Abstract
This article studies the effect of political regime type on economic growth in sub-Saharan Africa. Democracy promotes growth because it conditions government consumption so that consumption is used for public purposes rather than private needs and this in turn leads to faster growth. By conditioning consumption towards public goods and away from private goods, we should see that consumption in democratic regimes is associated with more public goods like roads and education while in authoritarian regimes consumption yields less of these goods. Likewise, consumption should be associated with falling fertility in democratic regimes and rising fertility in authoritarian regimes. Using several measures of growth, the empirical estimates from a large-n fixed-effects regression show that democracy conditions consumption so that the latter is associated with faster growth. Moreover, the empirical analysis indicates that government consumption in democratic regimes is associated with more education completion and lower fertility rates.
Introduction
Since gaining independence in the 1960s or thereafter, very few sub-Saharan African countries have experienced sustained economic growth. As a result of the inability to grow over long periods, many Africans still live in extreme poverty (Chen and Ravallion 2010). Scholars in political economy and international relations are interested in explaining why growth in Africa has lagged behind other parts of the world. Explanations have centered around factors related to geography, global or external factors, and politics. In terms of politics, many scholars see regime type, or the type of political system, as an important determinant of growth and development in Africa (Knutsen 2013). To this point, development economist William Easterly (2014) argues that in Africa, “Autocracy itself perpetuates poverty” while political scientist Larry Diamond (2005) claims that, “Africa cannot develop without democracy.” Increasingly, however, China’s explosive growth has led many to wonder if authoritarianism is the best model for African countries to grow and develop (Halper 2012). “Economic explosions in African countries like Rwanda and Ethiopia,” Hughes (2011) argues, “have lent credence to Chinese-style totalitarianism as a development model.” For Africa, then, is democracy the best way to promote growth or is authoritarianism a better way forward?
This paper examines this question focusing specifically on the connection between regime type and economic growth in sub-Saharan Africa. By providing people with the means to condition their leaders, democracy conditions leaders to pursue policies that encourage growth while consumption in authoritarian regimes is used for purposes less beneficial to the population at large and thus less catalytic for growth. In Africa, many regimes have been dominated by a single ruler or small clique, that have ruled in their own interests and not for the country at large and this has resulted in poor economic outcomes. To test this argument, a comprehensive empirical analysis shows that increasing government consumption in relatively democratic regimes is associated with more rapid growth while consumption in more authoritarian regimes is associated with a slowdown in growth. Notably, this result holds across several different measures of growth, and the empirical analysis provides evidence that consumption in democratic regimes is associated with improved education outcomes, and possibly lower fertility rates, while in authoritarian regimes it is associated with a decline in education completion rates and higher fertility.
Democracy, this analysis finds, encourages growth by focusing government behavior in ways that enhance economic activity. Democracy, in short, promotes growth by encouraging governments to do the things needed for growth like providing more education. Understanding the dynamics of growth in Africa, and how democracy affects growth, is important and of interest to a wide range of scholars (Acemoglu and Robinson, 2010; Ayittey, 2006a; Collier and Gunning, 1999; Easterly, 2009; Van de Walle, 2001). The question asked in this analysis about regime type and growth is important for a number of reasons but especially since countries in Africa are undertaking democratic reforms (Cheeseman, 2015) and many wonder if there will be benefits to political liberalization. Moreover, Africa is an interesting region to test the growth effects of democracy as most regimes there are to some degree authoritarian though increasingly democratic processes have been spreading across the region (Cheeseman, 2015). With the increasing variation in democratization, Africa remains an excellent context to examine if democracy encourages growth and thus to determine what exactly are the mechanisms connecting democracy and growth. This paper examines the democracy-growth nexus arguing that in Africa, and possibly beyond, democratic institutions exert important conditioning effects on the behavior of governments that in turn increase growth. The following article presents a literature review together with a theoretical discussion, empirical analysis, and conclusion.
