Download 2.1 Democracy and state capacity

Africa’s Growth Tragedy Revisited: Weak States, Strong Rulers*
Carl Henrik Knutsen
Department of Political Science, University of Oslo
November 2010
Contact: [email protected]
Abstract
This paper investigates how democracy and dictatorship affect economic growth in SubSaharan Africa, and whether the effect of democracy on growth depends on level of state
capacity. The paper particularly focuses on the adverse economic effects of dictatorial
regimes in countries with weak state institutions. In such contexts, leaders are free to pursue
policies that are macroeconomically inefficient, but which enhance leaders’ survival in office
and increase their personal wealth. The empirical analysis shows that democracy most likely
contributes to higher growth rates in Sub-Saharan Africa, and that democracy has a larger
positive effect on growth in Africa than globally. Moreover, statistical analyses, both on
African and global samples, show that democracy has a particularly positive effect in
countries with weak state institutions. The interaction between weak state capacity and
dictatorship is a vital factor underlying Africa’s many economic development disasters.
Keywords: Democracy, Dictatorship, State capacity, Economic growth, Africa
* Thanks to Daniel Bach, Mike Colaresi, Håvard Hegre, John Huber, Helge Hveem and Asmund Rygh for
valuable comments. Thanks also to participants at the Heidelberg Doctoral Colloquium on October 9-11 2009,
the Centre for the Study of Civil War’s Working Group meeting on October 15-16 2009 and the Columbia
University Political Economy Breakfast on February 9 th 2010 for comments. This paper is mainly written at the
Sciences Po, University of Bordeaux, with generous Mobility Fund support from the GARNET network of
excellence and at the Department of Political Science, Columbia University, with generous support from the
Norwegian Research Council’s Leiv Eiriksson Scholarship.
1
1. Introduction
One common argument is that authoritarianism improves development prospects in poorer
countries with weak state institutions (see Sen 1999). In such countries, authoritarianism is
argued to stabilize polities (e.g. Huntington 1968) and enhance accumulation of capital
investment (e.g. Przeworski & Limongi 1993). Several authors also question the suitability of
democracy in Africa more specifically (e.g. Chabal & Daloz 1999; Lumumba-Kasongo 2005).
However, this paper shows that democracy increases economic growth particularly in
countries with weak state institutions, and democracy is thus particularly conducive to growth
in Africa. Moreover, this paper’s results indicate there is no strong independent effect of state
capacity on growth, contrasting with arguments in several influential contributions (Evans
1995; Evans & Rauch 1999; Leftwich 2000; Fukuyama 2005). Rather, state capacity is vital
for growth only in dictatorships. Dictatorial regimes’ propensity to select “bad policies” is
aggravated in contexts of weak state institutional structures, like post-colonial Africa.
Selecting bad policies is often rational for survival- or consumption-oriented dictators (e.g.
Robinson 1998; Bueno de Mesquita et al. 2003; Acemoglu & Robinson 2006). Democracy
tends to limit the discretionary powers of rulers and channel their survival-oriented behavior
towards economic policies that benefit broader masses of people, and thus the national
economy. This explains why democracy outperforms dictatorship in contexts where there are
few other institutional checks on rulers’ behavior.
Consider Benin and Togo; two relatively similar countries with low state capacity, but which
after Benin’s democratization in 1990 have had different regime types. Both countries are
small, poor West African neighbors, with a relatively similar ethnic fragmentation structure,
French colonial history (although Togo was first colonized by Germany), a post-colonial
history of military rule (although Benin was less politically stable) and even a shared currency
(the CFA). These two countries thus constitute a very good controlled comparison for
investigating democracy’s economic effects in low-capacity states. Benin’s democracy after
1990 has been plagued by deficiencies (Magnusson 2005). However, Benin has had (at least)
partially free and fair elections since 1990 (Lindberg 2006). Benin has experienced
alternations of executive power, and civil liberties are also relatively well protected (Freedom
House 2008a). In Togo, long time ruler Gnassingbe Eyadema and his supporters managed to
block the introduction of democracy in the early 1990s, after initially yielding for pressure to
institute a multi-party system (Bratton & van de Walle 1997). Togolese elections have been
2
far from free and fair. Moreover, the courts are heavily influenced by the regime, freedom of
assembly has not been granted and there has been extensive government control over the
media (Freedom House 2008b). Figure 1 shows Benin’s and Togo’s GDP per capita in
constant 2000 US$ from 1960 to 2008.
Figure 1: Real GDP per capita in Benin and Togo from decolonization to present
Source: World Development Indicators.
According to data from the World Development Indicators (WDI) the picture of divergence is
even clearer when considering PPP-adjusted income: an average Beninese was 30 percent
wealthier than a Togolese in 1990. In 2008 however, he or she was 77 percent wealthier.
Moreover, the gross secondary school enrollment ratio in Benin improved from 9 to 32
percent between 1990 and 2005. The primary enrollment ratio in the same period doubled
from 48 to 96 percent. Moreover, the population-share with access to “improved sanitation
facilities” increased from 12 to 30 percent between 1990 and 2006. In Togo, the equivalent
number sank from 13 to 12 percent. Is Benin’s relative progress due to its more democratic
political system? If yes, does the positive economic effect of democracy hold more generally
in low-capacity states?
Africa’s most impressive economic success stories have been democratic Botswana and
Mauritius. Smart political and economic decision making helped these countries achieve rapid
and sustained growth. Moreover, democratization in several African countries in the early
3
1990s was succeeded by increased growth rates, notably including Ghana. Conversely, few
economies have performed as badly after 1960 as many African dictatorships have. Prominent
examples are Nigeria under various rulers, Zaire under Mobutu, Uganda under Idi Amin,
Ethiopia under Haile-Selassie and Mengistu, Kenya under Arap-Moi and Tanzania under
Nyerere. These regimes differed in several respects, like ideological basis, colonial history
and access to natural resources. But, they were all ruled by authoritarian regimes presiding
over weak-capacity states. A few dictatorships have performed relatively well economically.
The most prominent African example is Apartheid South Africa. South Africa, like its highgrowth dictatorial counterparts in East Asia, differed crucially from most other African
dictatorships on one account: it had a well-functioning bureaucratic apparatus and quite
effective state institutions.
Section 2 presents arguments for why dictatorship is particularly ill-suited for economic
growth in weak-capacity states, with empirical illustrations from Africa. Section 3 presents an
empirical analysis based on panel data techniques. The results are: a) democracy enhances
growth in Subsaharan Africa, a region characterized by low state capacity; b) the effect of
democracy is significantly larger in Africa than in the rest of the world; c) democracy’s effect
on growth is significantly larger in states with weak capacity, both within Africa and globally.
