Oil,domestic conflict,and opportunities for democratization

Abstract

The resource curse literature suggests two stylized facts about oil-producing states (‘petrostates’) that are not easily reconciled with each other. On one hand, petrostates experience more frequent civil wars than non-petrostates. On the other hand, petrostates have more robust and long-lasting autocratic regimes. This is puzzling because one might expect that one form of instability would lead to the other, as is typical in non-petrostates. If petrostates are more prone to domestic conflict than non-petrostates, and conflicts are opportunities for regime transition and democratization, why do we not observe such transitions more frequently in petrostates? I argue that despite frequent conflicts, rebels rarely succeed in violently overthrowing a petrostate regime or otherwise forcing regime transition. This is because oil generates financial resources that can be used by both an incumbent government and rebels to fund armed conflict, and an incumbent government typically has greater access to these resources. In an analysis of non-democracies for 1946–2004, I also find that oil inhibits democratization in petrostates, but only in the context of violent domestic conflicts. Peaceful pathways to democracy remain open in petrostates. These findings significantly alter our understanding of resource curse. Many scholars argue that oil inhibits democracy because of rentier politics, but this standard interpretation is incomplete. Oil appears to inhibit democratization only in the context of violent domestic conflicts. Ten of the eleven transitions to democracy in petrostates since 1945 occurred without significant domestic conflict.

Keywords

civil war democratization oil resource curse

Introduction

The resource curse literature suggests two stylized facts about oil-producing states that are not easily reconciled with each other. On one hand, oil-producing states (‘petrostates’) experience more frequent civil wars – or more broadly, violent domestic conflicts – than non-petrostates (Ross, 2004a,b, 2012; Fearon & Laitin, 2003; Collier & Hoeffler, 2004; Smith, 2008; Humphreys, 2005; Buhaug, Gates & Lujala, 2009; Lujala, 2010; for a dissenting view see Smith, 2004). On the other hand, petrostates are widely believed to have more robust autocratic regimes than non-petrostates, in the sense of long regime duration and a low rate of democratization (Ross, 2001, 2012; Andersen & Aslaksen, 2013; Bellin, 2004; Dunning, 2008; Morrison, 2009; Jensen & Wantchekon, 2004; Goldberg, Wibbels & Mvukiyehe, 2008; for dissenting views, see Haber & Menaldo, 2011; Herb, 1999). Thus existing research suggests that oil generates stability in one sense (i.e. persistent autocratic regimes) while simultaneously generating instability in another sense (i.e. frequent violent rebellions).

This is puzzling because it seems reasonable to expect that one form of instability would lead to the other. As this article demonstrates, a regime transition is generally more likely in the wake of violent domestic conflict. A regime transition is defined as a significant change in the political structures and practices of a state which affects the powers and/or the selection process of the executive leader. Regime transition is a necessary but not sufficient condition for democratization. Violent domestic conflict is neither a necessary nor sufficient condition for regime transition or democratization, but it raises the probability of both, at least in non-petrostates. From the American Revolution to the struggles in the Philippines (1986), Guatemala (1986), or Sierra Leone (2000), political violence often forms the crucible in which democracy is born. Yet if petrostates are more prone to domestic conflict than non-petrostates, and conflicts are opportunities for regime transition and democratization, why do we not observe such changes more frequently in petrostates?

This puzzle has gone largely unnoticed in part because scholars have generated two largely independent bodies of research on oil politics. The links between oil and democracy on one hand, and oil and violent domestic conflict on the other hand, have long been investigated. The interaction between the two phenomena has yet to receive sustained attention.

In this article, I propose and test an explanation for the puzzle. I argue that despite frequent conflicts, rebels rarely succeed in violently overthrowing a petrostate regime or otherwise forcing regime transition. This is because oil generates financial resources that can be used by both rebels and an incumbent government to fund armed conflict, and an incumbent government typically has greater access to these resources, which confers a significant advantage over rebel challengers. The military incumbency advantage closes one path to democracy. Still, democratization is possible through peaceful political methods. As this article explains, two outcomes are typical for petrostates: one in which violent rebellion is frequent but ineffective at toppling an autocracy, or one in which there is little violent domestic conflict and democratization is possible.

This argument is significant for two reasons. First, it is the first article to systematically investigate the outcome of petrostates’ domestic conflicts, in terms of regime transition, as compared to those in non-petrostates. Second, it has significant consequences for understanding how oil inhibits democratization. I find that oil does inhibit democratization in petrostates, but only in the context of violent domestic conflicts. Oil does not inhibit democratization in situations without violence. Thus the evidence supports a unified explanation of two of the best-known characteristics of petrostates: propensity for domestic conflicts and authoritarianism.

If correct, these findings significantly alter our understanding of resource curse. Many scholars argue that oil inhibits democracy because of rentier politics: oil income is not generated by taxing the domestic population, so it weakens the domestic accountability of the government and makes political representation less likely (Ross, 2001, 2012; Bellin, 2004; Goldberg, Wibbels & Mvukiyehe, 2008; Morrison, 2009). I argue that this standard interpretation is incomplete. Oil only inhibits democratization in the context of violent domestic conflicts. On the (few) occasions that a regime transition occurs in a petrostate without violent domestic conflict, democratization is common.¹ Ten of the 11 transitions to democracy in petrostates since 1945 occurred without significant domestic conflict.

The article is structured as follows. The first section describes the puzzle, drawing on the existing literature and some new empirical data. The second section develops the theory and hypotheses. The third outlines the empirical methodology and data. Section four presents the results, which show strong support for the theory: petrostates are less likely than non-petrostates to experience either a regime transition or democratization during a time of significant domestic conflict, but democratization is relatively common in petrostates during times without such conflict. The final section concludes.

Puzzle

Definitions

For the purposes of this analysis, a petrostate is any state that generates rents (profit) from its oil and natural gas production exceeding $100 per capita in a given year.² I intentionally use a relatively low threshold, which generates a set of 1,313 observations of petrostate-years across 58 states that produced oil in at least one year in the period 1946–2004. I use the dichotomous measure because of its simplicity and conceptual clarity. As part of the robustness checks, I use continuous measures of oil income, as well as other dichotomous measures using alternative thresholds.³ The choice of operationalization does not substantively affect the results.

The two dependent variables of this study are regime transitions and democratic transitions. Following Cheibub, Gandhi & Vreeland (2009), I consider a regime transition to have occurred whenever the state transitions

Table I.