Literature review
To date, there is no consensus on the growth effects of democracy in general, or for Africa per se. As it stands, a wide-ranging set of ideas exist about democracy and growth including that democracy limits growth, enhances it, or that it has no effect. In political economy, research indicates that relatively democratic regimes promote growth (Acemoglu et al., 2014; Baum and Lake, 2003; Feng, 1997; Feng, 2003; Gerring et al., 2005) and development (Kudamatsu, 2012; Stroup, 2007), as well as reducing volatility in growth (Mobarak, 2005; Quinn and Woolley, 2001). 1 However, some research finds a positive connection between authoritarianism and growth (Barro, 1998; Dick, 1974; Glaeser et al., 2004), with some evidence indicating that the effect of regime-type on growth depends on the region under consideration (Krieckhaus, 2006). Examinations of the regime-growth question argue that there is ultimately no consensus about this relationship (Knutsen, 2012; Przeworski and Limongi, 1993; Przeworski et al., 2000). Knutsen’s (2012) review of the research on democracy and growth concludes, however, that it is increasingly less plausible that democracy reduces or has no effect on growth.
Within Africa, several studies argue that authoritarianism is responsible for historically weak growth and the low levels of development (e.g. see Ake, 1996; Bates, 2005, 2010; Bates and Block, 2013; Van de Walle, 2001, 2003). In terms of quantitative analyses, Feng (1996) finds a positive relationship between democracy and growth for 40 African countries from 1960 to 1992. Bates and Block (2013) find that electoral competition is associated with more agricultural output and faster productivity growth in agriculture. 2 Similarly, Fosu (2008) finds that electoral competition is associated with more rapid growth, although this is true only at higher levels of electoral competition. Across several periods from 1986 to 2006, Lewis (2008) reveals that democratic regimes are associated with faster growth than non-democratic regimes, but this conclusion is limited because it is established by a series of bivariate correlations lacking appropriate controls. Through a number of rigorous empirical tests, Jaunky (2013) shows that there is a joint relationship between regime type and growth in Africa, thus raising concerns that existing empirical estimates are biased due to endogeneity. Finally, Masaki and Van de Walle (2014) conclude that African democracies grow faster than less democratic countries, and moreover that those countries with a longer tenure of democracy grow even faster than those with a shorter history of democracy. 3
Perhaps the most comprehensive analysis of democracy and growth in Africa, and the one most relevant for this analysis, is Knutsen’s (2013) analysis of democracy, state capacity, and growth. Using a wide-ranging and comprehensive empirical analysis, Knutsen finds that democracy enhances growth where state capacity is relatively low which means that democracy is likely to be conducive to growth in Africa where state capacity is typically low. When state capacity is low or weak (i.e. bureaucracies have trouble functioning), democratic regimes are likely to outperform dictatorships in terms of growth. In contrast, when state capacity is low, dictators act as they please and pursue spending that personally benefits them yet does nothing for the economy at large. Thus, authoritarian regimes should grow faster when state capacity is high compared to when it is low. Using a comprehensive empirical analysis, Knutsen finds that democracy has a stronger positive effect on growth when state capacity is low while dictatorships have more of an effect when capacity is high.
In many ways, the analysis here complements and builds on Knutsen’s (2013) analysis. Similarly, this analysis finds that the effect of democracy is conditional, but here the conditioning factor is argued to be government consumption instead of state capacity. While the conditioning factors are different, this paper complements the findings of Knutsen and others as it shows the effect of democracy on growth is robust to multiple indicators of growth, a point yet to be addressed in the research. Existing research has largely if not entirely relied on the World Bank for data on growth which is problematic because economic data reported to the World Bank is probably misreported (Jerven, 2013), and in particular it is likely biased upward when reported by authoritarian regimes (Magee and Doces, 2015). In Africa it is likely that the growth benefits of democracy are understated as authoritarian regimes are reporting inflated rates of growth. To address this issue, this analysis employs data from three different sources including the World Bank, Penn World Tables, and most importantly from satellite images of night-time lights. The latter source – night-time lights – is valuable as on the whole, it cannot be manipulated by governments. Together, all three sources show that the effect of government consumption on economic growth increases under democratic regimes and decreases under authoritarian regimes. Combined with earlier work, this analysis contributes to an emerging consensus that there is likely a growth-enhancing effect of democracy on growth in Africa and that this effect is conditional. The next section explains how regime type and consumption interact to affect growth.