All these results point to a strong interaction effect from regime type and state capacity on
economic growth. Section 4 concludes.
2. The political causes of economic development and stagnation, and
why many African countries are still poor
2.1 Democracy and state capacity
State capacity and regime type are distinct conceptually. Democracy relates to popular control
over political decision making and political equality among citizens (e.g. Beetham 1999).
Free and fair elections, broad participation rights and protection of civil liberties are crucial
institutional requirements for democracy. State capacity relates to state institutions’ ability to
independently implement official policies. The literature on state capacity provides different
definitions (e.g. Skocpol 1985; Migdal 1988; Caporaso & Levine 1992; Fukuyama 2005).
Here, I follow Skocpol, who relates state capacity to a state’s “ability to implement official
goals, especially over the actual or potential opposition of powerful social groups” (1985: 9).
This again implies that independent, rule-following bureaucratic apparatus is an important
4
factor underlying state capacity (e.g. Skocpol 1985; Evans 1995; Chabal & Daloz 1999), and
that personalization and informalization of political decision making goes together with low
state capacity.
Degree of democracy determines who are in charge of political decision making, which again
has implications for the shape of economic institutions and policies. Although earlier studies
often found negative effects of democracy on economic growth (see Przeworski and Limongi
1993), newer studies find either no significant, or a positive significant effect (e.g. Przeworski
et al. 2000; Baum & Lake 2003; Doucouliagos & Ulubasoglu 2008). Most episodes of
economic stagnation and decline globally have come in dictatorial countries (Przeworski et al.
2000), although some dictatorships, particularly in Asia after 1960, have experienced high
growth. Regarding state capacity, several political scientists argue that high capacity is crucial
for economic development (e.g. Evans 1995; Leftwich 2000; Fukuyama 2005). But, few
studies have tested this hypothesis statistically. One exception is Evans and Rauch’s (1999)
seminal study, which finds a positive effect of bureaucracies’ degrees of “Weberianness” on
economic growth. However, this study only uses cross-section data from 35 countries.
2.2 Why dictatorship works poorly when state institutions are weak
The East Asian Tigers were prime examples of high state capacity dictatorships that
experienced high growth (e.g. Wade 1990; Evans 1995; REFERENCE REMOVED).
However, African dictatorships have had poor economic development records. African postcolonial states were, in general, weakly institutionalized, leading to personalized and
informalized political decision making (e.g. Jackson & Rosberg 1982; Clapham 1996; Chabal
& Daloz 1999). Earlier studies of Africa’s “growth tragedy” have focused on geographic and
climatic factors (e.g. Sachs 2001), commodity dependence and decreasing or volatile terms of
trade (e.g. Sindzingre 2007), particular macro- and microeconomic policies (e.g. World Bank
1994) and ethnic fractionalization (Easterly & Levine 1997). However, political-institutional
factors may be at least as important. Indeed, specific institutional structures, including
democratic, can mitigate the negative effects of climatic- and disease factors (Acemoglu et al.
2001) and terms of trade shocks (Rodrik 1999). I argue that democratic institutions, which
were not very common in Africa before 1990, also mitigate the negative effect of weak state
capacity. If state institutions are weak, leading to few horizontal checks on rulers, the more
important becomes vertical accountability mechanisms induced by elections, free speech, free
media and other democratic institutions for restraining leaders’ incentives and abilities to
5
conduct bad economic policies. The combination of weak state institutions and dictatorship is
therefore particularly negative for economic development.
Bueno de Mesquita et al. (2003) argue that broad political support bases, or winning
coalitions, as exist in democracies, are conducive to policies that generate growth-enhancing
public goods, including property rights protection and provision of education to the masses,
rather than narrow redistribution of private goods to a few supporters. Democratic institutions
thereby induce political survival-oriented leaders to provide policies that are good for
economic development (see also e.g. Olson 2003; Acemoglu & Robinson 2006). Botswana
and Mauritius are examples of regimes that invested in broad public goods like infrastructure,
education and health care (Leith 2005; Bräutingham 1997), and these democracies have
grown fast after de-colonization.
Also regimes that have a small winning coalition whose future political and economic
prospects are not entirely in the leader’s hands may produce good economic outcomes
(Besley & Kudamatsu 2007). Such autonomous winning coalitions, for example incorporating
actors in relatively independent state- or party organizations, can pressure dictators to
promote growth enhancing policies, and will do so if they gain from growth. Present day
China, with its strong Communist Party organization and relatively independent and
institutionalized bureaucracy, is a dictatorship with an autonomous winning coalition. In
Africa, the oligarchic Apartheid regime in South Africa was a dictatorship with a relatively
autonomous winning coalition. In personalized, weakly institutionalized dictatorships, such
autonomy is difficult to attain for a winning coalition. In Zaire, for example, Mobutu “kept
ministers and senior officials in a constant state of flux, rotating them regularly, dismissing
them or imprisoning them to ensure they represented no threat” (Meredith 2006: 305). In such
contexts, a dictator is able to push through policies that enrich him or secure his own
continuation in office. The result is often economic stagnation. Personal rule, implying less
autonomous winning coalitions, also exacerbates the “Dictator’s Dilemma” (Mueller 1995:
416-417) on honest reporting of information, as subordinates have more to lose from falling
out with the dictator. More generally, dictators in weakly institutionalized states have poor
apparatuses for eliciting policy-relevant information. Thus, weak state structures are
detrimental to the informational basis on which rulers make policy decisions; even policies
resulting from development-enhancing intentions may fail.
6
Therefore, there are good theoretical reasons for why democracy is beneficial for economic
development in Sub-Saharan Africa, where state capacity has been low and personalized rule
has dominated. Chabal and Daloz (1999) argue that the neo-patrimonial character of regimes
and clientilistic practices are deep-rooted, and will continue to persist in African democracies.
However, even if clientelism is relatively persistent, it is not constant, and importantly,
clientelism’s economic effects may be dissimilar under democracy and dictatorship. First,
democracy may widen the number of clients from a few central players to a broader set of the
populace. This should increase public goods provision. Democracy may also alter the power
balance in patron-client relations. A vote, and other political and civil rights, may provide
clients with an extra tool that allows them to more successfully influence the patron. Although
many African elections have reinstated old dictators, Lindberg (2006) argues that these
elections discipline the behavior even of old dictators. While elections can be manipulated, a
small probability of losing office through elections may induce leaders to promote good
policies (Ferejohn, 1986). Indeed, leaders’ personal characteristics matter far less for
economic outcomes in democracies than in dictatorships (Jones & Olken 2005).
