Introduction: The Lee thesis
[M]any people in different countries of the world are systematically denied political liberty and basic civil rights. It is sometimes claimed that the denial of these rights helps to stimulate economic growth and is “good” for economic development. Some have even championed harsher political systems – with denial of basic civil and political rights – for their alleged advantage in promoting economic development. This thesis (often called “the Lee thesis” …) is sometimes backed by some fairly rudimentary empirical evidence (Sen, Reference Sen1999: 15).
Is dictatorship good for economic growth in developing countries, and particularly in Asia? Many believe so, and the last three decades’ incredible economic growth in authoritarian China is often invoked as exhibit number one. Conversely, several non-democratic leaders, academics, and journalists have argued that democracy is bad for economic growth in poor countries. The notion that democracy hinders economic growth, and wider development, is often dubbed the Lee thesis, after former Singaporean Prime Minister Lee Kuan Yew’s defense of this hypothesis. The underlying argument is that democracy, with its political rights, civil liberties, and political contestation, generates social instability or political gridlock, and democratic governments are therefore unable to generate rapid economic development. A strong, authoritarian leadership is deemed necessary to promote reforms, conduct efficient policies that are unaffected by special interest groups and generate high investment levels.Footnote 1
This paper investigates the relevance of the Lee thesis in the geographical region where it is arguably most likely to hold, Asia. Democracy is defined here as a political regime with a high degree of popular control over political decision making, and a high degree of political equality (Beetham, Reference Beetham1999). Both ‘dictatorship’ and ‘authoritarian regime’ refer to regimes that are not democracies, although some analysts distinguish between these concepts (Linz and Stepan, Reference Linz and Stepan1996). This paper thus investigates the general effect, in Asia, of popular control over politics, which is related to the existence of free and fair elections, checks on political elites’ power, political rights, and civil liberties. The second section presents arguments for why dictatorship might be particularly conducive to economic growth in Asia. The third section introduces general theoretical arguments indicating that dictatorships are conducive to economic growth, and some counterarguments. The fourth section discusses methodological problems related to inference and generalization from some Asian developmental regimes. The last section investigates the hypotheses that dictatorship is conducive to economic growth, high investment rates, and better education policies. Panel data techniques are used on a sample of 21 Asian countries, with data for some countries going back to the 1860s. The main conclusion is that relatively dictatorial regimes are no better at generating high economic growth than relatively democratic regimes. The Lee thesis as a general hypothesis does not find support, even in Asia.
Why should the Lee thesis be especially relevant in Asia?
Even though some authoritarian countries have been successful in generating economic growth, general empirical evidence for the Lee thesis is lacking. Indeed, statistical analyses of global samples indicate either no effect (e.g. Burkhart and Lewis-Beck, Reference Burkhart and Lewis-Beck1994; Helliwell, Reference Helliwell1994; Przeworski et al., Reference Przeworski, Alvarez, Cheibub and Limongi2000) or a positive effect (e.g. Leblang, Reference Leblang1997; Baum and Lake, Reference Baum and Lake2003; Knutsen, Reference Knutsen2011) of democracy on economic growth. Moreover, Rodrik (Reference Rodrik2000) finds that economic growth under democracy is of ‘higher quality’ than under dictatorship, as democracies have less cross-country variation in growth rates, experience less volatility in GDP growth over time, manage external shocks to the economy better, and pursue growth with more egalitarian distributional outcomes.
Although democracy on average produces equal or higher economic growth rates; dictatorships vary more in their economic performances (Rodrik, Reference Rodrik2000; Besley and Kudamatsu, Reference Besley and Kudamatsu2007). As Barro puts it, ‘history suggests that dictators come in two types: one whose personal objectives often conflict with growth promotion and another whose interests dictate a preoccupation with economic development’ (Reference Barro1997: 50). Some of the most profound economic miracles after World War II (WWII) have taken place in dictatorial Asian countries. Is there something special about Asian dictatorships? One explanation for why some dictatorships in Asia have grown so fast is that many Asian dictatorships, such as the military regimes in South Korea and the Kuomintang in Taiwan, faced severe external security threats (Knutsen, Reference Knutsen2011). Only through economic development could these regimes improve their military capabilities enough to deter foreign adversaries. However, in the following I will present two common arguments as to why authoritarianism is more conducive to economic development than democracy in Asia. The first is a cultural argument related to the concept of ‘Asian values’, and how these make democracy a bad fit for Asian countries. The second is a political-institutional argument, indicating that Asian autocracies are different from autocracies elsewhere, and that they have been particularly good for economic growth.
Answer I: Asian values
‘Asian values’ is argued to include ‘the principle of benevolence, familism (with patriarchal authority), nepotism, authoritarianism, national consciousness, community spirit, fervor of education, hard work, and frugality’ (Lee, Reference Lee2003: 31). Asian values, it has been argued, make democracy less fit than autocracy as a mode of governance on the continent (Kim, Reference Kim1994). It has further been argued that the combination of Asian values and autocracy is a good prescription for development (Sen, Reference Sen1999), based on the ‘interaction’ between cultural and political factors and how these affect the economy.Footnote 2 Asians are supposedly positive toward hierarchical government and not very appreciative of political and civil rights. The claimed hierarchical and patriarchal Asian family values are thus transferred to the national level. The result is an elitist argument about the need for a strong and intelligent national ‘father’ to make important political decisions. In other words, the ‘most able, and morally the most righteous and spiritually most advanced should share more responsibility of government’ (Tu Wei-ming cited in Sikorski (Reference Sikorski1996: 830)), indicating that elites are also better at generating economic development without popular involvement.