Regime type and growth
While there is significant geographical and cultural diversity across Africa, the experiences of these countries, especially in terms of their political systems and historical economic performance, is quite similar and for that reason focusing on this region alone makes theoretical sense. To this point, Ayittey (2006b) notes: “The basic African beliefs, political, legal, and economic institutions are strikingly and structurally the same across much of Africa.” Unlike other parts of the developing world, in Africa populations are rapidly urbanizing yet still rural in many ways, land is relatively abundant, and most African countries were colonized and decolonized around the same time. Economically most African countries are still relatively poor and in prominent sectors of the economy, like agriculture, traditional methods define the production process (Juma, 2011; Shelburne, 2009). Building on this context, the following subsections explain how democracy affects growth in Africa.
Consumption and growth
As a political system, democracy does many things but perhaps most importantly it allows those being ruled to condition their rulers. In the context of the economy, this gives people more control to condition what the government does in terms of its consumption. As an economic actor, governments exert important effects on the economy and its growth through the public sector’s consumption. At the macro-economic level (in a closed economy), total output is divided into private consumption, investment, and government consumption, represented by the well-known accounting equation Y=C+I+G. Importantly, government consumption provides the goods and services that are catalytic for growth like education, roads and health care. 4
Yet, government consumption does not always lead to the provision of those goods and services that enhance economic growth. Whether consumption leads to growth depends on the regime in power and whether the regime is accountable to the people being governed or not. In authoritarian regimes it is likely that consumption will be used for private purposes that enrich a single person or small group of people. In Africa, specifically, authoritarian governments often direct consumption toward prestige projects. When Félix Houphouët-Boigny was president of Côte d’Ivoire, he established Yamoussoukro, his birthplace, as the country’s capital and directed resources into very large projects, including building the world’s largest cathedral there. Similarly, the political leader of the Democratic Republic of Congo (DRC) (then Zaire), Mobutu Sese Seko, used government resources to build what has been coined a “jungle paradise” in his remote ancestral hometown of Gbadolite (Smith, 2015). Mobutu’s personal interpreter and translator, Olela Shungu, notes that: “Mobutu developed Gbadolite because he had grown up there” (Yocum, 2014), but in the rest of the country, Shungu notes people complained that “the lavish spending was much less popular with the rest of the population of the DRC.” Under Mobutu’s rule the Congo’s vast natural wealth was not spent on roads, education, and healthcare, but instead on unnecessary airports, private jets, and grand palaces (Hochschild, 1999; Knutsen, 2012). This type of consumption is common in Africa. In the case of Equatorial Guinea, Blas (2014) notes that: “While the government is spending heavily, little is being invested in ways that would foster long-term development.”
In Africa, relatively authoritarian regimes purposefully limit the provision of public goods, often picking projects that are “known losers”, as this is a strategy that provides political and economic gains for the regime and simultaneously keeps the citizens at bay (Robinson and Torvik, 2005: 198). Because the most authoritarian regimes in Africa govern primarily in their personal interest, building “white elephant” or prestige projects like remote airports and other prestigious yet wasteful ventures (Moss, 2007), the effect of government consumption on growth is likely to become increasingly negative under more authoritarian regimes. Not only will this type of spending crowd out private investment but it will lead to higher levels of debt, all of which limits growth. Moreover, it is associated with other political maladies like corruption, that also limits growth. Under the most authoritarian regimes, government consumption will thus be associated with slower growth because these regimes spend the least on growth-enhancing public goods like education, healthcare, and other public services. As a result, people do not have the complements needed to enhance their productivity and promote growth and investors do not have an incentive to finance growth-enhancing projects.
In contrast, democratic regimes are likely to use consumption for public rather than private uses leading to outcomes that are more conducive to growth. Rather than building churches in the middle of nowhere or roads to hometowns, democratic regimes promote education, healthcare, and other services that drive growth. As Knutsen (2013) notes: "Democracy incentivizes leaders to select a range of policies with positive consequences for economic growth." The policies pursued by leaders in democratic regimes manifest themselves in public goods that prove to be conducive to economic growth and development. This encourages growth as people are more productive and investors are more likely to undertake expensive projects when they have the necessary complements so that they can realize a return.