Probability of regime transition and democratic transition, 1946–2004

	Regime overthrow (%)	Democratic transition (%)	Obs.
All non-democratic states
Domestic conflict	9.29	3.78	872
No domestic conflict	3.95	1.79	3,520
…in non-petrostates only
Domestic conflict	10.98	4.62	692
No domestic conflict	4.43	1.83	2,845
…in petrostates only
Domestic conflict	2.78	0.56	180
No domestic conflict	1.78	1.48	674

The unit of analysis is all non-democratic state-years (e.g. Iraq, 1980). Differences between rows are statistically significant at p < 0.01 (except in ‘petrostates only’ subsample).

Sources: ACLP data (Cheibub, Gandhi & Vreeland, 2009); UCDP/PRIO Armed Conflict Dataset (Gleditsch et al., 2002).

between six major regime types: parliamentary democracy; semi-presidential democracy; presidential democracy; civilian dictatorship; military dictatorship; royal monarchy. Thus regime transitions can be autocratic-democratic, democratic-autocratic, autocratic-autocratic or democratic-democratic. A regime transition does not require a change in leadership, although it often happens. I use the term ‘democratic transitions’ to indicate only the subset of regime transitions from autocracy to democracy. The empirical operationalization is defined more precisely later.

Phenomenon to be explained and possibility of selection bias

As indicated in the introduction, regime transitions generally – and democratization specifically – are more likely in the wake of violent domestic conflict, at least in non-petrostates. Table I shows the average rate of regime transition (column 1) and democratic transition (column 2) per state-year among all non-democratic states. The top row shows the observations in which there was violent domestic conflict in either the year of the observation or the preceding year. The bottom row shows observations where there was no such violence. For observations with domestic conflict, the average rate of regime transition was 9.3%, more than twice as high as for observations in which there was no conflict (4.0%). Similarly, the rate of democratic transitions following recent domestic conflict is more than twice as high as when there is no domestic conflict. Clearly, violent domestic conflicts represent opportunities for regime transition and democratization.

Table I also shows the rates of regime transition and democratization in two subsamples. The pattern among non-petrostates is consistent with the overall pattern. Among petrostates, however, the pattern is quite different: democratization is actually less likely during times of recent domestic conflict, which is the opposite of the tendency in non-petrostates. This result highlights the central puzzle of the article: if conflicts are opportunities for regime transition and democratization, why do we not observe such changes more frequently in petrostates?

This puzzle is more striking in light of recent findings that oil income typically prolongs political leadership duration (Colgan, 2010, 2013b), but that effect depends on whether the state is non-democratic (Andersen & Aslaksen, 2013). In most developing countries, long leadership duration often implies an absence of regime turnover and democratization. Oil thus appears to bolster the robustness of autocratic regimes, but the conditions under which those regimes break down are left unexplained.

This article does not focus on what causes the onset of conflict. Identifying the causes of the onset of domestic conflict is a large subject that falls beyond the scope of this article. There is already much valuable work on that subject, and I simply seek to build on it (Cederman, Weidmann & Gleditsch, 2011; Fearon & Laitin, 2003; Collier & Hoeffler, 2004). Because violent domestic conflict is not randomly distributed among countries, one might worry that this approach creates a selection bias. Specifically, there might be underlying conditions that make it more likely for a country both to experience the onset of conflict and to democratize. For example, a country with weak security institutions might be more likely to experience domestic conflict (because law and order cannot be kept), and more likely to democratize (because the incumbent autocracy can be toppled relatively easily). If one then looked at a sample of states that experienced a lot of conflict (e.g. petrostates), a selection bias would make it more likely such states would democratize. Yet we observe the opposite result among petrostates: democratization is less likely during times of domestic conflict, not more (see Table I).

Thus at least this form of selection bias is not a problem. Even if it exists, the bias would run opposite to the phenomenon being explored in this article. The effect of oil is actually reversing the pattern, thereby overriding a plausible selection effect. Additional factors are needed to explain the observed pattern. Before proposing such an explanation, I consider first the existing literature on the resource curse.

Existing literature on oil and domestic conflict

There is significant evidence that oil makes domestic conflicts and civil war more likely (Fearon & Laitin, 2003; Collier & Hoeffler, 2004; Ross, 2004a,b; Le Billon, 2005; Buhaug, Gates & Lujala, 2009; Lujala, 2010). Fearon & Laitin (2003: 85) found that countries that derive at least one-third of their export revenues from fossil fuels face twice the risk of civil war onset compared with those countries that do not have such exports. A key feature of the existing research on domestic conflict is that it focuses almost entirely on how an oil economy alters the incentives and funding opportunities of the rebels, and says little about the incumbent government’s funding and incentives.⁴ Also, there remains some debate about the causal mechanisms linking oil to violent conflict.

The hypothesized mechanisms can be grouped into two basic categories, grievance and funding. The grievance hypothesis is that the political economy of a petrostate creates especially strong grievances among the population, causing them to rebel. Such grievances can stem from inequality, economic decline, forced migration, land expropriation, agricultural degradation, and/or environmental hazards associated with the oil industry (Karl, 1997; Humphreys, 2005; Wantchekon, 2002; Klare, 2002). The funding hypothesis suggests that oil (and possibly other natural resources) provides greater funding opportunities to rebels than those available in non-petrostates (Collier & Hoeffler, 2004; Ross, 2004a,b; Le Billon, 2005). In part, this is because the oil industry is geographically fixed by the oil reserves, whereas other industries (e.g. manufacturing) can often simply relocate if they are being harassed by local rebels (Ross, 2004b). In some cases, foreign actors may actually fund or assist the rebels directly, to form relationships that will pay off once the rebels win (Ross, 2004b). As Fearon & Laitin (2003: 81) point out, the grievance and funding mechanisms are not necessarily competing alternatives.

The finding that oil makes domestic conflicts more probable sits uneasily with research that suggests that oil and other non-tax revenue tends to make regimes more stable, in the sense of regime duration (Morrison, 2009; Smith, 2004). In recent work, Morrison (2010) seeks to resolve this apparent contradiction. He argues that while oil has a destabilizing effect on weak state capacity, oil has a stabilizing effect on states with strong state capacity. His argument suggests that the political effect of oil depends on the existing domestic politics of the state. One limitation of Morrison’s approach, however, is that it treats state capacity as strictly exogenous to oil, whereas other research suggests that the political economy of oil alters the capacity of the state, thus making state capacity endogenous to the question (Karl, 1997; Ross, 2012; Jensen & Wantchekon, 2004; Dunning, 2008).⁵ A more robust alternative is needed.