To be more specific, there are several types of “bad policies” that dictators, especially in
weakly institutionalized settings, may conduct. First, dictators may manipulate economic
institutions to their advantage, but to the detriment of the larger economy. For example, poor
property rights protection can be economically and politically beneficial to dictators (e.g.
North 1990; Olson 2003). Incentives for broad-based property protection are often weak in
dictatorships, and if there are few formal-institutional constraints on dictators, they are likely
to grab property. For example, in 1973, Mobutu decided to turn over farms, plantations and
commercial enterprises from foreign owners to Zairian nationals, followed by confiscation of
manufacturing plants. Mobutu benefited personally from these expropriations, but also
enriched his winning coalition on confiscated property (e.g. Wrong 2000: 92-93). In Uganda,
Amin expelled the Asian minority and “shops, the businesses, the property that the Asians
were forced to leave behind, even their personal possessions, were seized as spoils by Amin’s
cronies” (Meredith 2006: 236). Expropriation’s negative macroeconomic effects deter broad
or autonomous winning coalitions with a stake in the economy from expropriating.
Democracies and high-capacity dictatorships thereby provide better property protection.
Additionally, clearly defined public procedures, rules, laws and rights are more likely to exist
in both democracies and high-capacity dictatorships, and this reduces discretionary
expropriation.
7
Second, dictators ruling over weak-capacity states may have incentives to conduct poor fiscal
policy. Available resources for spending on productive public goods are reduced when
dictators’ siphon wealth from the budget for personal consumption and paying their winning
coalitions. This is far easier to do for dictators in weakly institutionalized states, where
political power is tied to person rather than institutional role. In the absence of strong and
independent state institutions, democracy provides an important safeguard to leaders’
appropriation of public resources. Democratic electorates, enlightened by a free press, would
never have approved of Bokassa’s coronation ceremony in The Central African Public, which
cost about a quarter of annual foreign earnings (Clapham 1996: 188) or of Omar Bongo’s 500
million dollar palace in Libreville (Meredith 2006: 381). Such projects mean fewer resources
for growth-enhancing spending on education, health and infrastructure spending. Stasavage
(2005) finds that African democracies spend far more on primary and secondary education
than African dictatorships. Przeworski et al. (2000) argue that in poor states, democracy may
not matter for economic growth through affecting public spending; there is in any case little to
spend. However, the first resources spent often yield the largest marginal return. Basic
mosquito bed-nets save more lives per dollar than expensive heart surgery, the first paved
road has a larger effect on goods-transportation than the second, and so forth. Therefore,
public revenue allocation is perhaps particularly important in poor countries. Dictators’
personal appropriation of resources may not crowd out productive public spending 1:1, but
rather be partly financed through higher taxation. The type of taxation pursued in countries
with low state capacity, where regular monitoring and taxing of income and consumption is
difficult, is extra distortionary. African dictatorships have relied on distortionary trade taxes
and license fees for generating revenues and rewarding political supporters (Ayittey 2005),
thereby increasing the costs of starting new businesses (La Porta et al. 1999).
Third, dictators often face political and economic incentives for conducting inefficient
industrial- and other microeconomic policies. If implementation of economic policy is
conducted by a relatively independent and rule-following bureaucracy, and leaders are
constrained by other state actors, relatively dictatorial countries can experience growthenhancing microeconomic policies (e.g. Wade 1990; Evans 1995; Kohli 2004). However, in
dictatorial countries with weakly institutionalized states, inefficient policies are likely
promoted. Bates (1981) treated these issues extensively. Especially the large agricultural
sector has suffered from inefficient policies in Africa, with the cocoa industry in Ghana being
8
one example. The Cocoa Marketing Board (CMB), set up to regulate the industry and buy
cocoa from farmers, fixed cocoa prices at very low levels. The regime could thus make a
healthy profit when exporting cocoa, whereas farmers received a fraction of world market
prices. Partly because of this, (official) Ghanaian cocoa production was halved from 1965 to
1980 (Meredith 2006: 186). Examples of inefficient industrial policies are plentiful in African
dictatorships (e.g. Ayiettey 2005), and contrast with the productivity-enhancing industrial
policies in high-capacity dictatorships like Taiwan and South Korea (e.g. Wade 1990; Evans
1995). However, the industrial policies promoted by weak-capacity African dictatorships also
contrast with industrial policies promoted in democratic Mauritius (Meisenhelder 1997).
Mauritius’ textile sector grew rapidly from the 1980s onwards, benefiting from reallocated
capital from the sugar industry, openness to foreign ideas and entrepreneurs and favorable tax
and business conditions in the country’s Export Processing Zones. Mauritian industrial policy
seemed tuned towards rewarding productivity enhancement and development of new sectors
rather than rewarding rent seeking behavior by already rich and powerful groups.
Fourth, also monetary and exchange rate policy has been conducted to the benefit of rulers,
but not to the overall economy, in African dictatorships. Printing money generates seignorage
to the regime, and reduces the regime’s real debt through increasing inflation. Printed money
can also be used to selectively reward political backers. However, hyperinflation is disastrous
to an economy; witness Zimbabwe. One would thus not expect extreme mismanagement of
monetary policy in countries with high degree of democracy, but neither would one expect it
in dictatorships with more independent and institutionalized central banks. Exchange rates
have often been set artificially high by African regimes (van de Walle 2005: 31). This
decreases the prices of imported luxury goods, consumed by the elites, but increases export
prices, which hurt local agricultural and manufacturing sectors. Contrast the many artificially
high exchange rates in African dictatorships with democratic Botswana’s balanced exchange
rate policies (Leith 2005: 76-81). However, the perhaps most spectacular monetary policy in
African history was conducted by Equatorial Guinean dictator Macias Nguema who allegedly
kept the entire foreign currency reserve and a large share of the local currency reserve in his
palace, and later in a bamboo hut where some of it rot to the ground (Meredith 2006: 240241).
To sum up, Africa’s dictators have been extremely successful. Several skilled and
strategically clever rulers have managed to become extremely wealthy and stay in power for
9
long. However, there is an inverse relationship between leaders’ personal success and their
countries’ economic success (Bueno de Mesquita et al. 2003). Expropriating property,
partaking in corruption, restricting entry to domestic markets, doling out profitable positions
to supporters (no matter how qualified) and saving in Swiss bank accounts, rather than
investing locally, hurt economic development. But, these activities may be quite rational for
self-interested dictators. If there are no strong, formal institutions to restrain them or an
autonomous winning coalition that desires growth, dictators can get away with such policies.
Thus, the worst combination for economic growth is likely a weak state and a strong ruler.