Botswana is an important and well-known example of how democracy conditions consumption in ways favorable to growth. Ayittey (2006b: 513) notes that: "At independence, Botswana’s prospects of surviving as a viable politico-economic entity were just about equal to those of Mali or Burkina Faso: landlocked, plagued by persistent droughts, poor resource wealth, and a small population with a low literacy rate." Yet, Botswana did grow and remains an African success. How did it do so? According to Ayittey (2006b: 514) two critical factors were that Botswana was a multiparty democracy and that it "pursued strikingly prudent economic policies." As a result, Ayittey (2006b: 514–515) argues that: "Revenues from minerals, customs union payments, and donor funds were devoted largely to investment in infrastructure and provide greater access to basic needs: water, health, and primary education." The success of Botswana shows the example of a democratic regime conditioning consumption in ways that proved conducive to growth. Alternatively, if we think of the counterfactual should Botswana have been an authoritarian regime, it is likely that the same consumption would have been directed to purposes private in nature and as a result growth would have been slower on average and more volatile. Accordingly, this discussion leads to the paper’s first hypothesis:
Hypothesis 1: In a comparison of democratic and authoritarian countries, government consumption will be associated with more rapid growth in democratic regimes.
Connecting democracy, consumption, and growth
The primary argument is that government consumption in democratic regimes will be more conducive to growth because it yields public goods rather than private goods. In a democratic regime government consumption might, for instance, be used to build infrastructure like roads and other growth-enhancing inputs, whereas in an authoritarian regime those same types of goods – if built at all – will be directed to private interests. In a joke about building roads, 5 a dictator facing a rebel attack asked Mobutu for advice, to which Mobutu asked if the rebels came by air or sea. The other ruler replies, "No they came by road." "Tsk tsk, my son," Mobutu responds, "Never build roads."
Beyond roads, consumption in democratic regimes is likely to lead to more education than in authoritarian regimes. While the quality of the education might not be better in democratic than authoritarian regimes (Knutsen and Dahlum, 2017), it is clear from research that democracies provide more education than authoritarian regimes including in Africa (Harding and Stasavage, 2014; Stasavage, 2005). In the context of the conditional claim being made in this paper, the argument here is that government consumption will be associated with more education in democratic regimes compared to authoritarian regimes, and this will be catalytic for growth in Africa where increased productivity depends on providing more people with basic skills. This leads to the paper’s second hypothesis:
Hypothesis 2: In a comparison of democratic and authoritarian countries, government consumption will be associated with more education in democratic regimes.
Finally, authoritarian regimes are likely going to condition consumption in ways that limit safety nets and lower levels of welfare spending (Avelino et al., 2005; Kaufman and Segura-Ubiergo, 2001; Rudra, 2002; Rudra and Haggard, 2005). Directing consumption to goods and services that help people live more secure lives is not in the interest of authoritarian rulers as it contradicts their incentive to maximize their private gains, and it can undermine their power and thus future ability to extract gains. In Africa, the state is not only derelict in providing safety nets and welfare spending, but in many cases it actively preys on its citizens making their lives even worse than if it did nothing at all (Ayittey, 2006a; Moss, 2007). Moreover, authoritarian regimes in Africa practice a form of patronage politics in which they purposefully favor certain ethnic groups over others. According to Lewis (2008: 95), “Authoritarian rulers [in Africa] commonly divert state revenues in order to maintain the support bases of their regimes.” To compensate for the state’s failings and purposeful neglect, people find ways to provide for their social insurance including maintaining community networks and having large families. In this context, children act as a form of social insurance that can compensate for the state’s failings. As government consumption increases in authoritarian regimes, people see that the state is not committed to providing goods that will enhance their security and thus they compensate by having more children. In democratic regimes, however, people see government consumption being directed to goods that increase their personal security as well as contributing to the quality of life for their children in terms of health and education (Becker et al., 1990) thus incentivizing them to have fewer children. Accordingly, this discussion leads to the paper’s third hypothesis:
Hypothesis 3: In a comparison of democratic and authoritarian countries, government consumption will be associated with lower fertility in democratic regimes.