Existing literature on oil and democracy

Quite separately from the literature on domestic conflict, many studies have investigated oil and democratization. Early studies developed the concept of the rentier state and the way in which oil corrupts domestic politics (Mahdavy, 1970; Karl, 1997). In a seminal article, Ross (2001) argued that ‘oil hinders democracy’, and many subsequent studies reached a similar conclusion (Jensen & Wantchekon, 2004; Bellin, 2004; Ulfelder, 2007; Gassebner, Lamla & Vreeland, 2008), though some disagree (Haber & Menaldo, 2011). Others argued that oil has a conditional effect (Herb, 1999; Dunning, 2008; Andersen & Aslaksen, 2013).

In an important recent article, Haber & Menaldo (2011) challenged the notion that oil inhibits democratization. In turn, their findings have been challenged by others, such as Andersen & Ross (2014). Without seeking to resolve that debate, I simply note that the majority of research that has looked at this question concludes that oil does inhibit democratization. I take that finding as the conventional wisdom, which creates the puzzle highlighted in the introduction.

Even among those who agree that oil hinders democracy, there is considerable disagreement about the causal mechanisms. Ross originally suggested three mechanisms, but in more recent work (2012) he finds evidence to support only one of the three: the rentier effect, which suggests that resource-rich governments use low tax rates, high public spending, and patronage to relieve pressures for democratic accountability. The rentier effect is the most widely cited causal mechanism (Bellin, 2004; Ulfelder, 2007; Goldberg, Wibbels & Mvukiyehe, 2008; Morrison, 2009; Hertog, 2010), but others have suggested that oil’s effect operates through foreign intervention (Bellin, 2004), capital immobility (Boix, 2003), or corruption (Fish, 2005: 133). The proliferation of hypothesized micro-mechanisms linking oil to authoritarianism generates considerable uncertainty about the precise nature of the causal relationship.

Thus there is an ongoing debate about the causal mechanisms through which oil affects regime type. Moreover, the links between oil and democracy on the one hand, and oil and domestic conflict on the other, have been investigated largely separately by scholars. Thus far, the literature has not considered how the interaction between the two phenomena might shed light on the causal mechanisms linking oil to authoritarianism.

Theory

I theorize that oil income, while used by both rebels and incumbent governments to fund armed conflict, favors the incumbents and makes it especially difficult for rebels to force regime transition through violent confrontation. This closes one path to democracy, but democratization is still possible through peaceful political methods. This section explains, first, why rebel victory is unlikely in a petrostate; second, why they fight anyway; and third, how democratization is possible in petrostates without recourse to prolonged rebel violence.

My argument focuses on the role of oil rather than other natural resources or other sources of non-tax income, such as foreign aid (Morrison, 2009). I do this for two reasons: petroleum is enormously important on its own, and analytically I wish to avoid the potential problems of unit heterogeneity that would come from including other forms of non-tax income in the analysis (Andersen & Aslaksen, 2013). For instance, foreign aid may not be fully analogous to oil income because donors providing aid sometimes apply political conditions and constraints along with the money, which oil reserves do not. In the conclusion, I discuss how my research could be extended to these topics.

I use the simplifying assumption that a regime transition during a time of violent domestic conflict represents a rebel victory. Of course, not all rebels aim at regime transition: some rebels have more modest goals (e.g. a shift in policy or revenue distribution). Still, the converse assumption seems reasonable: a regime transition during a time of violent domestic conflict typically represents a rebel victory. This could happen in different ways: sometimes regime transition occurs because rebels overthrow the incumbent and establish a new regime; other times rebels leave the incumbent in place but force significant regime changes.

Oil lowers the probability of rebel victory

Much scholarly attention has focused on how the resource curse creates incentives for rebels to fight their governments, but less attention has been paid to how oil (or other resources) affects the behavior of an incumbent government.⁶ Oil provides the government with significant income that is not derived from constituent taxes, which can be diverted to fight rebels relatively easily, in comparison to the funds available to non-petrostates. The fact that oil income is a non-tax source of revenue means that domestic populations are less likely to hold governments accountable for how it is spent, in part because of the significant information asymmetries involved.⁷ This gives the government an independent source of financial resources, which in turn generates political and military resources. There is evidence that petrostate governments spend more on military arms and personnel than non-petrostates, and that increases in arms expenditures are linked to oil booms (Ross, 2001; Chan, 1980; Colgan, 2011).

Consequently, when a petrostate regime is threatened by rebels, it can shift resources to meet the threat. How the regime responds will depend on the situation and on the regime’s preferences, leading to a variety of strategies. In Nigeria, authorities have tried to use amnesty offers that include cash and job opportunities to purchase peace in the Niger Delta. Faced with widespread protests in 2009, Iranian officials used the Iranian Revolutionary Guard and various state-sponsored militia groups to repress political activists. In Angola, the government simply outspent the rebels in weapons purchases and military expenditure. In Libya’s civil war in 2011, Gadaffi used oil money to employ thousands of African mercenaries to fight against domestic rebel forces (though Gadaffi ultimately lost, in part because of a NATO military intervention).

The extra resources available to the government in a petrostate increase its probability of victory against the rebels as compared to the counterfactual in which the government did not have those resources, all else equal. However, in a petrostate, the rebels also have increased funding due to oil, as discussed earlier. Strictly speaking, the net effect of oil on the outcome of the domestic conflict could go either way: it could help the government more than the rebels or vice versa. Yet in most cases, a government receives far more oil income than rebels do. Moreover, during times of peace, an incumbent government has almost exclusive access to the country’s oil income, allowing it to build up financial, political, and military reserves.⁸ The government’s oil income is typically so much larger than the rebels’ share that the relative balance of power favors the incumbent government. For example, rebel groups in Colombia extracted an estimated $140 million annually from the oil industry in the 1990s, but the government earned an average of $3.4 billion in oil profits after costs.⁹ Existing research provides significant empirical evidence of the link between government oil revenues and military expenditures (Ross, 2001; Chan, 1980; Colgan, 2011).

Of course, petrostate governments sometimes do lose power due to rebellions: the Shah of Iran in 1979 and King Faisal of Iraq in 1958 are examples. Gadaffi in 2011 is another example, though it is notable that the NATO intervention played a significant role in sustaining the Libyan rebellion. Petrostate regimes appear to be particularly vulnerable to transition at times of leadership change, or when an incumbent leader’s health is failing, often because the leader has eliminated the very political institutions that might provide stability in those times. Still, most of the time oil provides the government with the resources needed to quell political dissent. In sum, oil income gives petrostate regimes more resources to address potential rebellions than non-petrostate regimes, which in turn leads to a higher probability of victory in a domestic conflict.