4. Empirical analysis
4.1 Methodology and data
Below, I empirically investigate both the effect of democracy on growth in Africa, a region
with generally low state capacity, and the interaction between state capacity and regime type,
using both African and global samples. I utilize OLS with Panel Corrected Standard Errors
(PCSE), which takes into account heteroskedasticity and contemporaneous correlation
between panels, and AR1 autocorrelation within panels. However, although these models
include several control variables, there may still be omitted factors that bias results, for
example country-specific historical factors. Therefore, I also utilize Fixed Effects (FE)
models. Other models, for example Arrellano-Bond and 2SLS models, are used for robustness
checks.
I operationalize democracy with the Freedom House Index (FHI), an average of Freedom
House’s Political Rights and Civil Liberties indexes, which tap both the formal existence and
actual functioning of democratic institutions and rights. The FHI has data back to 1972, and
ranges from 1 (most democratic) to 7 (most dictatorial) (Freedom House 2008c). I also use the
Polity Index (PI), which incorporates formal-institutional factors such as competitive
elections, political participation rights and checks on the executive (see Jaggers and Gurr
2002). The PI ranges from -10 (most dictatorial) to 10 (most democratic). The PI, combined
with the other variables, yield time series that go back to 1960 for some countries. The
Pearsson correlation coefficient between the FHI and PI is -.91 globally and -.84 in Africa. As
dependent variable, I use annual growth in real GDP per capita (constant 2000 US$) from
WDI.
10
The models below incorporate several control variables. One is the logarithm of GDP per
capita level, as income may affect both subsequent growth and regime type (e.g Przeworski
2000). Another control is the logarithm of population size, also collected from the WDI. The
logarithm of regime duration, taken from Polity IV, is entered to control for political stability,
which may affect both regime type and growth (see Feng 2003). Since ethnic fractionalization
impact on economic growth (Easterly & Levine 1997), and perhaps regime type, I include the
ethnic fractionalization index from Alesina et al. (1997). I also include dummies for plurality
religion (Sunni Islam, Catholicism, Protestantism and indigenous religions) and colonizer
(British, French, Belgian and Portuguese). The models also incorporate decade dummies to
control for temporal effects; economic growth has been uneven in Africa, with a particularly
hard decade in the 1980s, and democracy has been more prevalent after 1990.
4.2 Democracy’s growth effect in Africa
The number of African electoral democracies has grown over the past two decades, albeit
unevenly (see e.g. Lindberg 2006). The number of countries that scores below or equal to the
middle value of the FHI (3.5) has increased from four in 1972 to eighteen in 2005. Despite
these changes, African democracies have consistently had higher economic growth rates than
African dictatorships. On average, African democracies grew at quite decent rates, also when
compared with countries on other continents. Figure 2 presents the 3-year smoothed average
GDP per capita growth for relatively democratic and relatively dictatorial countries in Africa.
To investigate whether the trends from Figure 2 are due to other factors (or coincidence), I run
regression analyses. The four columns to the left in Table 1 shows results for models based on
African samples, and the most inclusive covers 45 of the 48 Sub-Saharan states (Cape Verde,
The Seychelles and Somalia lack data).
11
Figure 2: Smoothed 3-year average real GDP per capita growth for African democracies
(FHI≤3.5) and dictatorships (FHI>3.5)
Sources: World Development Indicators and Freedom House.
All Africa-based models reported in Table 1 yield an estimated positive effect of democracy.
The FHI is significant at the 0.1%-level in the OLS with PCSE model, and at the 5%-level in
the FE model. According to the OLS with PCSE model, the estimated effect of going from
most dictatorial (7) to most democratic (1) on the FHI is 3.4 percent extra annual GDP per
capita growth. This is about the difference in post-1980 growth rate between Mauritius and
Rwanda, or Botswana and Sudan. The models using the more formal PI do not yield equally
strong results. The estimated effects are positive and quite sizeable, but insignificant.
However, several alternative specifications find a positive, significant PI-coefficient, for
example when applying Random Effects models or when using PPP-adjusted GDP per capita
growth as dependent variable. Regarding the control variables, one OLS with PCSE model
replicates the negative effect of ethnic fractionalization reported in Easterly and Levine
(1997). One FE model also indicates countries grow faster as they become more populous and
richer. Political stability is significant in various models, although the sign varies. Plurality
religion and identity of colonizer generally do not matter for growth.
12
Table 1: Models with linear effect of democracy. GDP per cap. growth as dep. variable1
OLS PCSE
b/(t)
-0.559****
(-3.70)
FHI
African Sample
OLS PCSE
FE
b/(t)
b/(t)
-0.547**
(-2.46)
FE
b/(t)
Africa*FHI
PI
0.042
(1.30)
OLS PCSE
b/(t)
0.221
(1.64)
-0.787****
(-4.12)
0.070
(1.41)
Africa*PI
Africa
Ln GDP pc
Ln Reg. dur.
Ln Pop.
Ethn. Fr.
Catholic
Protestant
Indigenous
Sunni
British
French
Portugese
Belgian
0.258
(0.75)
0.496**
(2.28)
0.253
(1.16)
-3.027**
(-2.09)
0.852
(0.53)
-1.082
(-0.83)
-0.977
(-0.65)
-0.771
(-0.55)
0.007
(0.01)
-0.224
(-0.14)
0.75
(0.42)
-1.958
(-0.85)
Dec60s
Dec70s
Dec80s
Dec90s
Constant
N
1
0.548
(0.58)
-1.147
(-1.25)
-0.723
(-0.86)
-0.006
(-0.00)
1310
0.345
(1.15)
-0.324*
(-1.66)
-0.139
(-0.66)
-1.682
(-1.49)
-0.811
(-0.57)
-2.340**
(-2.07)
-2.434*
(-1.87)
-2.339*
(-1.92)
-2.409
(-1.35)
-2.764
(-1.56)
-2.144
(-1.12)
-4.015*
(-1.82)
1.021
(1.08)
0.509
(0.58)
-1.239
(-1.40)
-0.238
(-0.29)
7.704*
(1.77)
1516
3.063****
(3.48)
-0.041
(-0.17)
6.647***
(2.88)
0.632
(0.83)
-0.614**
(-2.57)
-2.026
(-1.07)
5.443***
(3.04)
2.044*
(1.67)
0.418
(0.57)
-119.970***
(-3.23)
1310
-0.515
(-0.26)
-0.586
(-0.37)
-1.867*
(-1.73)
-0.626
(-0.91)
30.881
(1.01)
1516
4.163****
(3.75)
0.018
(0.09)
0.622****
(4.35)
0.217**
(2.44)
-2.442****
(-4.15)
-0.948*
(-1.75)
-1.616****
(-3.94)
-1.876**
(-2.55)
-1.265***
(-2.97)
-0.323
(-0.75)
-0.324
(-0.63)
0.766
(1.03)
-0.927
(-0.92)
*  p<0.10, **  p<0.05, ***  p<0.01 and ****  p<0.001.