Empirical analysis
Estimation equation
Following Mankiw et al.’s (1992) empirical specification of growth, 6 the core estimable equation for growth includes measures for initial income, investment, and human capital. Extending this equation to test the argument, the econometric specification includes measures for democracy, government consumption, and several additional controls. Equation (1) displays the statistical model used in this analysis:
Consistent with research on economic growth (e.g. Burnside and Dollar, 2000; Murdoch and Sandler, 2002), the dependent variable measures the rate of growth, taken as the difference between
Growth and initial income are taken from several sources. To be consistent with existing work in political economy, the first measure of growth and initial income comes from the World Bank’s Africa Development Indicators. While data on growth from the World Bank is common in political economy, there is research indicating that statistical results on the regime-growth question vary depending on whether growth is being measured using the World Bank’s data or the Penn World Tables (Krieckhaus, 2004). To address this concern, the Penn World Tables (Heston et al., 2011) measure of income and growth is also used in this analysis. An important concern in Africa is that economic data drawn from the World Bank, Penn Tables, and other similar sources is not credible due to a lack of capacity to accurately collect data, incentives to misreport, and the existence of a large informal sector that is omitted by official statistics (Jerven, 2013). Following recent research (Magee and Doces, 2015), the final robustness check addresses this issue using satellite images of night-time lights. Growth derived from satellite images of night-time lights is useful because it is less likely to be manipulated by politicians and bureaucrats and it is highly associated with the size of the economy, including both formal and informal sectors. Rich countries, as is widely known, have much brighter satellite images of night-time lights than poor countries. Growth based on night-time light usage is a good indicator of an economy’s size and its expansion, especially in developing countries where the informal sector is so large. Although it is not a perfect indicator of the growth of an economy, and can be manipulated by political leaders when they arbitrarily cut electricity to certain parts of the country as part of their political agenda measuring growth this way does offer an alternative measure to test the robustness of the relationship between democracy and growth in Africa. Ultimately, there is no perfect measure of growth so this analysis uses three different sources to show the robustness of the empirical results.
In equation (1), the Polity score captures the degree of democracy by country with higher/lower values indicating more/less democratic regimes. The Polity measure is one of the most widely used measures in empirical analyses of democracy capturing the degree to which a regime is democratic or authoritarian. Ranging from -10 (most authoritarian) to +10 (most democratic), the Polity measure is calculated based on a point system across several categories including the competitiveness and openness of executive recruitment, constraints on the chief executive, and regulation and competitiveness of political participation (Marshall and Jaggers, 2012). The Polity measure of democracy is shown to be highly correlated with other common measures of democracy, like Freedom House, and for the purposes of this analysis it is the best measure as it is largely driven by the degree of executive constraints and is looked upon as one of the best all-round measures of democracy (Hadenius and Teorell, 2005).
Also included in equation (1) are controls for initial income and other determinants of growth. Initial income is an important control not only because it is an important part of explaining growth (i.e. poor countries should grow faster than richer ones), but also because it helps address the issue of endogeneity. The degree to which regimes are democratic or authoritarian depends on the economy’s level of development (Epstein et al., 2006) and thus its past growth. Because economic development is associated with democracy there is likely an issue of endogeneity between regime type and growth, as Jaunky (2013) confirms in his analysis of Africa. Controlling for initial income at the beginning of each five-year period purges the association between democracy and the error term in equation (1) thus helping to address the endogeneity. In addition, controls for investment and trade openness, both of which are associated with growth, and also the process by which growth can affect democracy are included in the model to further isolate the effect of democracy on growth. To control for human development, a measure of life expectancy is included in the model, helping to address the possibility that growth encourages human capital acquisition and thus democratization. Controls for the amount of aid inflows, if the country is landlocked, and oil production are also included in the model. Finally, period controls are included in the specification with the reference being the initial period from 1960 to 1964.
To estimate the parameters in equation (1), a fixed effects estimator is used which controls for any time-invariant variables that might potentially be confounding factors. In equation (1), landlocked is an example of a variable that does not vary and so using the fixed effects estimator will not produce an estimate for the effect of landlocked on growth due to the fact that it does not vary over time. Other factors not appearing explicitly in the model like former colonizer or majority Muslim or Christian country will be controlled for as long as they are time invariant over each five-year period.