Why do rebels in petrostates fight if the probability of victory is low?

One might ask why rebels in petrostates are willing to engage in violent conflict if they can anticipate ex ante that the probability of full victory (i.e. regime transition) is relatively low. I suggest two answers to this question.¹⁰ The first is that some rebel groups (in both petrostates and non-petrostates) have goals that fall short of regime transition. Even if rebels have regime transition as a nominal goal, they could be satisfied with more intermediate objectives such as a larger share of government expenditures or modest increases in political participation. Rebels might choose to fight if there is a reasonable probability of achieving these intermediate goals.

The second answer is that it could be in the rebels’ material interests to fight their government even without any expectation of achieving even intermediate changes in the regime. Being a rebel is a paying job. Being a rebel might be an individual’s best economic option, especially in a country where other job opportunities are scarce or poorly paid. Of course, being a rebel is dangerous and illegal, so individuals might demand a wage premium to compensate for the risks. Still, the oil money available by theft or extortion could make rebellion attractive. In an extreme case, a rebel group’s public goal of political change might simply serve as a façade for a criminal gang.

Systematic evidence about the financial rewards of fighting in a rebellion or domestic conflict is very difficult to obtain. Nonetheless, in his analysis of 13 resource-related conflicts, Ross finds that looting played a significant role in at least ten of them (Ross, 2004a).¹¹ For instance, two rebel groups in Colombia, the ELN and FARC, extorted by various means an estimated $140 million annually from the oil industry in the late 1990s (Dunning & Wirpsa, 2004). How this total income translates into rebel wages is unclear, but it is enough to provide each rebel roughly $7,000–14,000 in gross income, in a country in which GDP per capita was $2,340.¹² The rebels’ oil revenue is in addition to other sources of revenue (like the illicit drug trade), which increases the financial incentive for a rebellion. Similarly, studies of the ongoing conflict in the Niger Delta suggest that rebels are ‘commanding monthly salaries of over N50,000 [$320 USD] – well above the wage that can be plausibly commanded by an educated youth in the formal sector’ (Watts, 2007: 640). It is thus plausible that rebellions provide financial rewards that cannot be easily matched by anything in the regular economy of many petrostates.

Is democratization possible in petrostates without violent conflict?

Thus far, I have argued that petrostate rebels have a lower probability of succeeding in forcing a regime transition than rebels in non-petrostates. As Table I demonstrates, domestic conflict in non-petrostates can lead to regime overthrow, which in turn can lead to democratization, but this pathway to democracy is mostly shut down in petrostates. Yet, in contrast to Ross and others, I argue that peaceful democratization is quite possible in petrostates.¹³

Even in a petrostate, repression comes at a cost: it hurts the state’s non-oil economy. When choosing the degree of political repression to impose, an autocratic regime must manage the trade-off between openness (to allow its non-oil economy to flourish) and tight control (to minimize the risk of political transition). For simplicity, I treat this as a dichotomous choice between (i) a totalitarian regime that is tightly repressive but economically stagnant and (ii) a limited autocracy with greater economic benefits but a real risk of political transition. Historically, many petrostate governments have chosen the former type, especially in the Middle East. In such cases, the opposition is divided, weakened, and/or forced underground by repression, regime transitions occur only due to extraordinary crises or gross mismanagement by the government, and even then democratization is almost impossible.

If the government chooses a limited autocracy, however, political opposition can form and operate relatively openly. Consider the limited autocracy as characterized by four actors: a leader, two members of a governing coalition that supports the leader, and an opposition actor. The leader stays in power by buying the support of the members of the governing coalition, in part by using oil rents. At times of political discontent, each member of the governing coalition has a choice. S/he can choose to support the leader in the hopes of continuing to receive oil rents from the regime, but s/he does so at the risk that the other member(s) of the governing coalition defects to the opposition, collapsing the regime and leaving her/him with nothing. Alternatively, the members of the governing coalition could choose to remove the leader and accept a pacted transition in which the oil rents are shared with the (relatively strong) opposition coalition (O’Donnell, Schmitter & Whitehead, 1986; Linz & Stepan, 1996). Crucially, key members of the military must regard the transition as politically legitimate, thereby rendering violent repression of the transition impossible. The nature of the new pacted democracy ensures that the old governing coalition retains a sufficient portion of the oil rents to make the transition attractive. The outcome of the transition for the individual leader, however, is indeterminate and can vary widely from case to case.

Some brief historical examples help illustrate the argument. In 1958, Venezuelan political elites ousted the dictator Perez Jimenez and transitioned toward a democracy structured by an explicit agreement between the three main political parties (Karl, 1997). The Punto Fijo Pact ensured that each of the parties would respect the outcome of elections but include members of the other parties in positions of power. Implicit in this pact was the notion that oil revenues would be used to fund government expenditures that benefited constituencies favored by each of the major parties (Dunning, 2008). With an agreement between elites in place, the coup against Perez Jimenez was relatively bloodless. In Nigeria, a transition to democracy was pushed forward in 1979 by the National Movement which brought together northern/Muslim and southern/Christian interest groups. The first elected president shared power with a broad coalition based on the National Movement. Obasanjo, the displaced military leader of Nigeria, was not punished and was later elected president after the country’s second democratic transition in 1999. Mexico finished a slow transition to democracy in 2000 when the ruling party (PRI) was defeated electorally. As Magaloni (2006) shows in the case of Mexico, a pacted democratic transition can leave the incumbent regime in power for a long time if the opposition cannot solve its collective action problem, even when the opposition is quite strong in the aggregate. Note that the outcome for the displaced autocratic leader varies widely in these cases: exile in Venezuela; political retirement in Mexico; and in Nigeria, an eventual return to politics. However, in all three cases the power-sharing and coalition politics led to oil-funded government expenditures being disbursed to a wide coalition.