13
0.231
(0.33)
-1.599**
(-2.35)
-1.021
(-1.47)
-1.69
(-0.54)
3984
Global sample
OLS PCSE
FE
b/(t)
b/(t)
-0.018
(-0.13)
-0.801****
(-3.65)
-0.039*
(-1.80)
0.111***
(2.88)
1.351***
(2.75)
0.156
0.667
(0.99)
(1.43)
0.307**
0.897****
(2.49)
(7.01)
0.165
-0.729
(1.59)
(-0.82)
-2.461****
(-5.19)
-1.376***
(-2.87)
-1.883****
(-5.05)
-1.695***
(-2.67)
-1.408****
(-4.06)
-0.624*
(-1.70)
-1.007**
(-2.19)
0.305
(0.49)
-1.480*
(-1.81)
0.946
(1.36)
0.287
0.035
(0.43)
(0.06)
-1.709***
-1.936****
(-2.58)
(-4.38)
-1.006
-1.403****
(-1.48)
(-4.53)
0.139
8.392
(0.05)
(0.53)
4778
3984
FE
b/(t)
-0.038
(-1.40)
0.186****
(4.32)
0.109
(0.31)
0.417****
(3.52)
-1.894***
(-2.98)
-0.585
(-0.91)
-0.922*
(-1.82)
-2.647****
(-6.89)
-1.768****
(-6.11)
31.968***
(2.81)
4778
I conducted several robustness checks.2 First, I ran Random Effects (RE) models, and the
results were quite similar to the OLS with PCSE and FE results for the FHI, and stronger for
the PI, as mentioned above. Second, I tested models using PPP-adjusted GDP per capita
growth (from WDI) as dependent variable. Although PPP-adjusted GDP may be a better
welfare measure as it takes into account local price levels, this measure has less data as its
times series are shorter than those of exchange rate adjusted GDP. However, OLS with PCSE
and RE models show a significant positive effect of democracy at the 5%-level for both the
FHI and PI, whereas the effect was still sizeable and positive, but insignificant in the FE
models.
Third, I tested whether Botswana and Mauritius may drive the results, because of these
democracies’ exceptional economic performance. However, even when excluding these
countries, the FHI models showed a positive, significant effect of democracy at least at the
5%-level. The Polity models also showed positive estimated effects, although these were
insignificant at the 5%-level. Despite this, excluding Botswana and Mauritius from the sample
does not cast serious doubts on the effects of democracy on growth in Africa.
Fourth, Africa has experienced several civil wars in its post-colonial history. Such conflicts
are likely endogenous to regime type (e.g. Hegre et al. 2001); but, prevalence of conflict may
affect both regime type and growth. I therefore tested models including a dummy capturing
civil war, with data from Gates and Strand (2006). I tested models that only controlled for
ongoing conflict, and models that also incorporated dummies capturing post-conflict countryyears (both 3- and 5-year post-conflict periods). Although the democracy indexes’ estimated
effects and t-values were slightly reduced, the positive, significant effect of democracy
survives in models controlling for intrastate conflict. Fifth, there may be geographical and
economic-structural factors that bias the effect of democracy. I therefore ran models that
controlled for landlockedness, absolute latitude, trade as share of GDP, and urban as share of
total population. Adding these controls only strengthened the result. Controlling for energy
production as share of GDP and metal exports as a share of exports from the WDI reduces the
number of units with about ¾. The insignificant effects in these models are therefore not
critical to the above results. In any case, the FE-models pick up country-specific
characteristics, such as the prevalence of diamonds in Sierra Leone and Botswana.
2
Tables with results from the robustness checks are available on request.
14
The results are not robust to specification of time lag. Most models that lagged the
independent variables with between two and five years showed a non-significant effect at the
5%-level, although many coefficients indicate effects of full democratization on growth of
about 1 percent. However, no models show a significant negative effect of democracy on
growth. Therefore, further testing is needed, and I ran simple Granger-tests to see if the
relation between democracy and growth comes from growth affecting democracy. However,
when regressing FHI in t on FHI and economic growth in t-1, the growth variable was
statistically insignificant with a t-value of -0.26. Moreover, when regressing growth in t on
growth in t-1 and FHI in t-1, the FHI coefficient was significant at the 1%-level. These
Granger-tests indicate that the relation between democracy and growth is due to democracy
affecting growth, rather than the other way around.
I further tested for the endogeneity of democracy by running 2SLS-regressions.3 Following
Helliwell (1994) I used historical values of the PI to instrument for the FHI; concretely, I used
a ten-year lagged PI-variable as instrument. The instrument is both correlated with the
possibly endogenous FHI, and does seemingly not violate the exclusion restriction. The
absolute t-value for the instrument in the first-stage is 10.4, showing that past democracy is an
important determinant of present democracy, and the lagged PI does not independently affect
growth: regression models (OLS with PCSE) including the lagged PI alongside the other
independent variables show an insignificant effect on present growth. This means 2SLS
regressions should give consistent estimates of democracy’s effect on growth, even if
democracy is endogenous. The estimated FHI-coefficient is indeed more sizeable in the 2SLS
analysis, and the positive effect of democracy on growth is significant at the 5%-level
(p=0.025). Thus, even if we account for endogeneity, democracy enhances African countries’
growth rates.
As a final robustness check, I ran Arrellano-Bond (AB) dynamic panel data models, which
incorporates lagged economic growth as a regressor in the baseline model above. One
interpretation of the democracy coefficient in such models is the effect of an increase in
degree of democracy in year t on the change in growth rates from year t-1 to t. The results
from the AB models were quite strong. The models incorporating only the one-year lag
3
To be precise, I applied RE2GSLS models.
15
yielded a positive, significant effect at the 1%-level for FHI, and 5%-level for PI. When
including both one- and two-year lags of growth as regressors, the FHI model still showed a
significant effect at the 1%-level, but the PI model yielded an insignificant effect.
Nevertheless, the positive effect of democracy on growth is quite robust.
The results above show a large effect of democracy on growth in Africa and may indicate a
stronger effect of democracy in Africa than globally. The four models to the right in Table 1
investigate the effect of democracy on growth in a global sample. The various models include
a Sub-Saharan Africa dummy and an interaction term that multiplies the dummy with the
democracy measure. The results indicate that the effect of democracy on growth is
significantly (at least 1%-level) more positive in Africa than the rest of the world. No model
finds a significant baseline effect of democracy on growth at the 10%-level, indicating the
global association between democracy and growth, at least after 1960, is mainly driven by
African experiences. Moreover, once political regime and stability, ethnic fractionalization,
colonial history and other variables are controlled for, there is no negative Africa-effect on
growth.
The strong, positive effect of democracy on growth in Africa is, as argued above, likely due to
the weak state institutions in many African countries, which induce dictators to select bad
policies to stay in power or increase personal consumption; democracy is relatively beneficial
to African economies because the alternative is so bad. An earlier study showed no
significant, neither positive nor negative, effect from democracy on growth in Asia
(REFERENCE REMOVED). Asian dictators have faced more autonomous winning
coalitions, for example a strong party organization, and stronger state institutions, which have
forced them to select growth-enhancing economic policies (Besley & Kudamatsu 2007).
Asian dictators have also to a larger extent than African been faced by severe external security
threats, inducing these dictatorships to industrialize and grow to modernize and strengthen
their militaries (REFERENCE REMOVED).
4.3 Democracy, dictatorship and state capacity
I argued that dictatorship was particularly detrimental to growth in Africa because of the
interaction between low state capacity and dictatorship. By the same logic, one should
observe an interaction effect between regime type and state capacity when studying intraAfrican variation. There are few good quantitative measures of state capacity (Bussman
16
2009). Many studies operationalize state capacity with tax income or public spending as share
of GDP. This is problematic, as state capacity is multi-dimensional and incorporates more
than states’ ability to tax (Fukuyama 2005). State capacity, defined as states’ ability to
autonomously design, conduct and implement policies, is related to institutionalization of
political decision making and implementation (Skocpol 1985), which is again tied to rulefollowing behavior of state agents rather than personal and contingent decision making.
A decent proxy for state capacity is thus the Bureaucratic Quality Index (BQI) from ICRG,
where “high points are given to countries where the bureaucracy has the strength and
expertise to govern without drastic changes in policy or interruptions in government
services... [and where] the bureaucracy tends to be somewhat autonomous from political
pressure and to have an established mechanism for recruitment and training” (The PRS Goup
2009). The BQI’s time series start in 1984, and several countries (e.g. Benin and Mauritius)
lack data altogether. The lowest possible score is 0 (assigned to for example Mali and Somalia
for several years), and the highest possible is 4 (assigned to South Africa from 1984 to 1994),
with differentiation allowed on the decimals. There is substantial intra- and between-nation
variation. Cote d’Ivoire for example had its BQI reduced from 3 to 0 during the period,
whereas neighboring Ghana had almost the opposite movement, particularly in the latter years
of Rawlings’ military dictatorship. The BQI and democracy measures are weakly correlated
in Africa. There is almost no correlation between the PI and BQI and the FHI has a
correlation coefficient of -0.13 with the BQI. This corroborates Kohli’s (2004: 401-402)
observation regarding the minimal degree of overlap between state characteristics and regime
type in the developing world. However, in global samples the respective correlation
coefficients are 0.48 and -0.59, likely due to high bureaucratic quality in Western
democracies. I enter interaction terms (democracy measure*BQI) and BQI linearly in the
models from Table 1.
17
Table 2: Interaction regime type and state capacity. GDP per capita as dep. variable
FHI
FHI*BQI
OLS PCSE
b/(t)
-1.743****
(-3.76)
0.777****
(3.29)
PI
PI*BQI
BQI
Ln GDP pc
Ln Reg. dur.
Ln Pop.
Ethn. Frac.
-3.914***
(-3.08)
1.291**
(2.08)
0.201
(0.63)
0.749**
(2.38)
-7.282**
(-2.24)
Catholic
Protestant
Indigenous
Sunni
British
French
Portuguese
Belgian
Dec80s
Dec90s
Constant
N
2.154
(1.36)
2.155*
(1.91)
2.554
(1.64)
-0.202
(-0.12)
-1.085
(-0.50)
2.024
(0.80)
1.179
(0.31)
0.002
(0.00)
-0.769
(-0.83)
-7.035
(-0.81)
634
African sample
OLS PCSE
FE
b/(t)
b/(t)
-1.478***
(-3.12)
0.651**
(2.43)
0.165**
(2.32)
-0.109**
(-2.36)
-0.412
-3.158**
(-0.99)
(-2.13)
0.088
2.031
(0.22)
(1.28)
-0.802**
-0.021
(-2.36)
(-0.06)
0.12
16.576****
(0.50)
(4.11)
-0.121
(-0.04)
2.557
(0.47)
3.365
(0.70)
3.785
(0.68)
3.473
(0.71)
-2.204
(-1.35)
-3.177
(-1.45)
-0.187
(-0.08)
-4.185
(-1.16)
-0.912
5.602***
(-0.92)
(3.11)
-0.657
1.753*
(-0.79)
(1.69)
-273.178****
(-4.15)
578
634
FE
b/(t)
0.158
(1.53)
-0.152**
(-2.32)
-0.13
(-0.27)
-2.271
(-1.35)
-0.990***
(-3.02)
7.170*
(1.91)
1.464
(0.89)
0.204
(0.22)
-98.838
(-1.62)
578
18
OLS PCSE
b/(t)
-0.843***
(-3.16)
0.253**
(2.52)
-1.063**
(-2.16)
0.102
(0.44)
0.355**
(2.13)
0.368****
(4.32)
-1.549***
(-2.80)
-3.108****
(-4.50)
-2.140****
(-3.43)
-3.072**
(-2.53)
-1.933***
(-2.72)
-0.27
(-0.53)
-0.202
(-0.34)
1.623*
(1.73)
-1.518
(-0.66)
-1.333**
(-2.47)
-1.024**
(-2.31)
1.167
(0.49)
2270
Global sample
OLS PCSE
FE
b/(t)
b/(t)
-1.042****
(-4.74)
0.369****
(3.58)
0.084*
(1.73)
-0.018
(-0.80)
0.137
-1.784****
(0.53)
(-3.64)
-0.044
2.467****
(-0.22)
(3.30)
0.003
0.294*
(0.02)
(1.84)
0.187
5.779****
(1.55)
(4.12)
-1.514**
(-2.37)
-2.644****
(-4.32)
-2.186****
(-3.80)
-1.276
(-0.98)
-1.501***
(-2.96)
-0.272
(-0.54)
-0.987
(-1.48)
1.39
(1.35)
-2.842
(-1.08)
-1.404**
0.791
(-2.56)
(1.49)
-0.993**
0.088
(-2.22)
(0.26)
2.013
-107.735****
(0.65)
(-4.42)
2186
2270
FE
b/(t)
0.160***
(3.26)
-0.068***
(-2.87)
0.037
(0.18)
0.449
(0.60)
-0.097
(-0.60)
3.337***
(2.59)
-0.21
(-0.42)
-0.35
(-1.13)
-55.800**
(-2.47)
2186
The results for the interaction models presented in Table 2 are striking. For the models based
on the African sample, the interaction terms are always statistically significant at least at the
5%-level, with the expected sign. Moreover, the linear FHI and PI terms have the expected
signs and are significant at least at the 5%-level in three of four models. The estimates
indicate that dictatorship is detrimental to growth, but only in countries with weak and
moderate levels of state capacity. The linear BQI-term is negative and significant, indicating
no positive growth effect of higher state capacity in democracies. This is surprising given the
strong theoretical arguments and interpretations of case-based evidence, pointing to the
(unconditional) importance of state capacity for economic development (Wade 1990; Evans,
1995; Leftwich 2000; Kohli 2004; Fukuyama 2005). Evans and Rauch’s (1999) results, based
on a smaller cross-country sample mainly consisting of non-democracies, indicated that a
“Weberian” bureaucracy enhances growth. The results above do not directly negate the
importance of state capacity, vitally including an independent and well-functioning
bureaucracy. However, they indicate that state capacity does not have a clear direct effect on
economic growth. Thus, Evans and Rauch’s (1999) focus on bureaucrats’ competence and
incentives may be overstated. Rather strong state institutions and independent bureaucracies
are vital for growth because they constrain dictators’ ability to, in self-interest, generate bad
economic policy.
If the interaction argument on state capacity and regime type is valid, we should observe the
pattern also globally. Table 2 also presents the global results, which confirm the picture from
the African sample. The linear democracy coefficient is significant with expected sign (10%level in one model and 1%-level in three models), and so is the interaction term in three of
four models (at least 5%-level). However, the coefficient-sizes are smaller in the global
sample. The estimates indicate a negative effect of dictatorship among medium-capacity
countries in the global sample, but there are no large growth differences between democracies
and dictatorships among high-capacity countries. Low-capacity dictatorships, however,
perform far worse than low-capacity democracies.
I tested the robustness of the above results, both for the global and African samples. For
example, RE models yielded results similar to those in Table 2, although the interaction term
was insignificant (p=0.16) for the PI-model in the African sample. I also ran models using
PPP-adjusted GDP, and these results were weaker for the African sample; only the OLS with
PCSE model using FHI yielded an interaction term significant at the 5%-level. However, for
19
the global sample the FHI models produced significant interaction terms (5%-level), but PI
models yielded one insignificant interaction term with a p-value of 0.23 (OLS with PCSE),
one significant at the 10%-level (RE) and one significant at the 5%-level (FE). I also tested
the interaction effect in models adding absolute latitude, trade as share of GDP, and urban as
share of total population as control variables. For the African sample, only the OLS with
PCSE models generated significant interaction terms. However, for the global sample, the RE
model using the PI found an interaction term significant at the 10%-level, whereas all other
models produced interaction terms significant at least at the 5%-level.
Finally, I tested Arrellano-Bond models using one and two year lags of the dependent variable
on the right hand side of the regression equation. The PI-interaction terms were insignificant
in these models, but the results for the FHI-interaction terms were strong: for the African
sample, the 1-lag model found a significant interaction effect at the 5%-level and the 2-lag
model at the 1%-level. For the global samples, the interaction terms were significant at the
0.1%-level for both the 1- and 2-lag models. To sum up, the above checks indicate that the
interaction between state capacity and regime type is not completely robust. The results are
somewhat weaker in models using the more formal-institutional PI. However, models based
on the FHI, which measures not only the existence of formal democratic institutions and
rights, but also how they are implemented or subverted in practice, yield strong results.
5. Conclusion
This paper’s argument and results indicate a simple policy implication for actors concerned
with economic development in poor, low-capacity states: Promote democracy! Skeptics have
questioned whether democracy can contribute to economic development in Africa (e.g.
Chabal & Daloz 1999). However, this study finds that African democracies outperform their
dictatorial counterparts in terms of economic growth. Democracy matters for economic
growth, but the effect is likely contingent; the economic benefits are larger in countries with
weak state institutions. Countries with low state capacity need democratic institutions in order
to provide minimum constraints on rulers and to channel rulers’ policies towards public goods
provision and decent economic management. Left unconstrained, dictators in pursuit of
securing their office and pocketing private wealth wreak havoc to their economies. This is a
large part of the reason why many African countries have experienced such miserable
economic fortunes. Luckily, Botswana, Mauritius and even recent developments in for
20
example Ghana indicate that African countries can prosper if they switch to more accountable
forms of government.
In the longer run, state capacity may be endogenous to regime type (Charron & Lapuente,
2010). For example, dictators may have an interest in building down state institutions’
capacity gradually (REFERENCE REMOVED), thus exacerbating the negative effects of
dictatorship over time. The literature on African political economies suggests that rulers may
have strong interests in “informalizing politics” (e.g. Chabal & Daloz 1999). The dynamics of
the regime type-state capacity-economic growth nexus thus remains to be explored in further
detail.
21
Literature
Acemoglu, D. & Robinson, J.A. 2006. Economic Backwardness in Political Perspective.
American Political Science Review 100(1): 115-31.
Acemoglu, D., Johnson, S. & Robinson, J.A. 2001. The Colonial Origins of Comparative
Development: An Empirical Investigation. American Economic Review 91(5): 1369-1401.
Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S. & Wacziarg, R. 2003.
Fractionalization. Journal of Economic Growth 8(2): 155-94.
Ayittey, G.B.N. 2005. Africa Unchained: The Blueprint for Africa’s Future. New York:
Palgrave-MacMillan.
Baum, M.A. & Lake, D.A. 2003. The Political Economy of Growth: Democracy and Human
Capital. American Journal of Political Science 47(2): 333-47.
Bates, R.H. 1981. Markets and States in Tropical Africa: The Political Basis of Agricultural
Policies. Berkeley: The University of California Press.
Beetham, D. 1999. Democracy and Human Rights. Cambridge: Polity Press.
Besley, T. & Kudamatsu, M. 2007. Making Autocracy Work. Working Paper.
Bratton, M. & van de Walle, N. 1997. Democratic Experiments in Africa. Regime Transitions
in Comparative Perspective. Cambridge: Cambridge University Press.
Bräutigam, D. 1997. Institutions, Economic Reform and Democratic Consolidation in
Mauritius. Comparative Politics 30(1): 45-62.
Bueno de Mesquita, B., Smith, A., Siverson, R.M. & Morrow, J.D. 2003. The Logic of
Political Survival. Cambridge, MA: MIT Press.
Bussman, M. 2009. Are strong and big governments good for peace? The effects of state
spending and revenues on civil war. Paper presented at 2009 ECPR Conference in Potsdam,
September 10-12.
Caporaso, J.A. & Levine, D.P. 1992. Theories of Political Economy. Cambridge: Cambridge
University Press.
Chabal, P. & Daloz, J.P. 1999. Africa Works: Disorder as Political Instrument. Bloomington:
Indiana University Press.
Charron, N. & Lapuente, V. 2010. Does democracy produce quality of government?
European Journal of Political Research 49(4): 443-70.
Clapham, C. 1996. Africa and the International System –The Politics of State Survival.
Cambridge: Cambridge University Press.
22
Doucouliagos, H. & Ulubasoglu, M.A. 2008. Democracy and Economic Growth: A MetaAnalysis. American Journal of Political Science 52(1): 61-83.
Easterly, W. & Levine, R. 1997. Africa’s Growth Tragedy: Policies and Ethnic Divisions. The
Quarterly Journal of Economics 112(4): 1203-50.
Englebert, P. 2000. State Legitimacy and Development in Africa. London: Lynne Rier.
Evans, P.B. 1995. Embedded Autonomy: States & Industrial Transformation. Princeton:
Princeton University Press.
Evans, P.B. & Rauch, J.E. 1999. Bureaucracy and Growth: A Cross-National Analysis of the
Effects of "Weberian" State Structures on Economic Growth. American Sociological Review
60(5): 748-65.
Feng, Y. 2003. Democracy, Governance, and Economic Performance: Theory and Evidence.
Cambridge, MA: MIT Press.
Ferejohn J. 1986. Incumbent Performance and Electoral Control. Public Choice 50(1-3): 5-25.
Freedom House. 2008a. Country Report: Benin. www.freedomhouse.org
Freedom House. 2008b. Country Report: Togo. www.freedomhouse.org
Freedom House. 2008c. Methodology. www.freedomhouse.org
Fukuyama, F. 2005. State Building – Governance and World Order in the Twenty-First
Century. London: Profile Books.
Gates, S. & Strand, H.. 2006. Modeling the Duration of Civil Wars: Measurement and
Estimation Issues. Working Paper.
Hegre, H., Ellingsen, T. Gates, S. & Gleditsch, N.P. 2001. Toward a Democratic Civil Peace?
Democracy, Political Change and Civil War, 1816-1992. American Political Science Review
95(1): 33-48.
Helliwel, J.F. 1994. Empirical Linkages between Democracy and Economic Growth. British
Journal of Political Science 24(2): 225-48.
Herbst, J. 1989. The Creation and Maintenance of National Boundaries in Africa.
International Organisation 43(4): 673-92.
Jackson, R.H. & Rosberg, C.G. 1982. Why Africa's Weak States Persist: The Empirical and
the Juridical in Statehood. World Politics 35(1): 1-24.
Jones, B.F. & Olken, B.A. 2005. Do Leaders Matter? National Leadership and Growth since
World War II. The Quarterly Journal of Economics 120(3): 835-64.
Kohli, A. 2004. State-Directed Development: Political Power and Industrialization in the
Global Periphery. New York: Cambridge University Press.
23
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. & Vishny, R.W. 1999. The Quality of
Government. Journal of Law, Economics, and Organization 15(1): 222–79.
Leith, J.C. 2005. Why Botswana Prospered. Montreal: McGill-Queen’s University Press.
Lindberg, S.I. 2006. Democracy and Elections in Africa. Baltimore: The Johns Hopkins
University Press.
Lumumba-Kasongo, T., ed. 2005. Liberal Democracy and its Critics in Africa. Dakar:
CODESRIA Books.
Mamdani, M. 1996. Citizen and Subject:Contemporary Africa and the Legacy of Late
Colonialism. Princeton: Princeton university Press.
Marshall, M.G. & Jaggers, K.. 2002. Polity IV Project – Dataset Users’ Manual. College
Park: Program Center for International Development and Conflict Management, University of
Maryland. Manual.
Meisenhelder, T. 1997. The Developmental State in Mauritius. The Journal of Modern
African Studies 35(2): 279-97.
Meredith, M. 2006. The State of Africa – A History of Fifty Years of Independence. London:
The Free Press.
Migdal, J.S. 1988. Strong Societies and Weak States. Princeton: Princeton University Press.
Miquel, G.P. 2007. The Control of Politicians in Divided Societies: The Politics of Fear.
Review of Economic Studies 74(4): 1253-74.
Mueller, D.C. 1995. Public Choice III. Cambridge: Cambridge University Press.
North, D.C. 1990. Institutions, Institutional Change and Economic Performance. New York:
Cambridge University Press.
Olson, M. 2003. Dictatorship, Democracy and Development. In: Stephen Knack, ed.,
Democracy, Governance and Growth. Ann Arbor: The University of Michigan Press.
Przeworski, A. & Limongi, F. 1993. Political Regimes and Economic Growth. The Journal of
Economic Perspectives 7(3): 51-69.
Przeworski A., Alvarez, M.E., Cheibub, J.A. & Limongi, F. 2000. Democracy and
Development: Political Institutions and Well-being in the World, 1950-1990. New York:
Cambridge University Press.
Robinson, J.A. 1998. Theories of “Bad Policy”. Policy reform 1(1):1-46.
Rodrik, D. 1999. The New Global Economy and Developing Countries: Making Openness
Work. Baltimore: The Johns Hopkins University Press.
Sachs, J.D. 2001. Tropical Underdevelopment. Cambridge MA.: National Bureau of
Economic Research. NBER Working Paper no. 8119.
24
Sindzingre, A. (2007). Financing the Developmental State: Tax and Revenue Issues.
Development Policy Review 25(5): 615-32.
Skocpol, T. 1985. Bringing the State Back In. In: P.B. Evans, D. Rueschemeyer & T.
Skocpol, eds., Bringing the State Back In. Cambridge: Cambridge University Press.
Stasavage, D. 2005. Democracy and Education Spending in Africa. American Journal of
Political Science 49(2): 343-58.
The PRS Group. 2009. International Country Risk Guide: Researcher Dataset.
van de Walle, N. 2005. Economic Reforms: Patterns and Constraints. In: E. Gyimah-Boadi,
ed., Democratic Reform in Africa: The Quality of Progress. Boulder: Lynne Rienner.
Wade, R. 1990. Governing the Market: Economic Theory and the Role of Government in East
Asian Industrialization. Princeton: Princeton University Press.
World Bank. 1994. Adjustment in Africa: Reforms, Results and the Road Ahead. Oxford:
Oxford University Press.
Wrong, M. 2000. In the Footsteps of Mr Kurtz. London: Fourth Estate.
25