argument yields several empirical implications in addition to "democracies produce more rapid technological change". 7 For instance, dictators may be expected to loosen restrictions on civil liberties in times of rapid technological change globally, thus absorbing more new global ideas: If the global technology frontier's growth rate is high, relaxing restrictions on civil liberties will yield a higher consumption gain for the dictator in exchange for a given decrease in survival probability. A dictator with a safe grip on power because of exogenous reasons can also allow more information to flow into his country in order to boost the national economy, and therefore his personal consumption. In addition, civil liberties restrictions depend on whether the dictator expects to consume resources earned while in 7 See Knutsen (2011b) for how these additional empirical implications are deduced. 10

11 power, or after he is dethroned. If a dictator's resources are immobile (see Boix, 2003) and cannot be enjoyed in exile, the dictator would have extra incentives to curb civil liberties because the expected value of consumption for a given set of policies is reduced. Moreover, a dictator's motivation also has implications for the preferred policy on civil liberties (see also Wintrobe, 1998) and the rate of technological change. Certain rulers are interested mostly in consumption and such rulers will also constrain civil liberties, as there is an indirect negative effect from lenient civil liberties policies on expected future consumption through an increased probability of losing office (Robinson, 2001; Acemoglu and Robinson, 2006a). Even so, they will do so to a lesser degree than rulers who are also inherently motivated to stay in office. Finally, some dictatorships may be better able than others to mitigate information-related problems. Even with restrictions on civil liberties, dictatorships with a high institutional capacity and bureaucratic quality may perform adequately because of better abilities in interpreting weak informational signals and for designing policies which enable the separation of politically and economically relevant information. This hypothesis is tested empirically in Knutsen (20116), though it only finds partial support. 3 Data Measuring technological change is difficult (see, e.g. Nelson, 2005). We focus here on technological change-induced economic growth, with one common measure for this being TFP growth. TFP is calculated as a residual when economic growth stemming from changes in factor inputs such as physical capital, human capital and labour are subtracted from total growth (see, e.g. Barro and Sala-i Martin, 2004; Baier, Dwyer and Tamura, 2002). I utilise the extensive TFP data from Baier, Dwyer and Tamura (2006). These data cover 145 countries and have a very extensive time series, with 24 countries having a time series of more than 100 years. The TFP data are estimated with 11

12 uneven intervals, averaging approximately a data point every 10th year. 8 I interpolate these series, assuming constant TFP growth rates within certain periods, and obtain estimates of annual TFP growth. There are several problems with TFP as a measure of technology-induced growth. First, TFP responds to short-term economic shocks, for example, the underutilisation of capacity in a recession or profit windfalls from oil-price hikes, yet the long-time series used here mitigates this problem. Second, TFP is a theoretical construct based on a particular macro production function, and is therefore reliant on the assumptions of a chosen model (see Rodrik, 1997). Furthermore, TFP growth may be systematically underestimated, as an investment and work effort will probably increase when the level of technology increases (see, e.g. Verspagen, 2005). If so, technological change is a cause of input accumulation, although the growth will be assigned to input accumulation. Nevertheless, there is no good reason why such biases should critically influence the relationship between democracy and TFP. TFP data are also troubled by unsystematic measurement errors, especially for the older data, in which GDP, capital stocks and human capital stock estimates are based on thin data material (see, e.g. Maddison, 2006; Baier, Dwyer and Tamura, 2006). If these errors are unsystematic, they do not bias coefficients, but should only increase the regression's standard errors. Therefore, the poor quality of the older TFP data does not cast serious doubts on the validity of the findings below. A tricky issue related to measuring technological change's impact on economic growth is the time lag (see, e.g. Verspagen, 2005; Crafts, 2003). When it comes to economic growth benefits from General Purpose Technologies (GPTs), Crafts argues that "[t]he lag before a GPT has its full effect on productivity is measured in decades not years." (Crafts, 2003, 19). There is also a time lag from the effect of political institutions and policies on innovation and idea diffusion: First, there may be lags tied to the effect from institutional structures on diffusion of ideas into the nation. Second, it 8 The data are based on multiple sources. TFP is calculated using income per worker rather than per person, and by assuming Hicks-neutral technology and a capital share of 1/3 (see Baier, Dwyer and Tamura 2002, 2006). 12

13 takes time before ideas diffuse widely within the national economy. Using patent citation data, Caballero and Jaffe (1993) estimate the diffusion lag of ideas between US firms to be between one to two years on average, whereas "Mansfield, Schwartz, and Wagner (1981) found that 60 percent of the patented innovations they studied were imitated within 4 years" (Segerstrom, 1991, 808). Third, it takes time before economic actors can efficiently utilise new ideas in actual production. The total time lag is therefore hard to guess, and I use two different models that operate with 5- and 10-year lags. For the political regime variable, one relevant operationalisation is the "Civil Liberties" (CL) index by Freedom House, which explicitly taps protection of civil liberties such as freedom of speech, media and assembly. 9 A 7 is given to countries with the worst protection of civil liberties, and a 1 to those with the strongest protection. However, the CL only has data which goes back to For that reason, I mainly rely on the Polity index (PI), which measures the degree of democracy operationalised according to participation and competition in elections and checks on the executive (Marshall and Jaggers, 2002). The PI goes from -10 (least democratic) to 10 (most democratic). Although the PI does not explicitly capture civil liberties, the correlation between the PI and CL ( ) was The PI has data which goes all the way back to Technological change is not only a function of regime type. Consequently, several other factors need to be controlled for. 10 The pre-existing level of (log) TFP is controlled for, as it is likely important to TFP growth; Idea gaps (Romer 1993) may create convergence effects analogous to those for capital stocks (Barro and Sala-i Martin, 2004). Further, the level of population may impact on technological change (Romer, 1990). Technological diffusion and creation may also be hampered when there is political instability. I therefore control for the log of regime duration with data from Polity IV. Geographic factors may also affect TFP growth (Hall and Jones, 1999), so I 9 One problem with using CL is that dictators adjust civil liberties partly as a response to, for example, global technology growth, which makes CL endogenous. 10 Only variables that also correlate with regime type may generate omitted variable bias if left out. See Chapter 4 in Knutsen (2011b) for a discussion of how these variables are related to democracy. 13

14 control for absolute latitude. Since economic openness also may impact positively on idea diffusion, I control for the Frankel-Romer trade instrument, which together with the latitude data are taken from Hall and Jones (1999). I also control for the ethnic fractionalisation index from Alesina et al. (2003). Easterly and Levine (1997) find that ethnic fractionalisation has an important effect on economic growth, and the difficulties of solving collective action problems, as well as social and lingual barriers, may reduce technology diffusion within fractionalised societies. I also add region and decade dummies to control for geographic- and time-specific effects. The decade dummies are particularly important because the global technology frontier's growth has varied quite much historically. The Fixed Effects analysis controls for country-specific effects, thus only investigating the effect from regime type on TFP growth based on within-nation variation. This further reduces the probability of omitted variable bias. 4 Empirical results 4.1 Main analysis I first run OLS with PCSE (see Beck and Katz, 1995), taking into account heterogeneous standard errors and contemporaneous correlation between panels and AR(1) autocorrelation within panels. The first models utilise the interpolated data. Table 1 shows the OLS with PCSE results: The PIL5 and PIL10 models show estimates from models using the PI, with some time series going back to the 19th century, with 5- and 10-year lags, respectively, on all independent variables. The PIS models are restricted to the time period from 1972, as are the CLS models, although these use CL as a regime measure. The results in Table 1 reveal a very strong support for the hypothesis that dictatorship reduces technology-induced economic growth: In all models, the regime coefficient has the expected sign. The models based on a long time series yield the strongest results, with both models showing significant positive effects from democracy (at least at the 5% level). Also, the models with a shorter time series yield a significant, positive effect from democracy, independent of the regime 14

15 measure when the 10-year lag is used. However, when using the 5-year lag, the coefficients are nonsignificant at conventional levels. The estimated effects are quite substantial: With the exception of one estimate, the models' estimates indicate an effect Table 1: OLS with PCSE models on interpolated data; TFP growth as dependent variable; Time dummies and constant omitted from table Model PIL5 PIL10 PIS5 PIS10 CLS5 CLS10 Variable b/(t) b/(t) b/(t) b/(t) b/(t) b/(t) Polity 0.036*** 0.022** ** (4.30) (2.34) (1.57) (2.15) FHICL *** (-0.68) (-2.99) Ln TFP *** *** *** *** ** *** (-5.22) (-5.03) (-2.81) (-3.24) (-2.56) (-3.46) Ln population *** ** *** ** (-3.65) (-2.37) (-2.69) (-1.42) (-2.35) (-1.37) Ln regime dur ** * (-0.15) (-2.09) (0.25) (-1.00) (-0.10) (-1.72) Ethn. fraction ** ** *** ** *** * (-2.18) (-2.08) (-3.14) (-1.97) (-3.33) (-1.92) Ln Fr.-Romer *** ** ** * ** (-3.09) (-1.20) (-2.23) (-2.26) (-1.93) (-2.36) Latitude (1.28) (0.68) (1.43) (-0.36) (1.25) (-0.31) E.Eur.+Sov.rep * * *** *** (-1.79) (-1.92) (-1.44) (-5.84) (-1.44) (-6.05) Africa SS *** *** ** *** ** *** (-4.83) (-4.61) (-2.45) (-3.38) (-2.44) (-3.37) Asia-Pacific *** *** * *** * *** (-3.43) (-2.67) (-1.93) (-3.39) (-1.78) (-3.34) MENA *** * *** (0.73) (-0.77) (-0.79) (-5.69) (-1.82) (-6.83) Latin America ** * * *** * *** (-2.35) (-1.92) (-1.94) (-5.22) (-1.84) (-4.85) N

16 of going from most dictatorial to most democratic between 0.4 to 0.7% extra annual TFP growth. If democracies also grow more because they enhance human capital accumulation (see, e.g. Tavares and Wacziarg, 2001; Baum and Lake, 2003; Doucouliagos and Ulubasoglu, 2008), democracies have a quite substantial growth advantage over dictatorships because of knowledge-related factors. Still, there may be country-specific characteristics biasing the above results. I therefore run Random Effects (RE) and Fixed Effects (FE) to check the robustness of the results. Indeed, the RE results shown in Table 2 indicate an even stronger effect from democracy. All regime coefficients, except in one model (CL with 5-year lag), are statistically significant at the 1% level, with absolute t-values from 3.15 to Moreover, the coefficients are larger than in Table 1, indicating that changing from most dictatorial to most democratic increases TFP growth with between 0.8 to 1.7% per year. Also, the results from the FE models given in Table 3 are very strong. Four out of six FE models find a significant effect from democracy on TFP growth at the 1% level, while one model finds a coefficient almost significant at the 5% level. The estimates are quite large, indicating a TFP effect from "full democratisation" of approximately 0.6 to 1.2% extra annual growth. 4.2 Robustness checks The results above are relatively robust when also controlling for historical colonising country and plurality religion dummies from Knutsen (2007). 11 Additionally, the results are relatively, though not completely, robust to using different lags. For example, the results generally hold for three-, but not two-year lags. Furthermore, the above argument indicates that dictators' survival probability, and thus the duration of the regime, is endogenous to technologically relevant policies. For that reason, I left out the log regime duration variable from the regressions, but the strong results also survived in this model specification. 11 Tables with results from various robustness checks are reported in Knutsen (2011b). 16

17 Table 2: Random Effects (robust standard errors) results; TFP-growth as dependent variable; Time dummies and constant omitted from table Model PIL5 PIL10 PIS5 PIS10 CLS5 CLS10 Variable b/(t) b/(t) b/(t) b/(t) b/(t) b/(t) Polity 0.040*** 0.040*** 0.044*** 0.083*** (6.26) (6.48) (3.15) (5.73) FHICL (-1.18) *** (-3.84) Ln TFP *** *** *** *** *** *** (-18.25) (-23.40) (-6.98) (-6.38) (-7.02) (-6.70) Ln population *** *** *** ** *** ** (-14.17) (-14.03) (-4.86) (-2.47) (-4.38) (-2.22) Ln regime dur *** *** *** *** (-4.06) (-7.21) (-1.02) (-3.01) (-1.11) (-3.90) Ethn. fraction ** *** *** *** *** (-2.24) (-1.25) (-3.73) (-3.08) (-3.74) (-3.10) Ln Fr.-Romer *** *** *** ** *** ** (-9.50) (-8.30) (-4.19) (-2.30) (-3.74) (-2.03) Latitude 0.024*** 0.025** (3.06) (2.33) (1.58) (0.57) (1.30) (0.38) E.Eur.+Sov.rep * ** * ** ** (-1.78) (-2.00) (-1.92) (-1.63) (-2.29) (-2.00) Africa SS *** *** *** *** *** *** (-6.02) (-6.16) (-3.32) (-3.13) (-3.94) (-3.80) Asia-Pacific *** *** ** *** *** *** (-3.85) (-4.11) (-2.39) (-2.99) (-2.92) (-3.56) MENA * (-0.30) (0.23) (-1.00) (-0.23) (-1.75) (-0.83) Latin America *** *** *** *** *** *** (-5.30) (-4.68) (-5.02) (-4.47) (-5.52) (-5.17) N

18 Table 3: Fixed Effects (robust standard errors); TFP-growth as dependent variable; Time dummies and constant omitted from table Model PIL5 PIL10 PIS5 PIS10 CLS5 CLS10 Variable b/(t) b/(t) b/(t) b/(t) b/(t) b/(t) Polity 0.036*** 0.037*** 0.030* 0.074*** (5.46) (5.93) (1.92) (4.65) FHICL *** (-0.23) (-2.98) Ln TFP *** *** *** *** *** *** (-20.86) (-27.55) (-8.32) (-7.31) (-8.43) (-7.79) Ln population *** *** *** *** *** *** (-17.88) (-17.13) (-8.36) (-3.54) (-7.91) (-3.41) Ln regime dur *** *** *** *** (-4.51) (-7.42) (-0.80) (-2.60) (-0.58) (-3.38) N The interpolation conducted on the sample above may be problematic because it expands the number of data points and introduces additional measurement error. Because of this, I calculate the average annual TFP growth rates for the periods between time points where Baier, Dwyer and Tamura (2006) provide TFP estimates. Some periods are as short as four years, and I include periods up to 20 years. However, 581 out of 795 periods are 10 years long, and an additional 82 are either nine or 11 years long. Only 17 are above 15 years, with one period counting as one observation. For the control variables, I use the values at the start of the time period. For political regime, I average the PI over the five years prior to the period and all years within the period, except for the five latest years, in order to take into account the effect's time lag. 12 These results also show that democracy enhances TFP growth. The effect from democracy is almost significant at the 5% level in the FE model, with a t-value of 1.96, and it is significant at the 1%-level in the OLS with PCSE and RE models. The point estimates are about equal to those obtained above, although a bit higher: According to these estimates, a change from a brutal dictatorship to a full democracy increases annual TFP growth between 1.0 and 1.4%. 12 There are too few observations on CL to run any meaningful panel analysis. 18

19 However, TFP growth is criticised as a problematic measure of technology-induced growth, and even more so as a measure of technological change. I therefore investigated whether democracy affects the ArCo index, which is supposed to capture technological capabilities (Archibugi and Coco, 2004). This index consists of three components; a technology creation index, a technology infrastructure index and a human skills index. 13 The index only has values for 1990 and 2000, and I ran OLS regressions, using the values from I tested three different versions of the democracy measures: one simply using the 2000 value, one using the average score for the 1990s and one using the average score from the 1980s and 1990s. I controlled for the various variables in the model above, substituting the TFP level with log GDP per capita (PPP) from Maddison (2006). These results indicate that democracy enhances technological capabilities: When using the CL as a democracy indicator, the aggregate measure and three components of the ArCo document a robust positive effect from democracy. The effect is not robust when using Polity, although the point estimate is always positive. However, the ArCo-index is restricted to between 0 and 1. We take this into account by running Tobit regressions (with 0 and 1 as censoring values, using 16,000 iterations) rather than OLS. This further strengthens the results, as all CL models now show an effect significant at the 1% level, and both PI models using historical averages show a significant effect at the 10% level, whereas the last model has a p-value of Conclusion This paper focused on how democracy and dictatorship affect the most important determinant of long-run economic growth, that of technological change. In dictatorships, the diffusion of economically relevant ideas and technologies is slowed down because dictators manipulate civil liberties and promote policies which inhibit idea diffusion. In an optimal world, dictators would want to promote technological change to increase their own personal consumption, yet in practice 13 See Archibugi and Coco (2004), and also Archibugi and Coco (2005) for methodological discussions. 19

20 they have to trade increased growth against the increased probability of being thrown out of office. Dictators are unable to perfectly separate politically dangerous from economic efficiency-enhancing information when setting their policies. Based on very extensive data material, the empirical results corroborated this hypothesis. Democracies have higher TFP growth rates, thus indicates a more rapid technological change. This means that in the long run, democracies should prosper more than dictatorships. One important policy implication for IGOs, NGOs and aid donor countries dealing with economic development and democratisation issues in developing countries can be derived from this chapter. One traditional argument against promoting democratisation in developing countries is that democracy may hamper economic development; one should postpone the "luxury good" of democracy until an economy is sufficiently developed (see Knutsen, 2011c). This chapter indicates that the proposed economic benefit of dictatorship is exaggerated. Technological change and longrun prosperity are best promoted by a democratic regime, and the premise that democracy will more or less automatically emerge after a period of development is questionable (see, e.g. Przeworski and Limongi, 1997; Acemoglu et al., 2008). Dictators sometimes claim their country will one day democratise, but that it is not yet ready. However, as indicated above, dictators will fight hard and apply various measures to stay in power. Promoting democratisation in developing countries is difficult, but there is no reason for expecting dictatorships to produce prosperity and dictators to subsequently give up power voluntarily. 20

21 References Abramowitz, M "Resource and Output Trends in the United States since 1870." American Economic Review 46:5-23. Abramowitz, M "Catching-up, Forging Ahead and Falling Behind." Journal of Economic History 46(2): Acemoglu, D. and J.A. Robinson. 2006a. "Economic Backwardness in Political Perspective." American Political Science Review 100: Acemoglu, D. and J.A. Robinson. 2006b. Economic Origins of Dictatorship and Democracy. New York: Cambridge University Press. Acemoglu, D., S. Johnson, J.A. Robinson and P. Yared "Income and Democracy." American Economic Review 98(3): Aghion, P. and P. Howitt "A Model of Growth through Creative Destruction." Econometrica 60: Alesina, A., A. Devleeschauwer, W. Easterly, S. Kurlat and R. Wacziarg "Fractionalization." Journal of Economic Growth 8(June): Archibugi, D. and A. Coco "A new indicator of technological capabilities for developed and developing countries (ArCo)." World Development 32(4): Archibugi, D. and A. Coco "Measuring technological capabilities at the country level: A survey and a menu for choice." Research Policy 34(2): Baier, S.L., G.P. Dwyer and R. Tamura "How Important Are Capital and Total Factor Productivity for Economic Growth?" Federal Reserve Bank of Atlanta. Working Paper a. Baier, S.L., G.P. Dwyer and R. Tamura "How Important are Capital and Total Factor Productivity for Economic Growth?" Economic Inquiry 44(1): Balzer, H Science, Technology and Education in the former U.S.S.R. In R.F. Kaufman and J.P. Hardt (eds.) The Former Soviet Union in Transition. Armonk: M.E. Sharpe pp

22 Barro, R.J. and X. Sala-i Martin Economic Growth. Cambridge, MA: The MIT Press. Baum, M.A. and D.A. Lake "The Political Economy of Growth: Democracy and Human Capital." American Journal of Political Science 47(2): Beck, N. and J.N. Katz "What to Do (and Not to Do) with Time-Series Cross-Section Data." American Political Science Review 89(3): Boesman, W Science and Technology in the Former Soviet Union: Capabilities and Needs. In R.F. Kaufman and J.P. Hardt (eds.) The Former Soviet Union in Transition. Armonk: M.E. Sharpe pp Boix, C Democracy and Redistribution. Cambridge: Cambridge University Press. Bueno de Mesquita, B., A. Smith, R.M. Siverson and J.D. Morrow The Logic of Political Survival. Cambridge, Massachusetts: MIT Press. Caballero, R.J. and A.B. Jaffe "How High are the Giants' Shoulders: An Empirical Assessment of Knowledge Spillovers and Creative Destruction in a Model of Economic Growth." NBER Working Paper NO Cohen, W.M. and D.A. Levinthal "Absorptive capacity: A new perspective on learning and innovation." Administrative Science Quarterly 35(1): Crafts, N.F.R "Quantifying the Contribution of Technological Change to Economic Growth in Different Eras: A Review of the Evidence." London: Department of Economics, LSE. Working Paper NO. 79/03. Denison, E.F The Sources of Economic Growth in the US and the Alternatives before Us. New York: Committee For Economic Development. Denison, E.F Why Growth Rates Differ: Postwar Experiences in Nine Western Countries. Washington DC: Brookings Institution. Doucouliagos, H. and M.A. Ulubasoglu "Democracy and Economic Growth: A Meta- Analysis." American Journal of Political Science 52(1):

23 Easterly, W The Elusive Quest for Growth: Economists' Adventures and Misadventures in the Tropics. Cambridge, MA: MIT Press. Easterly, W. and R. Levine "Africa's Growth Tragedy: Policies and Ethnic Divisions." The Quarterly Journal of Economics 112(4): Greenspan, A The Age of Turbulence: Adventures in a New World. Penguin Press. Greif, A "Contract Enforceability and Economic Institutions in Early Trade." American Economic Review 83: Grossman, G.M. and E. Helpman "Quality Ladders in the Theory of Growth." The Review of Economic Studies 58(1): Hachigian, N "The Internet and Power in One-Party East Asian States." The Washington Quarterly 25: Hall, R. and C. Jones "Why do some countries produce so much more output per worker than others." The Quarterly Journal of Economics 114(1): Halperin, M.H., J.T. Siegle and M.M. Weinstein The Democracy Advantage: How Democracies Promote Prosperity and Peace. New York: Routledge. Helpman, E The Mystery of Economic Growth. Cambridge, MA: The Belknapp Press of Harvard University Press. Kihl, Y.W. and H.N. Kim, eds North Korea: The Politics of Regime Survival. Armonk: M.E. Sharpe. Klenow, P.J. and A. Rodriguez-Clare The Neo-Classical Revival in Growth Economics: Has it Gone too Far? In NBER Macroeconomics Annual 1997, eds. Ben S. Bernanke and Julio J. Rotemberg. Cambridge, MA: MIT Press. Knutsen, C.H Democracy and property rights. Master's thesis. Oslo: University of Oslo, Department of Economics. Knutsen, C.H. 2011a. "Democracy, Dictatorship and Protection of Property Rights." Journal of Development Studies: Forthcoming. 23

24 Knutsen, C.H. 2011b. The Economic Effects of Democracy and Dictatorship. PhD thesis. Oslo: Department of Political Science, University of Oslo: Forthcoming. Knutsen, C.H. 2011c. "Investigating the Lee-Thesis: How Bad is Democracy for Asian Economies." European Political Science Review: Forthcoming. Kremer, M "The O-Ring Theory of Economic Development." The Quarterly Journal of Economics 108(3): Lake, D.A. and M.A. Baum "The Invisible Hand of Democracy: Political Control and the Provision of Public Services." Comparative Political Studies 34(6): Maddison, A The World Economy. Paris: OECD Publishing. Mankiw, G.N., D. Romer and D.N. Weil "A Contribution to the Empirics of Economic Growth." The Quarterly Journal of Economics 107(2): Marshall, M.G. and K. Jaggers "Polity IV Project - Dataset Users' Manual." College Park: Program Center for International Development and Conflict Management, University of Maryland. Manual. Mill, J.S On Liberty. London: Penguin. Mueller, D.C Public Choice III. Cambridge: Cambridge University Press. Nelson, R.R Technology, Institutions and Economic Growth. Cambridge, MA: Harvard University Press. Nelson, R.R. and E.S. Phelps "Investment in Humans, Technological Diffusion, and Economic Growth." The American Economic Review 56(1-2): North, D.C Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press. North, D.C Understanding the Process of Economic Change. Princeton: Princeton University Press. North, D.C. and B.R. Weingast "Constitutions and Commitment: The Institutions Governing Public Choice in Seventeenth Century England." Journal of Economic History 44(4):