In this account, the strength of the opposition and the degree of government repression are important exogenous factors, the origins of which lie beyond the scope of the article. Instead, this article simply predicts that violent domestic conflict and enduring autocracy tend to go hand-in-hand in petrostates: either both exist or neither does. Still, some conjectures are possible about the conditions under which the opposition is strong. Research suggests that opposition is weak when it cannot solve its collective action problem, and when it has difficulty recruiting supporters from all sections of society (Magaloni, 2006). It seems plausible that these challenges will decrease when elites are relatively homogenous and have significant exposure to democratic norms. In homogenous states, the opposition can focus political divisions on the extent of support for democracy, and work to persuade even the incumbent’s supporters. In more heterogeneous states, society is either fractionalized (making political unity within the opposition more difficult) or support for the regime is organized along ethnic, religious, or other cleavages. In the latter case, opposition might be unified but there is a clear additional cleavage between them and government supporters beyond the question of democracy, making it difficult to create the kind of a broad coalition needed for democracy. For example, many Latin American petrostates had experience with democracy prior to the influx of major oil incomes, and have religiously and socially homogenous elites. Significantly, most of the democratic transitions in petrostates have occurred in Latin America. Inequality tends to make the national political elite a relatively small, tight-knit group; religious homogeneity makes it easier for the opposition to solve its collective action problem. By contrast, in the Middle East and Africa, there is little historical experience with democracy, and opposition elites are often divided by religion or tribe. The online appendix also provides some (limited) quantitative data suggesting that democratization in petrostates is more likely when religious fractionalization is low and inequality is high, consistent with the idea of a more homogenous national elite.

What are the observable implications?

The hypothesis that petrostates experience frequent violent intrastate conflict has already been well tested in the existing literature (Fearon & Laitin, 2003; Collier & Hoeffler, 2004; Ross, 2004b). My work generates new testable hypotheses:

H1: At all times, the probability of a regime transition is lower in petrostates than in non-petrostates.

H2a: During periods of violent domestic conflict, the probability of a democratic transition is lower in petrostates than in non-petrostates.

H2b: During periods without violent domestic conflict, the probability of a democratic transition is not lower in petrostates than in non-petrostates.

Methodology

The empirical analysis in this article follows many of the methodological conventions used by Ross, and uses oil income data generously supplied by him (Ross, 2008). The analysis focuses on all non-democratic states during the period 1946–2004, the broadest period over which data are consistently available. The online appendix provides descriptive statistics.

The first independent variable, Petrostate, is measured dichotomously using the $100 per capita threshold described earlier. To measure Domestic conflict, I use the UCDP/PRIO Armed Conflict Dataset measure of intrastate violence defined as a use of armed force between two parties, of which at least one is the government of a state, that results in at least 25 battle-related deaths (Gleditsch et al., 2002). The relatively low threshold for violence used by UCDP is helpful in identifying a reasonably comprehensive set of cases in which violence might have generated a democratic transition. As a robustness check, I also used the higher threshold used by UCDP to identify civil wars (at least 1,000 battle-related deaths), and the results of the analysis were consistent.¹⁴

The analysis uses two dependent variables. The dependent variable is measured using the ACLP data developed by Przeworski et al. (2000) and updated by Cheibub, Gandhi & Vreeland (2009). To test H1, the dependent variable is Regime transition, a dichotomous variable coded as 1 during the year of a regime transition (as indicated when the ‘regime age’ variable in the ACLP dataset is equal to 1), and 0 in all other years. To test H2a and H2b, the dependent variable is Democratic transition, a dichotomous variable coded as 1 during the year of a transition to democracy, and 0 in all other non-democratic years. Observations of Democratic transition are not used (i.e. are coded as ‘missing’) in years in which a state is already a democracy. In the dataset, there are 291 cases of Regime transition, of which 95 are transitions from autocracy to democracy, 63 are transitions from democracy to autocracy, and the rest are within-type regime transitions (e.g. one autocracy replaces another autocracy). Consistent with these dichotomous measures of the dependent variables, the analysis uses a random-effects panel logit regression.

The dichotomous measure of democratic transitions is used in preference to Polity scores, another common measure of regime type, for two reasons. First, the dichotomous measure allows the analysis to focus more precisely on the transition to democracy, rather than any other changes in regime type. If Polity scores are used, a change from an absolute autocracy to a hybrid autocracy (for instance, –10 to –3 on the Polity scale) is quantitatively equivalent to a change from a hybrid regime to a democracy (+3 to +10 on the Polity scale), but these changes are not conceptually equivalent. Although in principle one could build a Polity-based dichotomous measure that focuses only on democratic transitions, in doing so one loses much of the rationale for using Polity in the first place (i.e. it is a continuous measure). Second, there are significant problems with using Polity scores to operationalize the concept of ‘regime transition’ (Colgan, 2012, 2013a). However, as a robustness check, I test the results using a new measure of democratic transitions generated by Boix, Miller & Rosato (2012).¹⁵

Several control variables were used. First is the variable Income, which measures the natural log of GDP per capita. Prior studies of democratization suggest that income is an important factor: when incomes rise, so does the likelihood that a state will be democratic (Londregan & Poole, 1996). This variable is measured using data from Fearon & Laitin (2003). Missing observations were filled in using data from the World Development Indicators.¹⁶

A second control variable is Economic growth, which measures the year-to-year change in a country’s income per capita. According to several studies, economic growth helps autocracies survive (Przeworski et al., 2000; Gassebner, Lamla & Vreeland, 2008). Third, the variable Previous transitions measures the number of previous democratic transitions a country has undergone since 1946. This variable accounts for a country’s history of regime transitions, as several studies suggest that when states have prior experience with democracy, a subsequent transition to democracy is more likely (Gassebner, Lamla & Vreeland, 2008). (For the regressions testing H1, the variable Previous transitions measures the number of any regime transitions since 1946, not just those that led to democracy.) Fourth, a dichotomous variable, Monarchy, indicates whether the existing regime is a monarchy, as this type of autocracy is often believed to be especially resistant to political change (Herb, 1999).¹⁷ Fifth, the variable Muslim represents the Muslim fraction of the population, using data drawn from Barrett (1982). The inclusion of this variable is not meant to suggest that democracy and Islam are incompatible. However, some scholars have suggested that the historical legacies in some countries with large Muslim populations could inhibit democracy, and this variable simply seeks to control for any such statistical correlation. Following Ross (2001), the statistical regressions also include a series of period dummies – one for each five-year period, beginning in 1946 – to control for temporal patterns and contemporaneous shocks, such as the end of the Cold War.