Results
Estimates for the parameters across the three sources of growth are presented in Table 1. The first model shows the estimates using the World Bank data. The estimate for the Polity score is -0.65 and it is significant at or below the 1% level (p < 0.01). This estimate suggests democratic regimes grow slower than authoritarian regimes, however the model includes an interaction term with government consumption such that the negative estimate for the Polity score is true for countries with no government consumption, which is unrealistic. The estimate for government consumption alone is negative but it is insignificant. Importantly, the interaction term between the Polity score and government consumption has a positive estimate. The positive estimate on the interaction term is consistent with the first hypothesis: for democratic countries (i.e. those with a positive polity score), increasing government consumption is associated with more rapid growth, but for authoritarian regimes (i.e. those with a negative polity score), increasing consumption is associated with less growth.
Regression analysis of regime type and growth.
Note: Fixed effects estimator with robust standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01.
Turning to the rest of the controls in model 1, we see that initial income, as expected, is negatively associated with growth indicating that in Africa the relatively rich countries grow slower than relatively poor countries. Investment, as expected, is associated with more rapid growth suggesting that African countries respond to the typical forces of growth. Trade openness, life expectancy, and foreign aid are all positively associated with growth but none of the parameter estimates are statistically significant. Landlocked does not have an estimate as it is time-invariant and thus controlled for by the fixed-effects estimator. Oil production is positively associated with growth but it too is not statistically significant. Finally, compared to the first period (1960–1964), the period controls show that in the late 1970s there was a slowdown in growth that lasted all the way into the first decade of the 21st century. Africa, as the period indicators illustrate, regressed over the latter part of the 20th century.
Building on these estimates, the second column in Table 1 provides the estimates using the Penn World data on growth. The estimate for the Polity score compared to the first model increases in absolute value, and the estimate on the interaction term also increases in magnitude. Similarly, the estimate for initial income also increases in absolute value suggesting that again countries with relatively high starting incomes grow slower in Africa than those with lower incomes. Unlike the first model, the estimate for oil production with the Penn data is positive and significant largely due to a substantial increase in the estimate rather than a change in the standard error. The period controls tell a similar story though some are no longer significant using the Penn data.
To help interpret the estimates on the Polity score, government consumption, and their interaction, Figure 1 models the effect of consumption on growth for authoritarian regimes and democratic regimes. 7 In this case, authoritarian regimes are defined as those with a Polity score of -8 while those that are democratic have a score of 8. The left hand side of Figure 1 shows that at low levels of consumption, authoritarian regimes have positive growth rates but as consumption increases the predicted rate of growth declines. At lower levels of consumption the 95% confidence interval does not include zero growth, though it does at higher levels of consumption. Nevertheless, the trend is clear that consumption in authoritarian regimes is associated with less growth. In contrast, the right panel of Figure 1 shows that for democratic regimes, government consumption is increasingly catalytic for growth. At low levels of consumption, the point estimate for growth is actually negative but with higher levels of consumption it increases and at the highest levels of consumption the estimate is positive and the 95% confidence intervals do not include zero growth.

Effect of government consumption on growth (Penn World Tables) by regime type.
The last model in Table 1 measures growth using satellite images of night-time lights. This provides an important test of the argument as this source of data is different from the World Bank and Penn data sets. The estimate for the Polity score is again negative and significant while the interaction term is also positive and significant. Like initial income, the estimate for initial light usage is negative indicating that countries with more initial light usage grow slower than those with less initial usage. Investment is positively associated with light usage as is oil production, both of which are also statistically significant.
To again help visualize the results, Figure 2 shows the effect of consumption on growth for authoritarian and democratic regimes. In the left panel of Figure 2, authoritarian regimes at low levels of consumption again show positive growth in terms of light usage, but that growth decreases as consumption increases to higher levels. For authoritarian regimes, the 95% confidence interval does not include zero growth at the low and middle levels of consumption, although the intervals do include zero at the higher levels of consumption. For democratic regimes, shown in the right panel of Figure 2, low levels of consumption are associated with negative growth but as consumption increases growth increases and the predicted values becomes positive at higher levels of consumption, although again the 95% interval includes zero growth. Using the lights data, then, it seems that authoritarian regimes reduce growth at low levels of consumption while democratic regimes increase growth at the same low levels of consumption. Again, however, the lights data show that at low levels of consumption authoritarian regimes grow faster than democracies.

Effect of government consumption on growth (lights) by regime type.
Consumption, education, and fertility
Having established that the effect of consumption on growth varies by regime-type, the final part of the empirical analysis tests the second and third hypotheses. Recall, that the argument claims that as consumption increases in democratic regimes it should be associated with more public goods provision like education, and less provision in authoritarian regimes that devote more of the consumption to private goods and away from public goods. In addition, it is claimed that authoritarian regimes devote consumption to private uses that cause people to feel less secure inducing them to compensate in many ways including having more children. Consumption in democratic regimes should cause people to feel more secure and thus have fewer children.
To test these claims the regression model is modified to include education and fertility as the dependent variables. Education is measured as outcomes for primary completion rates (Barro and Lee, 2010). The estimates in the first column of Table 2 show that the estimate for the Polity score is negative but that the interaction with government consumption is positive. For authoritarian regimes increasing consumption is associated with less education, while for democratic regimes it is associated with more education. To visualize this relationship, Figure 3 shows the effect of increasing consumption for authoritarian regimes in the left panel, which indicates a decreasing level of education as consumption increases. While the 95% confidence intervals overlap, the trend is still decreasing which is supportive of the idea that authoritarian regimes increasingly devote consumption away from public purposes. Unlike authoritarian regimes, government consumption in democratic regimes is associated with higher rates of education completion rates by over 10 percentage points from the lowest to highest levels of consumption.
Regression analysis of regime type, education, and fertility.
Note: Fixed effects estimator with robust standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01.

Effect of government consumption on education by regime type.
In considering the effect of consumption on fertility, the second model in Table 2 indicates that there is a positive estimate for the Polity score with a negative estimate on the interaction with government consumption. To display the effect of consumption on fertility by regime, Figure 4 shows that for authoritarian regimes, higher levels of consumption are associated with higher fertility. While the predicted values are increasing, the 95% confidence intervals overlap at all levels of consumption. For democratic regimes, shown on the right side of Figure 4, increasing consumption is associated with lower fertility rates. The estimates are at least suggestive that consumption in authoritarian regimes induces people to take precautions in the form of having more children, while people act in the opposite way in democratic regimes by having fewer children as consumption increases to higher levels.

Effect of government consumption on fertility by regime type.
Together the results provide evidence that democratic regimes grow faster than authoritarian regimes but this appears to be true only at relatively high levels of consumption. While democracy conditions consumption so that the latter is catalytic for growth, it is only at the higher levels of consumption where that effect appears to be positive. At lower levels of consumption, increasing consumption in democratic regimes is associated with less negative growth. In authoritarian regimes, low levels of consumption are associated with faster growth relative to democratic regimes, but as consumption increases growth increasingly slows. Perhaps the most interesting effect revealed in the empirical analysis is how government consumption in democratic regimes is associated with higher rates of education completion rates.
Conclusion
To conclude, I return to the original question being asked: does democracy promote growth in Africa? Or, is authoritarianism better suited to promoting long-run growth as we have seen in China? The answer emerging from this analysis is clear: authoritarianism is not necessarily better suited to promoting growth in Africa and when we account for the likelihood that the effect of regime-type on growth is conditional, there appears to be evidence that democracy is beneficial for growth in Africa. Here, the conditioning factor is government consumption and it is shown that consumption encourages growth in democratic regimes while it increasingly slows down growth in authoritarian regimes. It is important to note, however, that at low levels of consumption authoritarian regimes do grow faster than democratic regimes, but this trend reverses as consumption increases especially to relatively high levels.
In response to these findings, one might respond that the conclusion that democracy is better suited to promoting growth is only true at the highest levels of consumption, which is a level that is unrealistic for most African countries. Moreover, one might further respond that the result in favor of a democracy-growth connection depends on a conditioning factor. This is a fair point if the absence of the conditioning factor can be argued to not be necessary for growth, say membership in an international organization, where one can imagine growth still occurring at a zero value of the conditional factor. However, this point is less compelling in this case, unless we believe that government consumption plays no role in the growth process.
If we accept then that democracy matters for growth due to the way it conditions consumption then these findings call into question claims made about authoritarian-led development in Africa, many of which are based on episodes of rapid growth in countries outside of Africa. In particular, the results here suggest that the so-called China model of development is likely to be inappropriate for countries in Africa. An editorial in The East African (Odindo, 2013: 20) notes that the “ ‘Asian model’ of authoritarian leadership that orchestrates economic growth and prosperity is possible, but that in Africa . . . that model is not sustainable in the long run.” This analysis supports the editorial’s point, especially given the fact that to continue to grow African countries will need their governments to directly provide public goods like roads, education, and healthcare. It is likely that government consumption will play a role in the growth process and it is arguable that it will have to be at relatively high levels in the future, such that democratic regimes will have a growth advantage over authoritarian regimes.
Building on these findings, future research can do several things related to this article and the broader research it builds on. First, the results here need to be further considered with earlier findings regarding state capacity and other possible conditioning factors at play in the democracy-growth relationship. It is possible that consumption and state capacity are related in mutually reinforcing ways, a point which future research can address. Second, as African countries develop it will be important to consider if the relationship between regime type and consumption continues to hold at higher levels of development. The relationship might not hold at higher levels of development as special-interest groups disrupt the connection between democracy, consumption, and growth that is found in this analysis. Finally, research can study not only how growth is affected as African countries develop economically, but what happens to growth as they become advanced democracies. Ultimately, there are numerous questions that still need to be addressed regarding democracy and growth in Africa.
Footnotes
Appendix
Country and Period by Data Set.
| Country | World Bank | Penn World Tables | Lights |
|---|---|---|---|
| Angola | 6-7 | 6-7 | 7 |
| Benin | 1-10 | 1-10 | 7-10 |
| Botswana | 2-10 | 2-10 | 7-10 |
| Burkina Faso | 1-10 | 1-10 | 7-10 |
| Burundi | 1-10 | 1-10 | 7-10 |
| Cameroon | 2-10 | 2-10 | 7-10 |
| Central African Republic | 1-10 | 1-10 | 7-10 |
| Chad | 1-10 | 1-10 | 7-10 |
| Comoros | 5-10 | 5-10 | 7-10 |
| Congo, DRC | 1-10 | 1-10 | 7-10 |
| Congo, Republic | 1-10 | 1-10 | 7-10 |
| Côte d’Ivoire | 1-10 | 1-10 | 7-10 |
| Djibouti | 7-10 | 7-10 | 7-10 |
| Equatorial Guinea | 7-10 | 3-4, 7-10 | |
| Eritrea | 8-10 | 8-10 | 7-10 |
| Ethiopia | 6-10 | 5-10 | 7-10 |
| Gabon | 3-10 | 3-10 | 7-10 |
| The Gambia | 4-10 | 4-10 | 7-10 |
| Ghana | 1-10 | 1-10 | 7-10 |
| Guinea | 7-10 | 6-10 | 7-10 |
| Guinea-Bissau | 3-9 | 3-9 | 7-10 |
| Kenya | 1-10 | 1-10 | 7-10 |
| Lesotho | 2-10 | 2-10 | 7-10 |
| Liberia | 9-10 | 9-10 | 9-10 |
| Madagascar | 1-10 | 1-10 | 7-10 |
| Malawi | 1-10 | 1-10 | 7-10 |
| Mali | 3-10 | 2-10 | 7-10 |
| Mauritania | 1-10 | 1-10 | 7-10 |
| Mauritius | 5-10 | 4-10 | 7-10 |
| Mozambique | 5-10 | 5-10 | 7-10 |
| Namibia | 7-10 | 7-10 | 7-10 |
| Niger | 1-10 | 1-10 | 7-10 |
| Rwanda | 1-10 | 1-10 | 7-10 |
| Senegal | 1-10 | 1-10 | 7-10 |
| Sierra Leone | 5-10 | 5-10 | 7-10 |
| South Africa | 7-10 | 7-10 | 7-10 |
| Sudan | 1-10 | 3-10 | 7-10 |
| Swaziland | 3-10 | 3-10 | 7-10 |
| Tanzania | 7-10 | 7-10 | 7-10 |
| Togo | 1-10 | 1-10 | 7-10 |
| Uganda | 6-10 | 1-10 | 7-10 |
| Zambia | 1-10 | 1-10 | 7-10 |
| Zimbabwe | 4-10 | 7-10 |
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