Even if some scholars have delimited the Asian values-argument to East Asia, linking it to Confuciusianism, an expanded version of the argument also includes other parts of Asia. As Sen (Reference Sen1999: 232) notes, this expanded version is rather ignorant, as it does not appreciate the differences in cultures between Asian countries. However, limiting the Asian values thesis to East Asia is also problematic. Sen (Reference Sen1999: 231–238) shows that the argument relies on a selective reading of Asian classical philosophers, even Confucius, and historical patterns. Moreover, the empirical attitudes and values of Asian citizens do not differ substantially from citizens elsewhere, once socio-economic factors are controlled for (Inglehart and Welzel, Reference Inglehart and Welzel2006: 156). Historically, it is the political elites who have pushed for the notion of Asian values (Kim, Reference Kim1994; Inglehart and Welzel, Reference Inglehart and Welzel2006: 289). It is therefore legitimate to ask whether the proposition of Asian values and their political implications is promoted because of self-interest rather than correctness.
Answer II: Asia’s model of developmentalist autocracy
Another argument for why dictatorship is particularly conducive to economic development in Asia, is the existence of a particular type of developmentalist autocracy in many Asian countries (Leftwich, Reference Leftwich2000). Dictatorship is a sprawling category, defined primarily as an antidote to democracy, which contains several different regime types with different sets of institutions, structured according to different logics of governance (e.g. Linz and Stepan, Reference Linz and Stepan1996; Hadenius and Teorell, Reference Hadenius and Teorell2006). Here, I will not go into the interesting analytical possibilities related to a more fine-grained classification of regimes in the comparative politics literature, but it can, for example, be argued that the strong, dominant party systems existing in many Asian autocracies lead to specific economic policies conducive to economic development (Leftwich, Reference Leftwich2000). I will focus instead on the political economy literature on East Asia’s development miracles, and ideal-typical models of Asian-style developmentalist regimes.
The factors behind the development successes of the ‘Asian Tigers’ (Hong Kong, Singapore, South Korea, and Taiwan) have been vigorously debated. The World Bank’s (1993) report on the subject focused on economic factors like savings rates, human capital accumulation, efficient domestic markets, export orientation, and prudent macroeconomic policies. Critics of the World Bank’s report pointed to several explanatory factors lacking in its analysis (Gilpin, Reference Gilpin2001: 321–329). For example, the Asian Tigers carried out creative and selective industrial policies. A variety of policies and incentive schemes such as cheap credit and tax breaks were provided, especially to exporting firms.
Then there are the political aspects: why were all these economic policies generated in the first place? In these countries, bureaucrats and politicians have often been credited for acting autonomously from interest group pressure. They have been willing to carry out tough reforms and follow a long-term perspective, for example, by promoting policies that contributed to a high national savings rate. To what degree were these factors connected to these regimes’ authoritarian nature? Supporters of the Lee thesis would argue they were strongly connected.
Several ideal-typical models have been constructed from the successful East and Southeast Asian experiences. Wade (Reference Wade1990: 25–26) summarizes Chalmers Johnson’s (Reference Johnson1982) model, according to which the main features of Asian developmentalist regimes are: (1) economic development as the top priority of state action; (2) a commitment to private property rights; (3) elite bureaucracies as vital for economic policy formulation; (4) states engaged in consultation and coordination with the private sector; and (5) bureaucrats ‘rule’ while politicians ‘reign’ – politicians stake out broad political and economic strategies, while bureaucrats engage in specific policy formulation. The politicians hold the monopoly over political power, and the regimes are ‘softly authoritarian’. Another model, which focuses more on policy than institutional factors, is Chang’s (Reference Chang2006: 55) East Asian model, which includes: (1) pro-investment as opposed to anti-inflation macroeconomic policy; (2) control of luxury consumption; (3) strict control of foreign direct investment (FDI); (4) pursuit of infant-industry protection and export promotion; (5) intentional attempts to exploit economies of scale through export promotion; and (6) a dynamic productivity view of competition, rather than a static allocation view. Although Chang focuses less on political institutional aspects, some argue that the policies in this ideal-typical model are functions of the authoritarian political regimes in East Asian countries: dictatorships are better able than democracies to pursue certain policies, which again spur economic growth. This is the Lee thesis. Below, I consider three theoretical arguments that support the Lee thesis more in detail. These arguments relate directly to several dimensions identified in the models above.
Theoretical arguments for and against the economic superiority of dictatorship
Economic growth is the increase in economic production, or aggregate income, over a time period, and is usually operationalized as annual percentage change in GDP per capita. Despite vast differences between economic growth models (see e.g. Solow, Reference Solow1956; Romer, Reference Romer1990; Mankiw et al., Reference Mankiw, Romer and Weil1992), an increase in economic output is assumed to come from either increases in productive inputs, physical capital, human capital and labor hours, or through increased efficiency. An increase in efficiency means that an economy produces more output for a given amount of productive inputs. Increased efficiency may result from increased division of labor, improved functioning of markets and other allocation mechanisms, and technological change. How can political regimes affect input accumulation or efficiency? Political regimes matter because they systematically promote different policies, such as investment policies and education policies, or because they affect certain economic institutional structures, such as the property rights system and bureaucratic structures. Policies and institutional structures are again important for input accumulation and efficiency (see e.g. North, Reference North1990; Rodrik, Reference Rodrik2000). The arguments below sketch out causal chains where dictatorship affects the probability of a certain policy or institutional structure being adopted, which in turn affects either input accumulation or efficiency, which in turn affects economic output.
Argument I: Dictatorship and investment
One common notion is that dictatorships ‘are better able to marshal the limited resources available and direct them towards productive activities that will increase economic output’ (Halperin et al., Reference Halperin, Siegle and Weinstein2005: 3). It is argued that democracies are vulnerable to pressures for mass consumption, and thereby reduced savings, which again reduces investment. Investment in physical capital is, as mentioned, one source of economic growth in traditional economic growth theory (Solow, Reference Solow1956). However, also in ‘new growth theory’, which focuses on the importance of technological change, savings-, and investment rates matter since ‘[E]conomies with higher savings rates grow faster because they allocate (endogenously) more resources to inventive activities’ (Helpman, Reference Helpman2004: 45). Thus, if democracy undermines investment in physical capital, dictatorship will have an advantage in promoting economic growth.
There are different varieties of the argument, but a common assumption is that the populace is shortsighted and wants immediate consumption, which reduces savings and investment. Democratic politicians are under pressure to redistribute resources to both private and public consumption, since consumers are also voters. Voters are assumed to punish politicians in the next election if their demands for consumption are not met. Dictators can more easily neglect pressures for mass consumption. In his analysis of the proximate sources of growth in the Asian Tigers, Young (Reference Young1995) finds that capital investment was an important contributor to these countries’ high growth rates. However, promotion of investment was also crucial for economic growth in other Asian dictatorships, such as Suharto’s Indonesia (Ravich, Reference Ravich2000: 145–153).
Democracy likely increases consumption relative to dictatorship through several channels. One is the political redistribution of wealth. The rich, which as a group do not contain the ‘median voter’, are expected to save and invest a higher proportion of their income than the poor (Keynes, Reference Keynes1997: 96–98). Thereby, taxation of rich people in democracy combined with redistribution to the relatively poor majority is expected to lower savings and investment rates. Likewise, democracy increases consumption through increasing wages (Rodrik, Reference Rodrik1999a). Freedom of association, including the right to form collective labor organizations, is an important feature of democracy, at least when defined broadly (Beetham, Reference Beetham1999). The right for workers to organize typically increases their bargaining power in the labor market, thereby generating higher wages. Workers are assumed to consume a higher share of their income than wealthy capital owners. Aggregate consumption therefore increases when wages go up and capital income down. Reducing wages through co-opting or coercing labor is one way to increase capital accumulation in dictatorships. In Singapore, the share of national income going to wages was very low, between 0.3 and 0.4, from at least 1970 to 1990 (Przeworski et al., Reference Przeworski, Alvarez, Cheibub and Limongi2000: 172). Labor in Singapore has been aligned with, and can even be described as co-opted by, the ruling People’s Action Party government (Deyo, Reference Deyo1998: 202). The numbers of labor disputes in South Korea and Taiwan in the 1960s were stunningly low, often in the single digits per year. Moreover, only a fraction of these were related to demands for wage increases (Cho and Kim, Reference Cho and Kim1998: 140). This outcome is not solely due to intra-firm conditions, as ‘[S]tate agents and enterprise managers in private firms cooperated with each other to prevent workers’ collective actions’ (Shin, Reference Shin1998: 13). Freedom of association has also been only weakly present or indeed absent in Malaysia and Thailand when ruled by the military (Deyo, Reference Deyo1998: 188).
Dictatorships may also discourage consumption and increase savings through failing to provide social security, welfare arrangements, and other government insurance schemes. Dictatorial governments need not provide as much social security to the population as democratic governments, because of lower political accountability and responsiveness. The response of rational citizens is to save privately when public alternatives are non-existent. Wade links the high household savings rate in Taiwan to the lack of social security systems, with people saving for old age and sickness. Taiwanese households also piled up savings to support expensive education for their children (Wade, Reference Wade1990: 62–63). The lack of political responsiveness and accountability to the broader electorate in dictatorships allow political elites to bypass welfare, social security, and education programs. These are programs that a ‘median voter’ would likely care about and push for in a democracy.
In Taiwan, Wade (Reference Wade1990: 62) shows that national savings from 1956 to 1979 were mainly driven by high government and household savings. Only around 10% of national savings came from private corporations. The lack of social security, subsidized higher education, and housing programs not only pushed households to save more; they also allowed the government to save tax revenue. High government savings led to high public investment in Taiwan. Public sector investment made up about half of the economy’s investments in the 1960s and 1970s (Wade, Reference Wade1990: 173).
Dictatorships can also impose direct regulations that curb consumption, which arguably are unpopular and thus not viable in democracies. One example is the regulation of luxury consumption in East Asia. These countries ‘imposed heavy tariffs and domestic taxes on, and sometimes even banned the domestic production as well as the import of, certain “luxury products” ’ (Chang, Reference Chang2006: 25). One example is the ban on tourism in South Korea under military rule, which ended with political liberalization in the 1980s. In addition, state-owned banks did not provide loans for consumers and importers of foreign luxury goods, such as cars, washing machines, and video cameras (Shin, Reference Shin1998: 8). The tax system can also be designed to discourage consumption and encourage saving. Wade argues that the Taiwanese government did exactly this by trying to ‘harness the tax system for savings and investment objectives as well as revenue ones’ (1990: 175). Several incentives for private savings were provided, and selective, indirect taxes on luxury goods and amusement services were pursued. Indirect taxes, including customs duties, accounted for about 70% of total tax revenues in the late 1970s (Wade, Reference Wade1990: 60).
Argument II: Dictatorship and state autonomy
The alleged superior economic performances of Asian dictatorships have often been attributed to the dictatorial state’s autonomy (Przeworski and Limongi, Reference Przeworski and Limongi1993: 56). Autonomous states can ‘achieve relative independence (or insulation) from the demanding clamour of special interests (whether class, regional, or sectoral, where they exist) and … can and does override these interests in the putative national interest’ (Leftwich, Reference Leftwich2000: 168). Leftwich describes the South Korean, Taiwanese, and Singaporean states as ‘strong and relatively autonomous’ (2000: 99). These states have resisted political pressures from consumers, particular businesses and organized labor, and have thus been able to carry out policies beneficial to industrial growth in particular, and economic growth in general. Democratic polities on the other hand are argued to be prone to capture from special interest groups (Olson, Reference Olson1982). Lobbying directed towards politicians by interest groups will also generate wasteful rent seeking, which detracts resources and focus from more productive ventures. It is argued that authoritarian politicians are insulated from such pressures, which enables them to conduct policies that take into account the overall national welfare, including economic growth.
Wade (Reference Wade1990: 294–296) evaluates the Taiwanese economic bureaucracy, with its power delegated from the political rulers, as well protected from interest group pressures. In the three most authoritative industrial policy-making bodies, there was hardly any representation of business interests until the early 1980s. Moreover, when the government and bureaucracy cooperated with particular business interests, it was on the terms of the government rather than business (Wade, Reference Wade1990: 276–280). According to Wade, the weakness of the legislatures in Taiwan, South Korea, and Japan limited interest group influence on economic policy: a relatively low degree of democratic control over government, the argument goes, enhances bureaucratic autonomy. Bureaucrats can thereby focus on implementing growth-enhancing policies without disturbances.
A related argument deals with the quality of decision makers under democracy and dictatorship. An ‘enlightened ruler’ supported by a well educated and impartial technocratic-bureaucratic elite is more likely to understand the policies needed for long-term economic growth than a short-sighted populace or the man it ultimately elects to office. If the majority is unknowledgeable, or easily led by demagogy, the argument goes, it is better for enlightened autocratic elites to dictate policy. The argument does not however explain how one can ensure that enlightened rulers occupy office in dictatorships, and it assumes that rulers are not only enlightened but also have incentives to enact the proper policies.
Argument III: Dictatorship and reform
Another argument for why dictatorships are better at generating economic growth than democracies is that democratic systems are bad at legislating and implementing efficiency-enhancing economic reforms. Democracies are ‘conservative systems of power’ (Leftwich, Reference Leftwich2000: 171). In order to push through economic reforms, the argument goes, strong leadership is needed and political quarreling should be avoided. Dictatorships have a wider range of instruments and a broader menu of alternative actions when it comes to introducing reforms. The careful and cumbersome democratic process with its institutionalized dialogue, checks, and balances, and the representation of different interests, makes rapid reform difficult. This is especially the case if some of the represented interests are outright opposed to economic reform. Under democracy, the required support for implementing a reform might be far higher than a majority of the population (Pierson, Reference Pierson1998: 553–558). Reform opponents might be ‘veto players’ (Tsebelis, Reference Tsebelis2002) in the reform process. A dictator is assumed to have the means, and in some instances also the incentives, to carry out painful reform. Reform can probably also be conducted with greater speed under dictatorship, since procedural steps needed under democracy and time-consuming negotiations can often be passed over.
Two efficiency-enhancing economic reforms in Asia were the land reforms conducted in South Korea and Taiwan after WWII, which distributed land from previous landlords to tilling farmers. This not only contributed to reducing rural inequality; there are also good theoretical arguments for why such redistribution increased economic efficiency (Banerjee and Duflo, Reference Banerjee and Duflo2005). Leftwich (Reference Leftwich2000: 6–7) argues that democratic governments would have had problems in carrying out such rapid and widespread land reform; the nature of the regime in these countries may therefore have spurred rural development.
Dictatorships may also have longer time horizons than democratically elected governments, since dictators generally stay in power longer than democratic leaders (Bueno de Mesquita et al., Reference Bueno de Mesquita, Smith, Siverson and Morrow2003). Democratic politicians facing re-election might have weak incentives to promote reforms. Economic reforms are often initially painful, and the gains to structural economic reform might come with a substantial time lag. Changes in government through election might also lead to reform reversals as electoral fortunes change. Anticipated reform reversals can have detrimental economic effects, reducing investment (Rodrik, Reference Rodrik1991).
Why the Lee thesis might be wrong
Although argument I implied that dictatorship increases investment, there are some countering factors. In an open economy, there is no necessary relationship between savings and investment because of FDI, although the empirical correlation has been high (Feldstein and Horioka, Reference Feldstein and Horioka1980). However, this correlation may have weakened in recent decades, as FDI has increased dramatically. Empirical studies find a positive effect of democracy on property rights protection (Leblang, Reference Leblang1996; Boix, Reference Boix2003; Knutsen, Reference Knutsen2010) and the control of corruption, at least in consolidated democracies (Rock, Reference Rock2009a). Several studies show that FDI is negatively affected by expropriation and corruption (see e.g. Blonigen, Reference Blonigen2005). Thus, even if dictatorships force their citizens to save more, better investment climates in democracies might mitigate the dictatorial investment advantage through attracting FDI.
Argument II on dictatorship, state autonomy, and economic growth, can be disputed on several points. Even if dictatorial regimes are autonomous from the general population, autonomous dictators do not necessarily have an interest in pursuing growth-enhancing policies, which include the economic reforms discussed in argument III. Such an interest can stem from dictators being altruistic or from dictators’ self-interest being served by economic growth. Wintrobe (Reference Wintrobe1998) notes that history has seen few, if any, altruistic dictators. The second assumption is perhaps more often valid, for example, when a dictator’s legitimacy and probability of survival depends positively on growth. Nevertheless, several studies show how dictators’ self-interest is often enhanced by violating property rights and under-providing public goods (see e.g. Bueno de Mesquita et al., Reference Bueno de Mesquita, Smith, Siverson and Morrow2003; Acemoglu and Robinson, Reference Acemoglu and Robinson2006a). Dictators might also have incentives to block efficiency-enhancing reforms, as discussed in argument III, that undermine their political position (Acemoglu and Robinson, Reference Acemoglu and Robinson2006a). These policies are detrimental to economic growth, thereby casting serious doubt on the validity of arguments II and III. Another counterargument to II is that dictators are not always autonomous. Dictators also rely on important interest groups for securing their position in government. In practice, a dictator’s survival might, for example, rely on pleasing a particular ethnic group, the military, or a group of rich influential businessmen or landowners, depending on the dictator’s particular ‘winning coalition’ (see e.g. Bueno de Mesquita et al., Reference Bueno de Mesquita, Smith, Siverson and Morrow2003; Besley and Kudamatsu, Reference Besley and Kudamatsu2007). Thus, although dictators may often act autonomously of the preferences of a majority of citizens, they often cannot act autonomously of their political backers. A third counterargument to II deals with the relationship between autonomy and economic growth. Evans (Reference Evans1995) convincingly argues that autonomy is not necessarily conducive to growth. The linkage between state and society that best promotes growth is a particular form of ‘embedded autonomy’ (Evans, Reference Evans1995).
Additional arguments imply a ‘democratic growth advantage’ (see e.g. Przeworski and Limongi, Reference Przeworski and Limongi1993; Halperin et al., Reference Halperin, Siegle and Weinstein2005; Knutsen, Reference Knutsen2011). Democracy is believed to protect property rights better (North, Reference North1990; Knutsen, Reference Knutsen2010) and to curb corruption (Rock, Reference Rock2009a), which affect not only capital investment but also efficiency. Technological change is the most important factor for long-term economic growth (Romer, Reference Romer1990). Dictatorship, through restricting civil liberties like freedom of speech, media and travel, reduces information flows into, and within a country. Democracies thus allow for more widespread diffusion of ideas and technologies, which enhances technological change (Halperin et al., Reference Halperin, Siegle and Weinstein2005; Knutsen, Reference Knutsen2011). There is empirical evidence that democracy enhances total factor productivity (TFP) growth, a proxy for technological change-induced economic growth (Knutsen, Reference Knutsen2011). Interestingly, Young (Reference Young1995) shows that Singapore, South Korea, and Taiwan accumulated capital fast, but had unspectacular TFP-growth rates.
There is also evidence that democracy enhances human capital accumulation (e.g. Brown, Reference Brown1999; Lake and Baum, Reference Lake and Baum2001; Baum and Lake, Reference Baum and Lake2003). Democratic politicians are more accountable to their citizenries than dictators, and health care and education are not only productivity-enhancing but also desired by most citizens. Thus, even if dictatorships invest more, democracies are better at generating two other ingredients of economic growth, human capital, and technological change.
Problems of inference and generalization
As seen above, there are problems with the arguments implying the economic superiority of dictatorship, and there are also arguments pointing to the economic superiority of democracy. However, the successful authoritarian governments in Taiwan, South Korea and Singapore have arguably mitigated the ‘democracy advantage’ in several areas. As Przeworski et al. (Reference Przeworski, Alvarez, Cheibub and Limongi2000: 211) point out, together with Pinochet’s Chile, these are the only recent experiences of authoritarian countries providing decent property rights protection. Singapore is by far the least corrupt non-democracy in the world according to Transparency International’s surveys, and the societal ‘embeddedness’ of South Korean policy makers and bureaucrats (Evans, Reference Evans1995) may have mitigated problems of information transmission that pervade other dictatorships. Singapore, South Korea, and Taiwan were thus among the world’s fastest growing countries in the three decades after 1960. Thailand and Indonesia also enjoyed periods of decent economic growth under dictatorship. More recently, China has recorded a double-digit growth rate under a Communist one-party regime. Vietnam has also seen a booming export industry and a healthy economic growth rate over the last two decades. Therefore, one could argue, the case is made for the Lee thesis in the Asian context. Although dictatorship has created economically disastrous results elsewhere, is not the combination of Asian values and an Asian type of dictatorship a perfect recipe for economic growth? Even though there are many Asian authoritarian success-stories, there are several problematic methodological aspects related to treating the cases above as conclusive evidence for the Lee thesis in Asia.
As Geddes (Reference Geddes2003: 93–129) notes, the case-based literature on the suppression of labor rights and its economic effects provides a good example of case-selection on the dependent variable, and a resulting bias. Some countries, such as South Korea and Taiwan, have been systematically selected for study precisely because of their stellar economic performances. Results from these case studies have been erroneously generalized. This point also applies to the qualitative research on Asian development performances in general. What if the development disasters are also softly authoritarian and employ export oriented economic policies? Most poor countries after WWII were relatively authoritarian, and even if the countries that caught up with the West were drawn purely by chance, we would have seen more dictatorial countries grow at rapid rates than democratic.
Thus, the case for the Lee thesis may have been made too quickly in the Asian development literature. First, democratic Japan experienced astonishing economic development after WWII. In more recent years, India has also experienced high growth. Second, South Korea and Taiwan continued to grow relatively fast after democratization. Although growth rates in these two countries are not as high as those of earlier years, it is more difficult for already rich countries to grow fast (Barro and Sala-i-Martin, Reference Barro and Sala-i-Martin2004). Third, dictatorship has been the historical norm in Asia, without much economic success (perhaps with the exception of Meiji-Japan) before WWII. Fourth, there are several examples of Asian autocracies performing poorly over the last decades as well, such as North Korea and Myanmar. In the Philippines, the relatively democratic decade after Marcos’ fall saw a 2.5 percentage point increase in the annual GDP per capita growth rate when compared to the previous decade according to Penn World Tables data. More attention has been paid to authoritarian economic successes in the Asian development literature than to the failures. If we want to make general inferences on the effect of dictatorship on growth in Asia, we need to take into account all available evidence, rather than generalizing from a few selected cases. Indeed, Rock (Reference Rock2009b) found no negative effect of democracy on economic growth in an Asian sample. This result is based on analysis of 12 cross section units with time series going from 1960 to 2004.
Some would claim that generalization or testing is not the purpose of ideal-typical Asian developmental regime-models, and that such models simply yield a description of common traits of particular regimes in particular countries. However, the problem is whether or not we can really assert that autocracy has been empirically important for growth, even in, for example, South Korea. With few cases and many variables, we have a degree of freedom problem (King et al., Reference King, Keohane and Verba1994). It is difficult to disentangle what actually influenced growth in the East Asian development miracles. As seen from the ideal-typical models above, several dimensions were identified as important for the development of East Asian economies. It is difficult to evaluate the relative importance of each of these. In order to avoid selection bias and the degree of freedom problem, we should assess the whole sample of Asian countries over a substantial time period.
Setting the record straight: the general impact of dictatorship on Asian economies
This study uses the most extensive data set possible for investigating the economic effects of democracy in Asia. Some analyses cover more than 1200 country-year observations. Some countries have data from the 1860s to 2003, although most countries have far shorter time series. The sample covers 21 Asian countries outside the Middle East and the former Soviet Union.Footnote 3 Since colonies are not independent countries, colonial histories are left out. The data are mainly from the Polity IV project and Angus Maddison’s (Reference Maddison2006) data set. The first provides data on political regime type in terms of degree of democracy (Polity index), and duration of the regime in power. The second provides data on population, GDP per capita and economic growth. In addition, there are dummies for plurality religion, with data collected from different sources (Knutsen, Reference Knutsen2007). The Polity index ranges from −10 to 10 (most democratic). The index is additive, and the components are competitiveness of executive recruitment, openness of executive recruitment, constraints on the chief executive, competitiveness of political participation, and regulation of political participation (Marshall and Jaggers, Reference Marshall and Jaggers2005: 12–15). So-called interregnum-periods, mainly periods of anarchy or civil war (Marshall and Jaggers, Reference Marshall and Jaggers2005: 17), as well as periods of ‘foreign interruption’ (mainly occupation), are left out of the analysis. GDP per capita data are PPP-adjusted, and thus take into account local price levels. Maddison (Reference Maddison2006, Reference Maddison2007) relies on several sources for estimating GDP, and the estimates are plagued by measurement errors, particularly in early years where less information is available. Nevertheless, these are the only available economic data that cover the years before WWII. The Maddison data allow us to incorporate extra historical information, for example, by expanding the Chinese and Japanese time series. In addition, the Maddison data include estimates for countries left out of other data sets, such as North Korea, which mitigates the selection bias related to the under-representation of dictatorships with bad economic records. Elsewhere (Knutsen, Reference Knutsen2011), I have developed an interpolated sample, which constructed time series for GDP and population where there were gaps in the original Maddison data. The interpolations were only conducted if there were no significant changes in the degree of democracy over the time period in question, in order to avoid biases. In all models below, the independent variables are lagged with 2 years to reduce endogeneity problems.
Economic growth
I use pooled cross-section time-series data with country-year as unit of analysis, in order to incorporate both information from intra-national variation over time and information from cross-country comparisons. I first run ordinary least squares (OLS) with panel corrected standard errors (PCSE), which takes into account that standard errors can be heterogeneous across panels, autocorrelated within panels (AR1), and contemporaneously correlated between panels. The regression models include several control variables. Log of GDP per capita is controlled for, since income level is believed to affect both the probability of being democratic and economic growth. Log of population size is also controlled for because of the same reasons. The model also includes the log of the political regime’s duration to control for political stability. Dummies for plurality religion are entered to control for culture-specific effects. This is particularly relevant because of the ‘Asian-values’ argument, as such controls enable us to capture differences in effects of democracy within sets of culturally similar countries.Footnote 4 Dummies for time periods are also entered to control for time-specific effects. I use dummies for the time periods 1870–1913, 1914–45, and 1946–72 in some models, and decade-dummies in others, with the 1990s-dummy being extended to 2003. The regression equation is thus given by:
The empirical results from the OLS with PCSE analyses do not support the Lee thesis. The estimated effect of democracy on economic growth in Asia is almost zero. The models predict less than or about 0.1 percentage point extra annual GDP per capita growth when going from a ‘perfect’ democracy (+10) to a harsh dictatorship (−10). As seen in Table 1, the t-values are also almost zero, and we cannot reject the hypothesis of no effect of democracy on growth in Asia when analyzing the full sample of 21 countries, with some time series going back to the mid-nineteenth century.Footnote 5 China after 1979 and Singapore under Lee might be used as partial evidence for the Lee thesis, but the Lee thesis as a more general claim does not hold, even if tested on a purely Asian sample. Although some Asian dictatorships have grown fast, other dictatorships, such as North Korea and pre-1979 China, have not. Moreover, although some democratic economies, such as post-Communist Mongolia, have grown slowly, others have grown at a decent pace.
Results from OLS with PCSE analyses
PCSE = panel corrected standard errors.
The Asian economic miracles are a post-WWII phenomenon, and some would perhaps claim that the Lee thesis is constrained temporally. However, the results reported in Table 2 show that there are no large changes in the results when delimiting the sample to after WWII. The same is true for a sample using data only from after 1960. According to these analyses, there are no empirical reasons to believe that dictatorship in general improves economic growth rates in Asian countries. Although there is no support for the Lee thesis in Asia, there is no significant positive effect of democracy on growth in Asia either. This stands in stark contrast to the positive effect of democracy on economic growth found in global samples (Knutsen, Reference Knutsen2011). Thereby, the hypothesis that Asian dictatorships are better at generating economic growth than dictatorships elsewhere is not refuted by this empirical exercise. Asian dictatorships may still be considered developmental, but only when compared to the poor performances of dictatorships elsewhere.
Results from OLS with PCSE analyses with decade dummies
PCSE = panel corrected standard errors.
What if there are country-specific factors, for example, cultural or geographical factors, which were not included in the analyses above that influence both regime type and economic growth? If so, the models above are incorrectly specified. By including dummy variables for every country in a Fixed Effects (FE) analysis, we control for country-specific effects. FE thereby only utilizes intra-national variation; that is, variation in the degree of democracy and economic growth within nations over time. FE estimators are therefore not affected by differences in growth rates between democratic India and authoritarian China or North Korea, but only by changes in growth rates between democratic and dictatorial periods in the histories of individual countries. In addition to country-specific effects, the models control for time dummies, log GDP, log population, and log regime duration.
I ran FE regressions with robust standard errors on the data samples with the longest time series. The results are strikingly similar to those obtained in the OLS with PCSE analyses: there is no significant effect of democracy on economic growth, even at the 10% level. All the absolute t-values in Table 3 are lower than 0.7. Even if we control for country-specific effects, and only use information based on changes in regime and growth patterns within a country’s history, there is no support for the Lee thesis. We find no evidence that Asian countries have grown slower as democracies than as dictatorships. The models thus imply that democratization does not hurt economic growth. This reinforces the result from Rock, who finds ‘no autocracy growth advantage even for the so-called developmental states of East Asia’ (Rock, Reference Rock2009b: 947).
Results from fixed effects analyses (robust SE)
In FE regressions on samples with shorter time series, reported in Table 4, the estimated effect of democracy on growth decreases. The post-1960 analyses indicate that the replacement of a very democratic (+10) with a very dictatorial (−10) regime will give 1 percentage point extra annual GDP per capita growth. However, the estimate is very uncertain, and the negative effect of democracy on growth is statistically insignificant, even at the 10% level. The same is true for the post-1945 analysis. We will investigate possible temporal changes in effects below.
Results from fixed effects analyses (robust SE) with decade dummies
Tests were also performed on a different sample, using GDP data from the World Development Indicators (WDI). This sample contained only 17 countries, excluding, for example, North Korea, with time series from 1975 to 2004. There was no significant effect of democracy on economic growth in this sample either when running OLS with PCSE or FE, with decade dummies as time controls. OLS with PCSE models showed a negative estimated effect of democracy, with a t-value of −1.2. FE models, however, showed a positive estimated effect of democracy, with a t-value of 1.0. I checked the robustness of the results by substituting the Polity index with the Freedom House Index (FHI) and the dichotomous regime measure (AREG) from Alvarez et al. (Reference Alvarez, Cheibub, Limongi and Przeworski2004).Footnote 6 There was still no significant effect of democracy on growth, even at the 10% level, both when running OLS with PCSE and FE models. The sign of the estimated effect of democracy alternated between positive and negative in different models. Independently of whether we incorporate cross-national variation or not, and independently of whether we use a broad operationalization of democracy, including, for example, civil liberties (FHI), or a narrow operationalization, based on contestable elections only (AREG), there is no significant effect of democracy on growth. The first result presented therefore holds: there is no empirical evidence for the Lee thesis in Asia.
It might however be that rich Asian democracies like Japan, Taiwan, and South Korea drive these results, and that democracy reduces growth in poorer Asian countries. However, OLS with PCSE models that include an interaction term between polity and log GDP, show a negative significant (5% level) interaction term and a positive significant (5% level) Polity-coefficient, indicating that democracy increases growth in the poorer Asian countries, but not in the richer.Footnote 7 This result goes against the ‘premature democracy is bad for growth’ argument (Sirowy and Inkeles, Reference Sirowy and Inkeles1990). Political accountability and checks and balances on those in power might be extra important in poor countries, in order to mitigate these countries’ problems with poor protection of property rights, high corruption and the misallocation of scarce public funds to purposes other than basic public goods. These findings also have indirect implications for the ‘optimal path’ to stable democracy in the long run. Poor democracies are more likely to collapse than richer democracies (Przeworski et al., Reference Przeworski, Alvarez, Cheibub and Limongi2000). If dictatorships in poor developing countries had had higher growth rates than democracies, the optimal path could have been sticking with dictatorship until Lipset’s (Reference Lipset1959) requisites for stable democracy were established, and then transition. However, since poor Asian democracies grow at least equally as fast as poor dictatorships, there is no reason to postpone ‘premature’ democratization for democracy’s sake.
The relationship between democracy and growth might have changed over time. Rodrik (Reference Rodrik1999b) argues that democracies mitigate the economic problems and improve the benefits of economic openness: they provide institutions that allow for bargaining and the compensation of losers in international trade, and provide more flexible policies in response to external economic shocks, such as the 1997 Asian financial crisis. As trade and capital flows have increased substantially over the last three decades, also in Asia, the effect of democracy on growth might thus have become more positive. This hypothesis can be tested with a Chow-test (Greene, Reference Greene2003: 130–133). I run Chow-tests on the OLS with PCSE models to investigate whether the effect of regime type on economic growth is different before and after 1980. However, I find no significant change. The estimated effect of democracy was slightly positive before 1980 and slightly negative after 1980 in all models, although the difference is insignificant at the 10% level. The findings are qualitatively similar when I test for changes in effect from before and after 1990.
As seen above, there are arguments indicating that dictatorships enhance economic growth through increasing capital accumulation and increasing efficiency because of state autonomy and an ability to push through economic reforms. The lack of significant results could indicate the three arguments are not as relevant as their proponents believe, even in Asia. However, other arguments point to mechanisms through which democracy enhances growth, and the lack of an aggregate effect could result from several mechanisms cancelling each other out. We can assess the validity of argument I more directly through looking at how regime type affects investment.
Investment
One of the arguments backing the Lee thesis was that dictatorships are able to push up savings and investment rates. Although there is no effect of democracy on growth in Asia, there might be a negative effect of democracy on the share of GDP going to real capital investment. I used the same independent variables as above, including decade dummies, and ran OLS with PCSE models. The Maddison data set does not contain investment data, therefore I used WDI data on real investment as a share of GDP (INVSHARE), GDP per capita and population. The longest time series extends from 1975 to 2003 and 17 countries are included (424 observations). The effect of democracy on INVSHARE is negative and significant at the 5% level. Thus, there is reason to believe that dictatorships on average allocate a larger share of their GDP to investment. According to the estimate, a dictatorship (polity = −10) invests 2.2 percentage points more of its GDP on average than an otherwise equal democratic country (polity = 10). When we use the FHI instead of Polity, the negative effect of democracy on INVSHARE is only significant at the 10% level. The AREG measure, however, is insignificant at the 10% level. When we run FE analyses with INVSHARE as dependent variable, the Polity index still has an estimated negative coefficient, but is now insignificant at the 10% level. However, the FHI is now significantly negative at the 1% level. There is therefore at least some evidence for the hypothesis that democracy hampers investment in Asia, which indicates partial support for argument I.Footnote 8 Combined with the lack of significant effect on economic growth, this could imply that Asian democracies are better at generating human capital, or at promoting the efficient use of resources and technological change, than Asian dictatorships.
The relationship between dictatorship and investment may have changed over time due to the globalization of capital markets, as discussed above. If dictatorships are better at generating domestic savings, but democracies are better at attracting FDI, we would expect the effect of dictatorship on INVSHARE to have shrunk in recent decades. However, neither models using the Polity index, FHI, or AREG find a significant change in the effect of regime type on investment before or after 1980 when applying Chow-tests.