Some studies of democratic transitions have included country fixed effects in their statistical models to control for unobserved country-specific factors, and to focus on within-country variations over time rather than cross-country variations (Haber & Menaldo, 2011). However, as Michael Ross (2008: 7) writes, ‘these tasks are desirable, [but] the decision to include country fixed effects in a logit model creates an even larger problem: it eliminates from the sample all countries that have never transited to democracy, a group that includes most of the oil-rich developing countries. The result is a severe selection bias: countries that have transited to democracy remain in the sample, while countries that have not transited are dropped.’¹⁸ A fixed-effects analysis on ACLP data means that Algeria, Angola, Azerbaijan, Bahrain, Brunei, Cameroon, Egypt, Gabon, Equatorial Guinea, Iran, Iraq, Kazakhstan, Kuwait, Libya, Malaysia, Oman, Qatar, Russia, Saudi Arabia, Syria, Tajikistan, Turkmenistan, Tunisia, Uzbekistan, the United Arab Emirates, Vietnam, and Yemen are all dropped from the

Table II.

Regression analysis of regime transition, 1946–2004

	Model A	Model B	Model C	Model D
DV: Regime transition	Base	Interaction	Violent conflict	No violence
Petrostate	–0.946** (0.298)	–0.794* (0.365)	–1.919** (0.666)	–0.783* (0.367)
Domestic conflict		1.118** (0.207)
Conflict * Petrostate		–0.865 (0.604)
GDP per capita (log)	0.193* (0.097)	0.326** (0.111)	0.480* (0.232)	0.263† (0.135)
Economic growth	–0.042** (0.009)	–0.038** (0.010)	–0.043** (0.015)	–0.031* (0.014)
Previous transitions	0.227** (0.044)	0.128* (0.061)	–0.093 (0.119)	0.313** (0.065)
Muslim, %pop.	–0.056 (0.228)	–0.208 (0.270)	–0.392 (0.522)	–0.054 (0.312)
Monarchy	–0.871* (0.371)	–0.861* (0.399)	–0.045 (0.692)	–1.172* (0.509)
N	4,030	4,030	819	3,211
No. of states	126	126	123	80
Log-likelihood	–738	–722	–475	–229

All models use random-effects logit regression analysis for panel data. Non-democratic states only. All independent variables lagged by one year. Panel-adjusted standard errors are in parentheses; †p < 0.10, *p < 0.05, **p < 0.01. Dummy variables for each five-year period starting in 1946 are included but not shown.

analysis. The hypotheses in this article cannot be meaningfully tested in this way.

Results

Table II shows the statistical relationship between oil and Regime transition. Each column shows the results of a random-effects panel logit regression analysis. All independent variables are lagged by one year to reduce the potential for endogeneity. One striking feature of these models is that Petrostate is consistently negative and statistically significant, suggesting that it inhibits regime transition. This is true even when the Domestic conflict variable and its interaction term (which is just the product of the two variables, Conflict×Petrostate) are introduced. The coefficient for Conflict×Petrostate is not statistically significant, suggesting that oil income has no additional impact (positive or negative) on the probability of regime transition during times of domestic conflict. Other control variables behave as expected: for example, economically growing countries are associated with a lower probability of regime transition, whereas countries with recent previous regime transitions are more likely to experience them again than those states that have not had a recent transition.

To further illustrate this result, two additional analyses are conducted on the subset of observations in which there was (Model C) and was not (Model D) recent domestic conflict. Again, the models show that Petrostate is negatively and significantly associated with the probability of regime transition. These results are consistent with the notion that oil inhibits regime transition at all times, whether there is violent conflict or not, because the incumbent government has more resources with which to purchase political support and disrupt opposition. Thus the results support H1.

Table III shifts the dependent variable to Democratic transition, but the methodology is otherwise analogous. Model 1 generates a familiar result: oil is negatively associated with the probability of a democratic transition. Note that its statistical significance is somewhat weaker than is generally reported in other studies of the resource curse, but its democracy-inhibiting effect is clearly demonstrated in Models 2 and 3.¹⁹ The effects of the control variables are also familiar: the coefficients for Income and Previous transitions are positive and significant, while the coefficients for Economic growth and Monarchy are negative and significant, as expected. The Muslim percentage of the population is negative and weakly significant. Thus rich, non-monarchic autocracies that have low or negative economic growth and previous experience with democracy are likely to transition to democracies, especially when the state has zero or low oil income. These findings are consistent with much of the existing literature.

Model 1 confirms the finding that oil inhibits democratic transitions, yet a number of unanswered questions remain about exactly how oil has this effect. H2a posits that the democracy-inhibiting effect of oil is principally due to the low probability of regime

Table III.

Regression analysis of democratic transitions, 1946–2004

	Model 1	Model 2	Model 3	Model 4
DV: Democratic transition	Base	Interaction	Violent conflict	No violence
Petrostate	–0.709† (0.399)	–0.220 (0.446)	–3.208** (1.230)	–0.261 (0.441)
Domestic conflict		1.071** (0.320)
Conflict * Petrostate		–2.479* (1.152)
GDP per capita (log)	0.443* (0.174)	0.605** (0.192)	0.959** (0.360)	0.370† (0.207)
Economic growth	–0.042** (0.015)	–0.038* (0.016)	–0.050* (0.023)	–0.023 (0.023)
Previous transitions	0.685** (0.202)	0.464† (0.241)	–0.026 (0.365)	1.100** (0.186)
Muslim, %pop.	–0.784† (0.439)	–0.985† (0.505)	–0.263 (0.753)	–1.119* (0.558)
Monarchy	–2.446* (1.059)	–2.763* (1.114)	–21.236 (25813)	–2.042† (1.073)
N	4,030	4,030	819	3,211
No. of states	126	126	80	123
Log-likelihood	–372	–365	–111	–235

transition in petrostates during periods of violent domestic conflict. Model 2 introduces the variable Domestic conflict and its interaction term (Conflict×Petrostate).²⁰ The coefficient for Petrostate is no longer statistically significant under this specification. This supports H2b, and suggests that the standard view of the resource curse is incomplete.²¹ Instead, the democracy-inhibiting effect of oil appears in the coefficient of the interaction term Conflict×Petrostate, which is negative and significant. Domestic conflict itself is positive and strongly significant, as expected: such conflict is a proxy indicator of an opportunity for political change. In petrostates, however, such opportunities do not typically result in democratization, because rebels usually fail to force regime transition through violent campaigns. Post-estimation calculations suggest that the rate of democratization among petrostates which have experienced recent conflict is 0.004 per year compared with 0.014 for those without recent conflict, a difference which is statistically significant.²² Also, looking only at countries with recent domestic conflict, the rate of democratization of petrostates is significantly different from non-petrostates.²³ The raw data on the number of democratic transitions for each type of state is reported in the online appendix.