Human capital
Although dictatorship increases investment in Asia, there was no detectable effect of dictatorship on economic growth. It might be that democracies allocate their investments more efficiently. There may also be other reasons. As seen above, one general assertion is that democracies accumulate more human capital by providing well functioning education systems for their populations (Brown, Reference Brown1999; Lake and Baum, Reference Lake and Baum2001; Baum and Lake, Reference Baum and Lake2003). Moreover, studies of other regions, such as Africa (Stasavage, Reference Stasavage2005), have shown that democracy improves education policies. I test whether democracy had a positive average effect on gross school enrollment ratios, using data from the WDI. For primary schooling, there are data for 17 countries, but the short time series yield only 120 observations. I ran OLS with PCSE models with the same independent variables as above. The result is somewhat surprising, as democracy, as operationalized by Polity, FHI, and AREG, has a negative significant (5% level) effect on primary school enrollment in Asia. Political economic models indicate that democratization expands education to larger parts of the citizenry due to political pressures (Acemoglu and Robinson, Reference Acemoglu and Robinson2006b). One explanation for this puzzling result might be that many Asian post-WWII dictatorships have been Communist. Communist regimes have historically scored high on primary school enrollment and literacy. In addition, Brown (Reference Brown1999: 686), drawing on arguments related to this paper’s arguments II and III, proposes that when a dictator wants economic development and ‘attempts to “build” a nation, authoritarian regimes are more efficient at mobilizing an entire society toward universal enrollment at the primary school level’.
However, democracies outperform dictatorships on secondary school enrollment. The estimated coefficients are large, and all democracy indicators are significantly positive, at least at the 1% level. The same result holds for tertiary school enrollment, with all democracy indicators being significantly positive at the 0.1% level. These analyses, although the data material is limited, indicate that while dictatorships in Asia were better than democracies at generating high primary school enrollment, they were far worse at generating high secondary and tertiary school enrollment. The latter results are in line with theoretical expectations. Secondary and tertiary school enrollment is probably extra important for productivity in knowledge-intensive economies focused on production in complex manufacturing and service sectors. As Asian countries aim to become ‘knowledge economies’ in the future, dictatorships may then perform worse than democracies. However, if Brown’s (Reference Brown1999: 686) argument as presented above is valid, it might be that some dictatorships will manage to mobilize high levels of secondary and even tertiary enrollment as their economies develop further, should their leaders want to do so.
Conclusion
Leftwich calls the belief that democracy is necessary for economic development an ‘orthodoxy’ that is not well-founded in classical political science literature or empirical evidence (2000: 129). He argues that if economic development, and particularly growth, is to emerge in many developing countries, ‘it will be under the auspices of a political system that is a lot less pleasant than democracy’ (15). Some scholars studying East- and Southeast Asian politics conclude that dictatorship is the best regime for promoting economic growth, and that democracy is a luxury good for rich countries (Sen, Reference Sen1999: 15). This is the Lee thesis. This paper reviewed two proposed reasons for why dictatorship is more conducive to economic growth in Asia than elsewhere: Asian values and Asian developmental dictatorships. However, there are flaws in the Asian values argument, and it is unclear whether the developmental characteristics of Asian regimes are connected to their dictatorial nature. The paper then sketched three general arguments that indicate the economic superiority of dictatorship: dictatorships invest more, are more autonomous from social groups, and can conduct economic reforms more easily. Although the arguments seem reasonable, there are strong counterarguments. Moreover, there are other factors indicating the economic benefits of democracy.
The South Korean, Taiwanese, Singaporean, and recent Chinese experiences provide convincing anecdotal evidence for the Lee thesis. However, in order ‘to assess the impacts of political regimes, we must examine their full record, not just the best performers’ (Przeworski et al., Reference Przeworski, Alvarez, Cheibub and Limongi2000: 4). The statistical analyses in this paper provide no evidence for the Lee thesis, even in Asia. Panel data models using data from up to 21 Asian countries found no significant effect of dictatorship on economic growth, regardless of the time period investigated. However, as the theoretical discussion predicted, dictatorships invest a larger fraction of their GDP. Somewhat surprisingly, dictatorships also have higher primary school enrollment. Dictatorships are however worse when it comes to generating secondary and tertiary school enrollment.
Nevertheless, Asian dictatorships perform better than dictatorships globally when it comes to the statistical relationship with economic growth. Globally, dictatorship is associated with lower growth rates (Knutsen, Reference Knutsen2011). However, even in Asia ‘there is not a single reason to sacrifice democracy at the altar of development’ (Przeworski, Reference Przeworski2004: 1). The Lee thesis is at best a proposition that bears relevance for a particular ‘species’ of dictatorship, and at worst it is flat out wrong. Future statistical research should look at the relationship between different types of dictatorships and economic growth.
This paper’s findings have some interesting implications. First, the paper illustrates Geddes’ (Reference Geddes2003) methodical point: one should be cautious when generalizing from a few cases selected on the dependent variable. Second, Inglehart and Welzel (Reference Inglehart and Welzel2006) find that Asians do not have such authoritarian values as some have suggested. This paper reinforces the criticism of the rosy picture of Asian authoritarianism, as its often postulated positive economic impact is questioned. Third, the paper has implications for policy makers working with democratization issues. One argument against promoting democratization in developing countries is that democracy is inappropriate in such contexts, because of negative effects on political stability and economic outcomes. Some policy makers, journalists, and academics still consider dictatorship a useful instrument for bringing economic development to poor countries. The post-1979 Chinese miracle is used as anecdotal evidence for the economic superiority of dictatorship today in the same way that South Korea and Taiwan were used a couple of decades ago. This paper indicates that the proposed economic benefit of dictatorship, even in Asia, is highly exaggerated. Conversely, democracy’s effects on economic development in poor countries is underrated. Policy makers and others can promote democracy in developing countries without a guilty conscience about the economic consequences.
Acknowledgements
I would like to thank participants at the 2009 Norwegian National Political Science Conference’s ‘Comparative Politics II’ panel, and participants at the 2009 ISA Conference’s ‘Globalization, Political Regime, and Prospects for Democratization in East Asia’ panel, for comments on a previous version. I would especially like to thank Helge Hveem and John Ravenhill for fruitful comments.