Two additional regression models support the finding that oil only inhibits democratization during times of violent domestic conflict. Among the observations in which there was recent domestic violence (Model 3), oil has a powerful negative effect on the probability of democratic transition, consistent with the expectation of H2a. Among the observations in which there was no recent domestic conflict (Model 4), however, the size of the coefficient for the oil income variable diminishes considerably, and it is no longer statistically significant at standard levels of confidence. Thus the evidence supports H2a and H2b.

The magnitude of these effects is large. Overall, the probability of democratic transition in petrostates is roughly half what it is in non-petrostates (1.3% vs. 2.4%).²⁴ The effect is especially pronounced when there is domestic violence (0.6% vs. 4.6%, see Table I). The online appendix (Table A2) shows the 11 cases of democratic transitions in oil-producing states since 1945, and there is only one instance of a petrostate transitioning to democracy following significant civil violence (Romania, which was only marginally a petrostate).²⁵ By contrast, non-petrostates regularly transition to democracy as a result of a violent domestic clash: there were 32 such instances in the time period under analysis. When there is no violent domestic conflict in the state, however, the probability of democratic transition is about the same in petrostates and non-petrostates (1.5% vs. 1.8%, a difference which is not statistically significant).

The key difference in the results between Table III and Table II is that oil negatively inhibits regime transition (see Table II) at all times, but inhibits democratic transition (see Table III) only when there is domestic conflict. This is again consistent with H2b: peaceful regime transitions are uncommon in petrostates, but when they occur they are likely to lead to democratic transitions. Indeed, peaceful regime transitions in petrostates led to democracy in ten of 13 cases (77%), a rate that is significantly higher than the rate of democratization in non-petrostates (56 out of 126 cases, or 42%). However, caution is warranted when drawing inferences from this comparison due to the small sample size. Hypothesis H2b is more modest: it claims only that the probability of democratization under nonviolent conditions is not lower (and possibly higher) in petrostates than in non-petrostates.

Robustness checks

The empirical results were subject to a battery of robustness tests. First, all models were re-analyzed using the linear measure of oil income, as well as the dichotomous measures using alternative thresholds described earlier. Second, the time period of the analysis was restricted to the period 1960–2002, to facilitate comparisons to the analysis conducted by Ross (2001). Third, the Previous transitions variable was replaced by a dichotomous measure indicating whether the state had experienced any prior democratic period since 1946. Fourth, various additional control variables were inserted into the regressions, including a measure of ethnolinguistic fractionalization, indicators for OECD states or Cold War years, various types of autocratic regimes, and the sociolegal institutional legacy from previous colonial or socialist masters.²⁶ None of these changes made a substantive difference to the results discussed above.²⁷

Regional dummy variables were also inserted into the analysis, following the World Bank regional classification system. Interestingly, the coefficient for Petrostate was no longer statistically significant for Model 1 in Table III, suggesting that the standard ‘resource curse’ result is somewhat fragile. (Indeed, other robustness tests of Model 1 also suggest the fragility of the standard ‘resource curse’ result; see online appendix.) However, the results in all of the other models in Table III do not change substantively when the regional control variables are included. Thus the principal empirical results shown here are robust.

Finally, the democratization hypotheses were tested using a survival analysis based on a Cox proportional hazard model. Again, the results were consistent with the main findings and are shown in the online appendix.

Conclusion

This article identifies a major new fact about politics in petrostates: the democracy-inhibiting effect of oil only applies in the context of domestic conflict. Democratization does happen in petrostates, but it generally occurs only when the opposition operates nonviolently. Fighting for democracy or regime transition does not work in petrostates. Despite frequent rebellions, autocratic petrostate regimes are very rarely compelled through violence to alter the regime. Oil generates conditions conducive to violent conflict, while at the same time making it unlikely that the incumbent regime will undergo a transition. My argument resolves the apparent contradiction in the literature by showing how oil generates stability on one level (regime continuity) while simultaneously generating instability on another (violent rebellions). Thus this article contributes to the ongoing investigation of the causal mechanism linking oil to authoritarianism.

One opportunity for future research is to extend the argument presented here to other natural resources (e.g. diamonds, illicit drugs, etc.) and other forms of non-tax income such as foreign aid. This article focuses on oil because the sheer importance of oil to the modern global economy makes it the logical starting point of any such inquiry. The global trade of oil and natural gas generates revenues that are somewhere between ten and 100 times larger than the next largest natural resource, and 20 times larger than the combined foreign aid of all OECD states.²⁸ Still, in some countries, foreign aid or other resources play a large role in the local economy. It would be interesting to test whether the logic described in this article also applies in those cases.²⁹

This article offers hope for democratization in petrostates. Opposition groups should think seriously about the potential for a nonviolent process. The weight of that decision is not theirs alone, however: the incumbent government must be responsive. Still, when violent conflict is not occurring, opposition groups in petrostates have as much chance of transitioning their country towards democracy as those in non-petrostates.

Footnotes

Replication data

The online appendix, dataset, and do-files for the empirical analysis in this article can be found at .

Acknowledgments

I wish to thank Sarah Bush, Inken von Borzyskowski, Graeme Blair, Johannes Urpelainen, three anonymous reviewers, and participants at the EPSA 2012 meeting in Berlin for their helpful comments on early drafts of this article.

Notes

References

Andersen

Jørgen Juel

Aslaksen

Silje

(2013) Oil and political survival. Journal of Development Economics 100 (1): 89–106.

Andersen

Jørgen Juel

Ross

Michael

(2014) The big oil change: A closer look at the Haber-Menaldo data. Comparative Political Studies 47 (7): 993–1021.

Barrett

David

(1982) World Christian Encyclopedia. New York: Oxford University Press.

Bellin

Eva

(2004) The robustness of authoritarianism in the Middle East. Comparative Politics 36 (2): 139–157.

Boix

Carles

(2003) Democracy and Redistribution. Cambridge: Cambridge University Press.

Boix

Carles

Miller

Michael

Rosato

Sebastian

(2012) A complete data set of political regimes, 1800–2007. Comparative Political Studies 56 (12): 1523–1554.

Buhaug

Halvard

Gates

Scott

Lujala

Päivi

(2009) Geography, rebel capability, and the duration of civil conflict. Journal of Conflict Resolution 53 (4): 544–569.

Cederman

Lars-Erik

Weidmann

Nils

Gleditsch

Kristian Skrede

(2011) Horizontal inequalities and ethno-nationalist civil war: A global comparison. American Political Science Review 105 (3): 478–495.

Chan

Steve

(1980) The consequences of expensive oil on arms transfers. Journal of Peace Research 17 (3): 235–246.

10.

Cheibub

José Antonio

Gandhi

Jennifer

Vreeland

James Raymond

(2009) Democracy and dictatorship revisited. Public Choice 143 (1–2): 1–35.

11.

Colgan

Jeff D

(2010) Oil and revolutionary governments: Fuel for international conflict. International Organization 64 (4): 661–694.

12.

Colgan

Jeff D

(2011) Venezuela and military expenditure data. Journal of Peace Research 48 (4): 547–556.

13.

Colgan

Jeff D

(2012) Measuring revolution. Conflict Management and Peace Science 29 (4): 444–467.

14.

Colgan

Jeff D

(2013a) Domestic revolutionary leaders and international conflict. World Politics 65 (4): 656–690.

15.

Colgan

Jeff D

(2013b) Petro-Aggression: When Oil Causes War. Cambridge: Cambridge University Press.

16.

Collier

Paul

Hoeffler

Anke

(2004) Greed and grievance in civil war. Oxford Economic Papers 56 (4): 563–595.

17.

Dunning

Thad

(2008) Crude Democracy: Natural Resource Wealth and Political Regimes. New York: Cambridge University Press.

18.

Dunning

Thad

Wirpsa

Lisa

(2004) Oil and the political economy of conflict in Colombia and beyond: A linkages approach. Geopolitics 9 (1): 81–108.

19.

Fearon

James

Laitin

David

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97 (1): 75–90.

20.

Fish

M Steven

(2005) Democracy Derailed in Russia: The Failure of Open Politics. Cambridge: Cambridge University Press.

21.

Gassebner

Martin

Lamla

Michael

Vreeland

James Raymond

(2008) Extreme bounds of democracy. Journal of Conflict Resolution 57 (2): 171–197.

22.

Gleditsch

Nils Petter

Wallensteen

Peter

Eriksson

Mikael

Sollenberg

Margareta

Strand

Håvard

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39 (5): 615–637.

23.

Goldberg

Ellis

Wibbels

Erik

Mvukiyehe

Eric

(2008) Lessons from strange cases: Democracy, development, and the resource curse in the US states. Comparative Political Studies 41 (4–5): 477–514.

24.

Haber

Stephen

Menaldo

Victor

(2011) Do natural resources fuel authoritarianism? A reappraisal of the resource curse. American Political Science Review 105 (1): 1–26.

25.

Herb

Michael

(1999) All in the Family: Absolutism, Revolution, and Democracy in the Middle Eastern Monarchies. Albany, NY: State University of New York Press.

26.

Hertog

Steffen

(2010) Princes, Brokers, and Bureaucrats. Ithaca, NY: Cornell University Press.

27.

Humphreys

Macartan

(2005) Natural resources, conflict, and conflict resolution: Uncovering the mechanisms. Journal of Conflict Resolution 49 (4): 508–537.

28.

Jensen

Nathan

Wantchekon

Leonard

(2004) Resource wealth and political regimes in Africa. Comparative Political Studies 37 (7): 816–841.

29.

Karl

Terry Lynn

(1997) The Paradox of Plenty. Berkeley, CA: University of California.

30.

Klare

Michael

(2002) Resource Wars: The New Landscape of Global Conflict. New York: Holt.

31.

Le Billon

Philippe

(2005) Fuelling War: Natural Resources and Armed Conflict. Abington, IL: Routledge for the International Institute for Strategic Studies.

32.

Linz

Juan

Stepan

Alfred

(1996) Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore, MA: Johns Hopkins University Press.

33.

Londregan

John

Poole

Keith

(1996) Does high income promote democracy? World Politics 49 (1): 1–30.

34.

Lujala

Päivi

(2010) The spoils of nature: Armed civil conflict and rebel access to natural resources. Journal of Peace Research 47 (1): 15–28.

35.

Magaloni

Beatriz

(2006) Voting for Autocracy: Hegemonic Party Survival and Its Demise in Mexico. New York: Cambridge University Press.

36.

Mahdavy

Hussein

(1970) The patterns and problems of economic development in rentier states: The case of Iran. In: Cook

Michael A

(ed.) Studies in Economic History of the Middle East: From the Rise of Islam to the Present Day. Oxford: Oxford University Press, 428–467.

37.

Morrison

Kevin

(2009) Oil, non-tax revenue, and the redistributional foundations of regime stability. International Organization 63 (1): 107–138.

38.

Morrison

Kevin

(2010) Oil, conflict, and stability. Working paper (http://politics.as.nyu.edu/docs/IO/14563/Morrison.pdf).

39.

O’Donnell

Guillermo

Schmitter

Philippe

Whitehead

Laurence

(1986) Transitions from Authoritarian Rule, Vol. 4: Tentative Conclusions About Uncertain Democracies. Baltimore, MA: Johns Hopkins University Press.

40.

Przeworski

Adam

Alvarez

Michael

Antonio Cheibub

José

Limongi

Fernando

(2000) Democracy and Development. Cambridge: Cambridge University Press.

41.

Ross

Michael

(2001) Does oil hinder democracy? World Politics 53 (3): 325–361.

42.

Ross

Michael

(2004a) How do natural resources influence civil war? Evidence from thirteen cases. International Organization 58 (1): 35–67.

43.

Ross

Michael

(2004b) What do we know about natural resources and civil war? Journal of Peace Research 41 (3): 337–356.

44.

Ross

Michael

(2008) But seriously – does oil really hinder democracy? Unpublished paper, UCLA.

45.

Ross

Michael

(2012) The Oil Curse: How Petroleum Wealth Shapes the Development of Nations. Princeton, NJ: Princeton University Press.

46.

Smith

Alastair

(2008) The perils of unearned income. Journal of Politics 70 (3): 780–793.

47.

Smith

Benjamin

(2004) Oil wealth and regime survival in the developing world, 1960–1999. American Journal of Political Science 48 (2): 232–246.

48.

Tripp

Charles

(2002) A History of Iraq. Cambridge: Cambridge University Press.

49.

Ulfelder

Jay

(2007) Natural-resource wealth and the survival of autocracy. Comparative Political Studies 40 (8): 995–1018.

50.

Wantchekon

Leonard

(2002) Why do resource dependent countries have authoritarian governments? Journal of African Finance and Economic Development 5 (2): 57–77.

51.

Watts

Michael

(2007) Petro-insurgency or criminal syndicate? Conflict & violence in the Niger Delta. Review of African Political Economy 34 (114): 637–660.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB