country's oil wealth, they suggest, this democratizing effect will shrink or disappear. Does oil ... including the influ
Trustees of Princeton University
Does Oil Hinder Democracy? Author(s): Michael L. Ross Reviewed work(s): Source: World Politics, Vol. 53, No. 3 (Apr., 2001), pp. 325-361 Published by: Cambridge University Press Stable URL: http://www.jstor.org/stable/25054153 . Accessed: 16/07/2012 16:29 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact
[email protected].
.
Cambridge University Press and Trustees of Princeton University are collaborating with JSTOR to digitize, preserve and extend access to World Politics.
http://www.jstor.org
DOES OIL HINDER
DEMOCRACY?
ByMICHAEL L.ROSS*
Introduction scientists believe that oil has some very odd proper ties. Many POLITICAL to become
more
to this
tion
studies
show
democratic.
rule:
if rising
that when
incomes
tend rise, governments Yet some scholars imply there is an excep can be traced to a incomes oil country's
wealth, they suggest, this democratizing effect will shrink or disappear. oil really have
Does
properties? What
antidemocratic
about
other min
erals and other commodities? What might explain these effects? The
claim
specialists
to
is often used by area do not mix that oil and democracy states of the Arab Middle the high-income explain why
East have not become democratic. If oil is truly at fault, this insight could help explain?and perhaps, predict?the political problems of oil around the world, exporters states of Central the oil-rich
and such as Nigeria, Venezuela, Indonesia, If other minerals have similar prop Asia. account for the absence or weakness of de
this effect might help states in sub-Saharan in dozens of additional Latin Africa, mocracy and Southeast claim Asia. Yet the "oil impedes democracy" America, has received little attention outside the circle of Mideast scholars; erties,
moreover,
it has not been
carefully
tested with
regression
analysis,
ei
ther within or beyond theMiddle East. I use pooled and 1997
1971
time-series
cross-national
to
three
explore
aspects
data from
113
states between
of the oil-impedes-democracy
claim. The first is the claims validity: is it true?Although the claim has been championed byMideast specialists, it is difficult to test by examining only
cases from
theMiddle
East because
*
the region provides
scholars with
Previous versions of this article were presented to seminars at Princeton University, Yale Univer of California, Los Angeles, of the and at the September 2000 annual meeting sity, and the University inWashington, American D.C. For their thoughtful comments on ear Political Science Association Indra de Soysa, Geoffrey Garrett, Phil Keefer, Steve lier drafts, I am grateful to Pradeep Chhibber, Lowi, Ellen Lust-Okar, Lant Pritchett, Nicholas Sambanis, Jennifer Widner, Michael Knack, Miriam reviewers. I owe special thanks to Irfan Nooruddin for his research and three anonymous Woolcock, assistance and advice and to Colin Xu for his help with the Stata. I wrote this article while Iwas a vis Bank inWashington, D.C. The views I express in this article, and all re iting scholar at The World errors, are mine alone. maining
WorldPolitics 53 (April2001), 325-61
326 WORLD
POLITICS
Table 1 Index of Oil-Reliant 1. Brunei (1994)
Statesa
47.58
46.14
2. Kuwait
3. Bahrain 45.60 4. Nigeria (1991) 45.38 5. Congo,
Dem.
45.14
Rep.
6. Angola (1996)
45
38.58
7. Yemen
38.43
8. 9. 10. 11. 12. 13.
33.85 Qatar (1994) 33.85 Libya (1988) 29.74 Iraq (1983) 23.48 Algeria 21.44
14.
Venezuela
Oman Saudi Arabia
18.84
15. Syria 15.00 16.
13.46
Norway
17. Iran (1983) 11.95 18. Ecuador 8.53 19. Malaysia 5.91 20.
Indonesia
21.
Cameroon
5.69 5.63
22. Lithuania 4.48 23. Kyrgyz Republic (1996) 24.
4.25
3.14
Netherlands
3.13
25. Colombia
ismeasured
aOil reliance
by the value
of fuel-based
exports
divided
by
GDP. Most
figures
are based on data for 1995 fromWorld Bank (fn. 71).Figures for Brunei, Nigeria, Qatar, Libya,
Iraq,
and Iran are the most
gyz Republic
are not
little variation ments
have
available,
recent
1996
available.
figures
are
Since
1995
figures
for Angola
reported.
on the dependent been authoritarian
variable: virtually allMideast govern since gaining Moreover, independence. in the for the absence of democracy explanations
there are other plausible the influence of Islam and the regions distinct Mideast, including influence and colonial history. Does oil have a consistently negative once one accounts other variables? for these and mocracy Second,
and Kyr
I examine
is geographic.
For
on de
two dimensions. the claim's generality along reasons the oil-impedes-democracy obvious
has been explored most carefully byMideast teen states most
culture
reliant
on oil wealth
One claim
specialists: ten of the fif
are in the Middle
East
region
(see
Table 1). But is oil an obstacle to democracy only in theMideast, or does it harm oil exporters everywhere? If the hypothesis is true for all oil-rich
DOES OIL HINDER DEMOCRACY? Table 2 of Mineral-Reliant
_Index 1. Botswana
327
States'1_
35.11
2. 3. 4. 5. 6. 7. 8. 9. 10.
Zambia Bahrain Chile
Angola (1996) 11.5 Papua New Guinea 10.13 Togo (1991) 7.79 Bolivia 5.53 Congo, Dem. Rep. (1983) Jordan 5.28
11.
Peru
24.97 16.39 12.63
7.00
3.84
12. Central African Republic 3.16 13. Iceland 3.11 14. Zimbabwe 3.00 15. Norway
2.49
16. Belgium 2.23 17.
Canada
2.22
18. Australia
19. 20. 21. 22.
2.20
1.96 1.87 Jamaica Slovak Republic 1.74 South Africa 1.69 Lithuania
23. Morocco 24.
1.65
Cameroon
1.62
25. Kyrgyz Republic
1.56
aMineral reliance ismeasured by the value of nonfuel mineral exports divided by GDP. Most figures are for 1995 based on data fromWorld Bank (fn. 71). The figures for Congo and
Togo
are the most
recent
available;
the 1996
figure
is
reported
since no
for Angola,
fig
ure for 1995 is available.
states, then its importance has been underappreciated by other political scientists. If it holds only for states in theMideast, why is this so? is sectoral: do other types of minerals dimension and on of commodities have similar effects While types governments? in the Middle oil exporters tend to be concentrated of East, exporters The
other
other
nonfuel minerals aremore geographically dispersed (seeTable 2). Have these wealth? found
states, too, been rendered have Or does petroleum in other
less democratic antidemocratic
because properties
of resource that are not
commodities?
Finally, I explore the question of causality: if oil does have antidem ocratic
effects,
explanations:
what
is the causal
a "rentier
effect,"
mechanism? which
suggests
I test that
three
possible resource-rich
WORLD POLITICS
328 use
tax rates
to relieve pressures for and patronage resource that which effect," argues accountability; "repression to boost their retards democratization by enabling governments
governments
low
a
greater wealth
a "modernization funding for internal security; and that growth based on the export of oil and minerals the
social
and cultural
changes
that
effect," which fails to bring to democratic produce
tend
holds about gov
ernment.
I also have
two broader to
study democracy Many
"global"
studies
aims. The
first
the Middle
incorporate of democratization
is to encourage scholars who East into their analyses.
have
avoided
en
the Mideast
tirely.1 Influential studies by Przeworski and Limongi and Przeworski, Alvarez, Cheibub, and Limongi simply drop the oil-rich Mideast states from
their database.2
scholars doing
There
of democracy so can only weaken
is, however, to exclude these any general
no
reason for analytical from their research, and It also tends to margin
sound
states
findings.
alize the field ofMiddle East studies. My
second
aim
is to address
the literature
on the "resource
curse."
Many of the poorest and most troubled states in the developing world have,
high paradoxically, body of evidence growing
levels
of natural
resource
that resource wealth
wealth.
There
is a
itself may harm a coun resource greater natural
for development. States with try's prospects counter wealth tend to grow more slowly than their resource-poor are also more to suffer from civil wars.4 This article likely parts.3 They that there is a third suggests as well oil and mineral wealth tends to make
to the resource component states less democratic.
curse:
1 eds., O'Donnell, See, for example, Guillermo Philippe C. Schmitter, and Lawrence Whitehead, Rule: Prospects for Democracy Transitions from Authoritarian (Baltimore: Johns Hopkins University Press, 1986); D. Larry Diamond, Juan J. Linz, and Seymour Martin Lipset, eds., Democracy inDevel Ronald Inglehart, Modernization andPostmod oping Countries (Boulder, Colo.: Lynne Rienner, 1988); emization Press, 1997). (Princeton: Princeton University 2 and Facts," World Politics 49 Adam Przeworski and Fernando Limongi, "Modernization: Theories and Fernando Limongi, Cheibub, (January 1997); Adam Przeworski, Michael Alvarez, Jos? Antonio 7 (January 1996); idem, Democracy and De "What Makes Democracies Endure?"JournalofDemocracy in the World, 1950-1990 Institutions and Well-Being (New York: Cambridge Uni velopment: Political versity Press, 2000). 3 and Economic Growth," "Natural Resource Abundance Jeffrey D. Sachs and Andrew M. Warner, no. 517a Institute for International Develop Discussion (Cambridge: Harvard Development Paper ment, 1995); idem, "The Big Push, Natural Resource Booms and Growth," Journal ofDevelopment "Does Mother Nature Corrupt? Nat Economics 59 (February 1999); Carlos Leite and JensWeidmann, IMF ural Resources, Corruption, and Economic Growth," (1999); Michael Working Paper, WP/99/85 L. Ross, "The Political Economy of the Resource Curse," World Politics 51 (January 1999); R. M. Auty, Resource Abundance and Economic Development (Oxford: Oxford University Press, 2001). 4 "On Economic Causes of Civil War," Oxford Economic Papers 50 Paul Collier and Anke Hoeffler, or Curse: Are Civil Wars Driven 1998); Indra de Soysa, "The Resource (October by Rapacity eds., Greed and Grievance: Economic Agendas in Civil Paucity?" inMats Berdal and David M. Malone, Wars (Boulder, Colo.: Lynne Rienner, 2000).
DOES OIL HINDER DEMOCRACY?
329
claim and the limi
I begin by outlining the oil-impedes-democracy
of previous work on the topic. I then draw on earlier case stud states to ies of oil-rich that might three causal mechanisms specify more next sec authoritarian. The how oil makes governments explain tations
a model the research design. tion presents of regime types and describes tests and follow I then present the results of the validity and generality a discussion that with of the results of tests on the causal mechanisms and a conclusion.
of the
The Concept
"Rentier
State"
of the Mideast often describe most of the governments specialists a as and North Africa "rentier states," since they derive large fraction of than half of the government's their revenues from external rents.5 More
Area
revenues in Saudi Arabia, Bahrain, the United Arab Emirates, Oman, Kuwait,
and Libya
Qatar,
governments rents from
have,
of Jordan,
at times, and Egypt
Syria, for pipeline payments the Suez Canal. Workers'
through source of foreign
exchange
Algeria, andMorocco,
in
come
the sale of oil. The
from
earn
large locational fees, and passage have been an important Tunisia, Syria, Lebanon,
variously transit crossings,
remittances
Egypt,
Yemen,
although these rents go (at least initially) to pri
to Israel, and Egypt, foreign aid that flows a rent. type of economic in the early twentieth the term "rentier used century states that extended to the loans to non European
vate actors, not the state. The Jordan may also be considered Economists state"
to refer
European governments.6 Mahdavy
iswidely credited with giving the
a state that receives substantial its current meaning: or concerns Beblawi "foreign individuals, governments."7 one a state is rentier where that this definition, suggesting term
rents
from
later refined the rents are
accrue to the state, and where directly they paid by foreign actors, where a few are in the generation of this rent (wealth), the ma engaged "only
jority being only involved in the distribution or utilization of it."8
5 Throughout Banks definition
this article I use the term "Middle East"
to include North Africa. I adopt theWorld of this region: Algeria, Bahrain, Egypt, Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman, Qatar, Saudi Arabia, Syria, Tunisia, United Arab Emirates, and Yemen. Libya, Malta, Morocco, 6 to Lenin, "The rentier state is a state of parasitic, decaying capitalism, and this circum According of the countries concerned." V. I. stance cannot fail to influence all the socio-political conditions in Robert C. Tucker, ed., The Lenin Anthology the Highest Lenin, "Imperialism, Stage of Capitalism," (New York: W. W. Norton, 1975). 7 in Rentier States: The "The Patterns and Problems of Economic Development Hussein Mahdavy, Case of Iran," inM. A. Cook, ed., Studies inEconomic History of theMiddle East (London: Oxford Uni versity Press, 1970), 428. 8 Beblawi and Giacomo Lu in Hazem "The Rentier State in the Arab World," Hazem Beblawi, excludes that this definition 1987), 51. Note ciani, eds., The Rentier State (New York: Croom Helm,
WORLD POLITICS
330 about
Claims those
the rentier
that
oil wealth
suggest causes that suggest oil wealth economic moting development. cle focuses
state
can be
makes
sorted
states
into
two
less democratic
categories: and those
to do a poorer governments job of pro arti Often the two are conflated. This
on the first claim.
to Anderson, "The notion of the rentier state is one of the According sci of Middle contributions East regional studies to political major now use a version some scholars of of ence."9 Indeed, this democracy to account states of the Middle for the otherwise argument puzzling East. Huntington,
for example,
that the democratic
suggests
trend may
East since many of these states "depend heavily on
bypass theMiddle
oil exports, which of the state bureaucracy."10 enhances the control out countries Others have adapted the "rentier state" idea to oil-rich
side theMiddle
East.11
has not per se inhibits democratization as most been subjected statistical tests, however, quantitative it as an explanatory variable. And studies of democracy overlook simply states have odd prop that oil-rich the handful that even acknowledge The
claim
that oil wealth to careful
erties do little
to
for ex and his collaborators, if their "ratio of fuel exports eccentric cri fifty percent"?an
explain why. Przeworski countries from their database
ample, drop to total exports in 1984-1986 exceeded terion that excludes six oil-rich states,
all of which
are located
on the
a
includes dummy study of democracy a minimum net oil exports represent of two one to thirds of total exports and are at least equivalent approximately is statistically percent of world exports of oil."13 The Barro oil dummy as in the But correlated with and democracy. significant negatively
Arabian
Peninsula.12
variable
for states
analyses
of Przeworski
Barro's
"whose
et al., the dummy
variable
uses
an
arbitrary
cut
remittances. As Chaudhry notes, large flows of remittances have different political implica tions than do large oil rents. See Kiren Aziz Chaudhry, The Price ofWealth: Economies and Institutions in theMiddle East (Ithaca, N.Y.: Cornell University Press, 1997). 9 East and North Africa," Comparative Politics 20 (Octo Lisa Anderson, "The State in the Middle ber 1987), 9. 10 in the Late Twentieth Century (Norman: The Third Wave: Democratization Samuel P. Huntington,
workers'
of Oklahoma Press, 1991), 31-32. University 11 "Rent Capitalism, State, and Democracy: A Theoretical See, for example, Olle T?rnquist, Propo on Southeast Asia, sition," in Arief Budiman, ed., State and Civil Society in Indonesia, Monash Papers no. 22 (1990); in and Neocolonialism Douglas A. Yates, The Rentier State inAfrica: Oil Rent Dependency the Republic of Gabon (Trenton, N.J.: Africa World Press, 1996); Terry Lynn Karl, The Paradox of of California Press, 1997); John Clark, "Petro (Berkeley: University Plenty: Oil Booms andPetro-States of the State 8 (July 1997); idem, "The Nature and Evolution Politics in Congo,"Journalof'Democracy 32 (Winter 1998). in Zaire," Studies in Comparative International Development 12 et al. (fn. 2,2000), 77. See Przeworski 13 107 (December 1999). Robert J. Barro, "Determinants of Democracy," Journal ofPoliticalEconomy
DOES OIL HINDER DEMOCRACY?
331
to "oil states" and "non-oil states" and im between distinguish some thresh or no influence on until has little that oil regime type plies old is reached.
point
also studies of the oil-impedes-democracy Qualitative hypothesis have been country-level have important limitations. The vast majority case studies of oil-rich states in the Mideast. many have been Although a is nevertheless rich and analytically the Mideast nuanced, empirically difficult
place have
ernments
absence
The
to test this claim, since gov virtually all oil-rich Mideast since gaining been highly authoritarian independence. as on on the well variable?as of variation dependent
It has control variable?has made Islam, an important testing difficult. to tasks that would also allowed Mideast help neglect specialists the key claim?defining sharpen and refine the oil-impedes-democracy in falsifiable the causal arguments variables better, specifying terms, and cases to which their the relevant domain of arguments apply. outlining As
a result,
the notion
of conceptual
overstretch:
of the rentier assertions
state has about
suffered
from
a bad case
of oil on Mid
the influence
dle East politics have become so general that their validity has been di luted. As Okruhlik imply One
so much
that
state has come
"The idea of the rentier observes, it has lost its content."14
way to restore the usefulness it statistically. I thus evaluate
of an overstretched one
core facet
concept of the rentier
to
is by state
testing three questions. claim?with concept?the oil-impedes-democracy a oil and authoritar valid correlation between First, is there statistically can the for? Second, ianism once other germane variables are accounted
claim be generalized both beyond theMiddle East and beyond the case of oil? Finally, if oil thwarts democracy, what is the causalmechanism? Proponents
of the oil-impedes-democracy
hypothesis
naturally
sug
gest both that it is valid and that it can be generalized to oil exporters outside theMiddle East. Some also imply that other types of com modities have similar effects. Nothing in Beblawi s definition, which is restricts the set of rentier among Mideast specialists, accepted to cover many to oil exporters. In fact, the definition appears on the tend to generate that (1) minerals mineral exporters grounds rents, (2) the rents are largely captured by states via export taxes, cor extraction and (3) mineral taxes, and state-owned porate enterprises, same little labor. The however, definition, implies relatively employs of agricultural will not be rentier states. that exporters commodities
widely states
14 Gwenn
Okruhlik,
"Rentier Wealth,
tics31 (April1999), 308.
Unruly
Law, and the Rise of Opposition,"
Comparative
Poli
332
WORLD
POLITICS
is because (1) agricultural commodities generally do not produce
This
cases go in most rents, (2) export revenues directly is more the state, and (3) agricultural production a hence employs fraction of the population larger
to
not private actors, labor intensive and
for a given
value
of
exports.15
Causal Mechanisms At
least
tween
three
Mideast
the alleged link be might explain rule. The first comes largely from
causal mechanisms and authoritarian
oil exports
specialists and might be called the "rentier effect." A close
case studies a second mechanism: suggests reading of a third fect." Modernization theory implies possible the "modernization effect."
The Rentier
Effect comes from the work of Middle
The first causal mechanism scholars, who have pondered eral they argue that governments that might otherwise pressures ability. Case
a ef "repression I call cause, which
studies
describe
issue for over
this
use
their oil revenues
lead to demands
three ways
East
two decades.16
this may
In gen to relieve social
for greater occur.17
account
The first is through what might be called a "taxation effect." It sug gests
revenues from the sale of derive sufficient governments or not at all, and to tax their less heavily likely populations
that when
oil, they
are
the public in turnwill be less likely to demand accountability from? in?their and representation government.18 in studies of the evolution is grounded of The of the argument logic in Histori institutions and France. democratic early modern England ans and scientists political arose tion in government
for representa have argued that the demand to raise to the in response sovereign's attempts
15 con states are politically Note that, by contrast, dependency theory suggests that developing to advanced industrialized strained by their reliance on the export of ?///types of primary commodities states. See, for Cardoso and Enzo Faletto, Dependency and Development example, Fernando Henrique of California Press, 1979); Peter Evans, Dependent Development: in Latin America (Berkeley: University Princeton University The Alliance ofMultinational, State, and Local Capital in Brazil(Princeton: Press, and Democracy: The Cross-Na "World System Position, Dependency, 1979); Kenneth A. Bollen, tional Evidence," American Sociological Review 48 (August 1983). 16 (fn. 8), notes that "theories of the ren Perhaps they have thought about it too carefully. Chaudhry tier state far outstrip detailed empirical analysis of actual cases" (p. 187). 17 to clar Case studies often conflate these three effects. I treat them here as separate mechanisms ify their logic. 18 Giacomo Luciani, Luciani (fn. 8).
"Allocation
vs. Production
States: A Theoretical
Framework,"
in Beblawi
and
DOES OIL HINDER DEMOCRACY? taxes.19 Some Mideast tween
scholars
in tax levels
variations
found
have
looked
for similar
333 correlations
in the demand for political the govern that the discovery of oil made to the traditional merchant less accountable and variations
accountability. Crystal ments of Kuwait and Qatar class.20 Brand's study of Jordan
that a drop in foreign in the 1980s led to greater pressures mittances for political not all Middle have been tion.21 Yet East specialists argues
Waterbury
be
argued
that "neither historically
nor in the twentieth
aid and re representa persuaded: is
century
theremuch evidence [in theMiddle East] that taxation has evoked de mands
account for their use of tax monies. that governments Predatory in the countryside, but there has has produced revolts, especially no translation of tax burden into pressures for democratization."22
taxation been
A
of the rentier effect might
second component effect":
oil wealth
"spending in turn age, which dampens for example, Entelis, argues
be called the
on may lead to greater patron spending latent pressures for democratization.23 that
the Saudi Arabian
government
used
its oil wealth for spending programs that helped reduce pressures for democracy.24
Vandewalle
government.25
Mexican
And
Kessler
a similar
about the Libyan argument and Bazdresch and Levy find that the
makes
oil boom of the 1970s helped prop up?and
long?one-party
rule.26 While
all authoritarian
perhaps pro
governments
may
use
19 States inWestern Europe (Princeton: Princeton Uni Charles Tilly, ed., The Formation ofNational Donald Lien, "ANote on Taxation, Development, versity Press, 1975); Robert Bates and Da-Hsiang and and Representative Politics and Society 14 (January 1985); Philip T Hoffman Government," 1450-1789 Government, (Stanford, eds., Fiscal Crises, Liberty, and Representative Kathryn Norberg, Calif.: Stanford University Press, 1994). 20 inKuwait and Qatar (New York: Jill Crystal, Oil and Politics in the Gulf: Rulers andMerchants Press, 1990). Cambridge University 21 in a Rentier Economy: The Case of the Laurie A. Brand, "Economic and Political Liberalization and Liberal Hashemite of Jordan," in Iliya Harik and Denis J. Sullivan, eds., Privatization Kingdom Indiana University ization in theMiddle East (Bloomington: Press, 1992). 22 in The Potential for Political Liberalization John Waterbury, "Democracy without Democrats? theMiddle East," in Ghassan Salame, ed., Democracy without Democrats? The Renewal ofPolitics in the Muslim World (New York I. B.Tauris, 1994), 29. 23 a formal model that makes a similar point, that resource wealth Lam andWantchekon develop can the distributive influence of an elite. Ricky Lam and Leonard impede democracy by enhancing as a Political Dutch Disease" of Political Sci Wantchekon, (Manuscript, Department "Dictatorships ence, Yale University, January 1999). 24 in the Arabian Peninsula: and the Prospects for Democratization "OilWealth John P. Entelis, in Naiem A. Sherbiny and Mark A. Tessler, eds., Arab Oil: Impact on the The Case of Saudi Arabia," Arab Countries and Global Implications (New York: Praeger, 1976). 25 Dirk Vandewalle, (Ithaca, N.Y.: Cornell Univer Libya since Independence: Oil and State-Building sity Press, 1998). 26 Carlos Bazresch
in and Santiago Levy, "Populism and Economic 1970-82," Policy inMexico, in Latin America and Sebastian Edwards, eds., The Macroeconomics of Populism R?diger Dornbusch P. Kessler, Global Capital and National Politics: of Chicago Press, 1991); Timothy (Chicago: University sFinancial System (Westport, Conn.: Praeger, 1999). Reforming Mexico
WORLD POLITICS
334
their fiscal powers to reduce dissent, these scholars imply that oil wealth providesMiddle East governments with budgets that are exceptionally Rulers large and unconstrained.27 same tactics as their authoritarian enues
could make
in the Middle
East
may
follow
the
but oil rev
elsewhere, counterparts at fiscal more effective. pacification
their efforts
The third component might be called a "group formation" effect. It a that when oil revenues with government provide enough use to the will its of the formation money, government prevent largesse social groups that are independent from the state and hence that may implies
be inclined to demand political rights. One version of this argument is inMoore's
rooted
claim
the formation
that
of an independent
bour
geoisie helped bring about democracy inEngland and France.28 Schol ars the cases of Algeria, and Iran have all examining Libya, Tunisia, states observed oil-rich the formation of social blocking independent a state all is that the groups; argue necessary precon thereby blocking dition of democracy.29 A second version of the group-formation effect draws on Putnam's that the formation argument above the family and below cratic governance.30 Scholars
of social
institutions that lie capital?civic to promote more demo the state?tends cases of the Iran, Iraq, studying Algeria,
and the Arab Gulf states have all suggested that the government's oil wealth has impeded the formation of social capital and hence blocked a transition
to
democracy.31
Whether Mideast formation
group First suggests
states use their oil revenues to deliberately inhibit
In the case of of some disagreement. Libya, "there is not a consistent the of policy against development is a matter
27 Lisa Anderson, "Peace and Democracy in the Middle East: The Constraints of Soft Budgets," Journal of International Affairs 49 (Summer 1995). 28 Social Origins ofDictatorship and Democracy (Boston: Beacon Press, 1966). Moore, 29Barrington see Clement in theMaghreb," On Algeria, "Petroleum and Political Development Henry Moore, in Sherbiny and Tessler (fn. 24); on Libya, see Ruth First, "Libya: Class and State in an Oil Economy," in Petter Nore and Terisa Turner, eds., Oil and Class Struggle (London: Zed Press, 1980); also on Libya, see Vandewalle (fn. 25); on Tunisia, see Eva Bellin "The Politics of Profit inTunisia: Utility of the Ren tier Paradigm?" World Development 22 (March 1994); and on Iran, see Hootan Shambayati, "The Ren tier State, Comparative 30 Robert
Interest Groups, and the Paradox Politics 26 (April 1994).
of Autonomy:
State
and Business
in Turkey
and Iran,"
Putnam, Making Democracy Work: Civic Traditions inModern Italy (Princeton: Princeton Press, 1993). University 31 see John P. Entelis, "Civil Society and the Authoritarian in Algerian On Algeria, Temptation Politics," inAugustus Richard Norton, ed., Civil Society in theMiddle East, vol. 2 (Leiden: E. J. Brill, 1995); on Iran, see Farhad Kazemi, "Civil Society and Iranian Politics," in Norton; on the Gulf states, on see Zuhair Humadi, see Jill "Civil Crystal, "Civil Society in the Arab Gulf States," inNorton; Iraq, ed., Toward Civil Society in theMiddle East? Society under the Ba'th in Iraq," in Jillian Schwedler, of civil society scholars have argued that the weakness 1995). Other (Boulder, Colo.: Lynne Rienner, that oil wealth is the in theMiddle East has hampered a transition to democracy, without suggesting source of this weakness.
DOES OIL HINDER DEMOCRACY?
335
an
but the growth of this class is in practice con bourgeoisie, indigenous ventures and its links with interna strained by the state s own economic tional capital."32 Chaudhry, argues that in the 1970s the by contrast, revenues to their oil that Mideast used governments develop programs were to In the all cases, "explicidy designed depoliticize population....
governments deliberately destroyed independent civil institutions while to facilitate aims of the state."33 the political designed the and effects taxation, Collectively, spending, group-formation a states fiscal constitute the rentier effect. Together that they imply its regime that fund themselves influence type: governments policies others
generating
revenues and have through oil that thoritarian; governments are more small relatively likely
The Repression A
close
reading
larger budgets fund themselves to become
are more
through democratic.
likely taxes
to be au and are
Effect of case studies
from
the Mideast,
Africa,
and Southeast
Asia suggests that oil wealth and authoritarianism may also be linked as states may want in resource-rich Citizens democracy by repression. as citizens resource allow their much but wealth elsewhere, may gov so block the more on internal ernments to and security popula spend notes of Iran's tions democratic that much aspirations. Skocpol she calls oil wealth was spent on the military, producing what pre-1979 a "rentier absolutist
in his study of the 1990s oil boom in state."34 Clark, allowed the of Congo, finds that the surge in revenues to build up the armed forces and train a government special presidential to East de order.35 And Gause argues that Middle guard help maintain the Republic
mocratization
has been inhibited in part by the prevalence of the
state.36 security) resource wealth two reasons lead to might why forces. One may be pure self-interest: larger military given the oppor an authoritarian to better arm itself pressures, against popular tunity so. A second reason may be that resource will do government readily a causes ethnic or reflect wealth larger military might regional conflict; mukhabarat There
(national are at least
the government
s response. Mineral
wealth
is often
geographically
con
32 First (fn. 29), 137. 33 of the Rentier State," Com Kiren Aziz Chaudhry, and the Lineages "Economic Liberalization 9. parative Politics 27 (October 1994), 34Theda Skocpol, "Rentier State and Shi'a Islam in the Iranian Revolution," Theory and Society 11
1982). (April 35
Clark (fn. 11,1997). 36 in the Arab in Political Liberalization F. Gregory Gause II, "Regional Influences on Experiments and Democrati in Rex Brynen, Bahgat Korany, and Paul Noble, eds., Political Liberalization World," zation in theArab World, vol. 1, Theoretical Perspectives (Boulder, Colo.: Lynne Rienner, 1995).
336
WORLD POLITICS
an in a region by populated resource extraction may or exac ethnic promote religious minority, as federal, actors compete erbate ethnic tensions, and for local regional, to and mineral These lead forces rights. disputes may larger military states such as An in resource-rich, less democracy fractured ethnically centrated.
If it
happens
to be concentrated
or
the Democratic of Congo, Burma, Indonesia, Nigeria, Republic New Sierra and South Africa. This mechanism Leone, Guinea, Papua gola,
would be consistent with the research of Collier and Hoeffler Soysa, who find more likely.37
resource
that natural
The Modernization
wealth
tends
to make
and de civil war
Effect
a third Finally, explanation which holds that democracy
can be derived is caused
from modernization
a collection by
of social
theory, and cul
and tural changes?including urbanization, specialization, occupational in turn are caused by economic levels of education?that devel higher different clusters of social and scholars emphasize opment.38 Different cultural from impact
on the likelihood
comes
the most
carefully shaped position that two types of social change have that a state will become democratic:
changes. Perhaps who argues Inglehart,
a direct
1. Rising education levels,which produce amore articulate public that is bet ter
equipped
to
organize
and
communicate,
and
2. Rising occupational specialization, which first shifts the workforce secondary autonomous
sector
and
then
workforce,
into
the
accustomed
sector. These tertiary changes to for themselves thinking
produce on the
into the amore job
and
having specialized skills that enhance their bargaining power against elites.39 not address the of re modernization question theory does Although an source wealth if economic is that per se, develop implicit corollary ment it will not and social changes, these cultural does not produce result
notes: aIs the As Inglehart linkage between not: if se? due to wealth and democracy per Apparently then resulted from simply becoming wealthy, automatically
in democratization.
development
democracy if In other words, Kuwait and Libya would be model democracies."40 not lead to and education levels does resource-led higher growth 37 See Collier and Hoeffler (fn. 4); de Soysa (fn. 4). 38 Economic and of Democracy: Development Seymour Martin Lipset, "Some Social Requisites "Social 53 (March 1959); Karl W. Deutsch, Political Legitimacy," American Political Science Review American Political Science Review 55 (September 1961); In Mobilization and Political Development," glehart (fn. 1). 39 (fn. 1), 163. 40 Inglehart Ibid., 161.
DOES OIL HINDER DEMOCRACY?
337
it should also fail to occupational specialization, bring about de the rentier Unlike the and modernization effects, mocracy. repression state: it is a social mechanism, not a effect does not work the through one. political The and modernization effects are largely com rentier, repression, on the use of fiscal rentier The effect focuses plementary. government's greater
measures fect
to
the public keep the government's
stresses
the repression ef demobilized; politically use of force to keep the public demobi effect looks at social forces that may keep
lized; and the modernization All the public demobilized. be them, may simultaneously
Model
or any combination
three
explanations, valid.41
Specification
and Research
of
Design
a model to claim, I present predict a feasible generalized least-squares a in time-series with cross-national data set, which pooled on over one states data all sovereign with populations hundred
test the oil-impedes-democracy types and test it using regime
To
method cludes
thousand between 1971 and 1997. The model includes five causal vari ables nants
that according of democracy.
and mineral The
to studies are the most robust determi previous a state s oil measure It also includes variables that
to see if they add explanatory is: model regression
wealth
basic
= + Regime.t a1 b/Oil.i5) + + bjlslam.) b/OECDJ where
+ +
b/Minerals. b?Regime.J
power.
) + b/Log Income. J + + b/YearJ... bJYearJ
and / is the year.
/ is the country
The dependent variable, Regime, is derived from the Polity98 data set constructed 0-10
states
entiates
between
variable
differentiates
cators
contain
each country
two Gurr and Jaggers compile and AUTOC; the former differ while the latter that are relatively democratic,
and Jaggers.42 by Gurr DEMOC scale variables,
interval
between
authoritarian
states. of
Since
the two
information
separate, types nonoverlapping them into a single measure year, I combine
indi about
by subtracting
41A fourth explanation has been offered by U.S. vice president Richard Cheney, a political scientist see fit to put oil and gas reserves where there are by training: "The problem is that the good Lord didn't in in David "Oil and Politics Mix democratic Cited Ignatius, governments." Suspiciously Well America," Washington Post, July 30,2000, A31. 42 1. Ted R. Gurr and Keith Jaggers, Each of the variables is defined more precisely in Appendix 1999 (con 1800-1998," "Polity 98: Regime Characteristics, http://www.bsos.umd.edu/cidcm/polity/, sulted March
1,2000).
338
WORLD
POLITICS
the autocracy measure from the democracy then rescale it measure.431 as a 0-10 variable, with 10 representing "most democratic.,, are the measure Oil and Minerals the ex variables; independent they value of mineral-based fuels natural and coal) and gas, port (petroleum, ores and metals as fractions of GDP. the export value of nonfuel exports, These sources
variables
of fuels capture both the importance of export revenue and their relative importance
as
and minerals in the domestic
economy.44
The
side of the equation also includes five control vari to capture most the factors associated with designed robustly are available for most of the countries for which indicators regime type, as the natural and years. The first is Income, measured log of per capita GDP corrected for purchasing interna power parity (PPP), in current right-hand
ables
as a correlate tional dollars. Per capita income has been widely accepted re its validity has been confirmed in more of democracy since Lipset; cent tests and Lewis-Beck, and Poole, Prze by Burkhart Londregan worski and Barro.45 and Limongi, The
second
control
variable
centage
of the states
gested
that states with
democratic tested
population
denotes
in 1970.46 Previous
large Muslim
than non-Muslim Islam
is Islam, which
populations
states.47 Of
the Muslim studies
have
per sug
tend to be less
all the religious categories the same data set)
the same way with
(measured by Barro, on a had by far the largest and most influence statistically significant states Islam in this model has special importance regime type.48 Placing 43 Here
I am following the practice of John B. Londregan and Keith T. Poole, "Does High Income Promote Democracy?" World Politics 49 (October 1996). 44 are similar to the indicators used and by Leite Oz/and Minerals (fn. 3,1995) by Sachs andWarner andWeidmann (fn. 3) in their studies of the influence of resource wealth on economic performance. While and agricultural goods into a single vari Sachs andWarner combine fuels, nonfuel minerals, able, I consider them as separate variables to see if their regression coefficients (and hence their influ ence on regime types) differ. 45 S. Lewis-Beck The (fn. 38); Ross E. Burkhart and Michael Lipset "Comparative Democracy: Economic Development 88 (December Thesis," American Political Science Review 1994); Londregan and Limongi and Poole (fn. 43); Przeworski (fn. 2); Barro (fn. 13). 46 In virtually all cases, the figure for 1980 (the only other year for which data were available) was identical to the 1970 figure. 47 of Democracy Salame (fn. 22); Seymour Martin Lipset,"The Social Requisites Revisited," Ameri can 59 (Februaryl994); Manus Midlarsky, for "Democracy and Islam: Implications Sociological Review Civilizational Conflict and the Democratic Peace," International Studies Quarterly 42 (December 1998). 48 Barro (fn. 13). Observers offer different arguments to explain the negative correlation between Sharabi, Neopatriarchy: A Theory (-.38). See, for example, Hisham democracy and Islamic populations inArab Society (New York: Oxford University Press, 1988); Bernard Lewis, "Islam ofDistorted Change 271 (February 1993); and Michael Hudson, "The Political and Liberal Democracy," Atlantic Monthly to Arab Democratization: Culture Approach It Back In, Carefully," in Brynen, The Case for Bringing are (fn. 36). Although they negatively correlated for the period covered by this data Korany, and Noble set (1971-97), it is not obvious that they will continue to be negatively correlated in the future. Two
DOES OIL HINDER DEMOCRACY? because
many
states with
great mineral
wealth
339
also have
large Muslim
populations, not only in theMiddle East but also in parts of Asia (In donesia, Malaysia, Brunei) andAfrica (Nigeria). The simple correlation between Oil and Islam is 0.44. The third control variable is OECD, a dummy that is coded 1 for that are members of the Organization for Economic newer members Mexico and Development (excluding researchers have found Korea) and 0 for all others. Previous states
Coopera and South
tion
that the ad
vanced industrialized states of the OECD are significantly more likely to be democratic
II era than
in the War post-World even after the influence
veloping world, for.49 There accounted
is no consensus
of income on
why
the states of the de and other
factors
are
this is so. It has variously
s unique historical trajectory;50 the cultural
been attributed to theWest
influence of Protestantism;51 the residual effects ofWestern colonialism on non-Western a "world the states;52 and system" that constrains states in of non-Western the any prospects "periphery."53 Conceivably antidemocratic effects from Oil and Minerals be spurious and might states and mineral-exporting The OECD dummy helps account for any of a on the mech effects, without position taking
reflect the location merely in the non-Western world. these Western-specific
of most
fuel-
anisms behind it. The
fourth
control
variable
is
Regimet5,
which
is the dependent
vari
able lagged by five years. Placing it on the right-hand side of themodel on a state s the most influence First, purposes. important own regime type may often be its peculiar history; Regimef5 helps cap or cultural ture any historical features that may be country-specific missed the other variables. Second, by right-hand-side including has
three
turn the into a change model, transforming helps equation Regime in a variable from regime type to the change the dependent country's over a ensure that the re regime type given five-year period. This helps
states with large Islamic populations, Nigeria and Indonesia, have recently moved toward democracy, and some of the most important prodemocracy forces in other Islamic states (including Algeria, Egypt, are often classified as Islamic "traditionalists" or "fundamentalists." It is instruc Jordan, and Malaysia) tive to recall that until the "third wave" of democratization and began in the mid-1970s, democracy were Catholicism correlated. negatively 49 and Poole (fn. 43); Przeworski and Limongi See Burkhart and Lewis-Beck (fn. 45); Londregan (fn.2). 50 See Moore (fn. 28). 51 See Lipset (fn. 38); Huntington (fn. 10). 52 and Opposition (New Haven: Yale University See Robert A. Dahl, Polyarchy: Participation Press,
1971). 53
The Modern See Immanuel Wallerstein, and Lewis-Beck (fn. 45).
(fn. 15); Burkhart
World-System
(New York: Academic
Press,
1974); Bollen
WORLD POLITICS
340 gression
will
indeed
measure
both
time-series
and
cross-sectional
changes in regime types.Third, Regime 5helps address the problem of serial
correlation
tional
data
that
to bedevil
tends
pooled
time-series
cross-sec
sets.54
one includes a set of twenty-six Finally, dummy variables, for each year covered by the data (1971-97), auto less one to mitigate are to correlation. These ef capture two types of time-specific designed the model
fects.The first is the coldwar, which may have blocked many transitions to states at second are contagion effects that influenced democracy. The different times in Southern and Eastern Latin and America, Europe, to to sub-Saharan Africa, where early transitions democracy appeared states. boost the likelihood of subsequent in transitions proximate run with a feasible tests were The process generalized least-squares using
Stata
6.0.55
Since
I include
side of the equation,
a
lagged dependent I correct for first-order
variable
on
the
autocorrelation
right-hand a allows the degree of autocorrela process, which using panel-specific tion to vary from country to country. I use a and control variables. The five-year lag for all independent lag more confidence arrow is that the in the right di causal gives pointing im rection; it also enables me to look for factors that have an enduring on not pact regime types. As I illustrate below, using shorter lags does the results of the it basic but does increase the absolute model, change
value of the coefficient of the lagged dependent variable relative to the other
explanatory
variables.
rent
regime type becomes in the previous year, while
Hence
with
overwhelmingly the influence
a one-year a lag, country's a function of its regime of other
variables
cur
type is artificially
suppressed.56
Results For the basic model country-year observations
described below, Stata is able to utilize 2,183
observations from
from
158 states. The
states, out of a possible 3,752 data for each of the variables are sum 113
marized inAppendix 2. 54 James A. Stimson, "Regression in Space and Time: A Statistical Essay,"American JournalofPolit icalScience 29 (November 1985); Nathaniel Beck and Jonathan N. Katz, "What to Do (and Not to Do) with Time-Series Cross-Section Data," American Political Science Review 89 (September 1995). 55 Beck and Katz (fn. 54) recommend using ordinary least squares with "panel-corrected standard errors" when working with panel data if the number of units is less than the number of time points. In this data set the number of units (113) exceeds the number of time points (27). 56 Variables Can Suppress the Explanatory Power "Why Lagged Dependent Christopher H. Achen, of Other Independent Variables" (Paper presented at the annual meeting of the Political Methodology Section of the American Los Angeles, Political Science Association, July 20-22,2000).
DOES OIL HINDER DEMOCRACY?
341
3 Table and Democracy3 Wealth is regime) variable (dependent
Resource
.253***
Regime
(.0203)
Oil
(.0051) (.00778)
(.00317)
.922***
.119***
(.00506) -.0438***
(.0081) .935***
-.0031***
(.00208)
.176*
(.308)
(.00543) -.0455***
(.00804) .965***
(.107)
-.0178***
-.0173***
(.0021)
(.000665)
1.47***
-.0393***
(.106)
(.0342)
-.018***
OECD
-.0339***
-.00718*
(.105) Islam
.246***
(.0204)
(.0024)
-.0459***
Income (log)
.25***
(.0203)
-.0078***
-.0346***
Minerals
.894***
(.00846)
1.42***
(.305)
(.0781)
Food
(.00211) 1.44***
(.308)
.0244*
(.0102) .042
Agriculture
(.0239) Observations States 113 Log likelihood
2498 115 -3283
2183 -3133
2182 113 -3129
2178 113 -3123
*
** *** at the 0.001 at the 0.05 level; at the 0.01 level; level significant significant significant are entered with in column aAll and control variables five-year lags, except independent are entered with a one-year errors are in Standard below the 2, where they lag. parentheses run with Stata Least corrected coefficients. Feasible Generalized 6.0; Squares regressions a is run with Each for first-order autocorrelation process. regression using panel-specific dummy
variables
for every
year
(but one)
covered
by
the data.
The results of the basic model are reported inTable 3, column 1.All of the variables are highly significant with the expected signs.57Both effects; these effects are of is some theMinerals coefficient
Oz/and Minerals
antidemocratic
roughly what larger.58
although
have strong the same magnitude,
57 Most
of the coefficients for the year dummies are also significant: for years 1971-89 the coefficients not sig negative and range from marginally to highly significant, for 1990 the coefficient is negative but nificant, and for years 1991-96 the coefficients are positive, although all but one (1994) are not significant. 58 These results were unaffected by the inclusion of other variables that are sometimes significant in status as a former British colony, Catholic attainment, including educational democracy regressions, run with a random the last variable was significant. When and trade openness. Only population, run with a fixed test produces a chi2 of 466 and a P value of 0.000. When effects process, aHausman for the lagged dependent effects process, however, none of the right-hand-side variables?except variable and Log Income?are significant. are
342 WORLD POLITICS -2
H
'So -1
0
-
-F-1-1-1-,-1
200 10 Value
of Oil Exports
30 (U.S. Dollars,
50
Billion
40
per Year)
Figure 1 Impact of Oil Exports on Regime a This
on the 0-10 variable a shows the net predicted impact of oil exports Regime, for a per capita income of $1,720 dollars a year, which of million with country twenty hypothetical people is the sample mean. Note the scale on the Y-axis is negative. figure
of oil and that the antidemocratic suggest properties a are substantial: standard deviation rise in the single a .49 over the in the 0-10 democracy index produces drop
The
results
mineral
wealth
Oil variable
five-year period, while leads to a .27 drop. A
a standard state
that
deviation is highly
rise in theMinerals reliant
1995 level ofAngola, Nigeria, or Kuwait?would scale due
democracy
on mineral
to its oil wealth
alone. A
variable
on oil exports?at
the
lose 1.5 points on the
state that was
equally
de
lose 2.1 points. that the impact of any new oil or however, implies, mineral wealth may be partly offset by a rise in income. To complicate on is nonlinear, and the the influence of Oil andMinerals matters, Regime on the state's of level of their income.59 prior magnitude impact depends pendent The model
exports would
also
As Figure 1 shows, the marginal influence of Oil on Regime is larger
and it drops oil exports are a small fraction of the economy, Barro and Przeworski reliant on oil. While country grows more
when
imply
that oil wealth
matters
only when
59 These effects occur because Income is entered an oil discovery will influence both the numerator
exports
reach
as the et al.
extraordinarily
in the model as a logarithmic function and because and the denominator in the Oil variable.
DOES -2
OIL HINDER
343
DEMOCRACY?
11
yj
,-,-,-,-,-,-!
0 5000
10000
Initial per Capita
Income
Figure Impact
of
$10 Billion
Rise
Annual
20000
15000
per Capita
30000
25000
(U.S. Dollars)
2
in Oil
Exports
on
Regime,
by Initial
Income
aThis figure shows the net predicted impact of a $10 billion rise in oil exports on the 0-10 variable of twenty million, country with a population by initial per capita income, for a hypothetical Regime, with no prior oil exports. Note the scale on the Y-axis is negative.
high
levels,
this test more
in
democracy test also The
suggests
the opposite: barrel for barrel, ones. countries than in oil-rich
oil harms
oil-poor cause greater dam implies that oil and mineral wealth ones to in in countries than rich 2). (see Figure age poor democracy a a income is whose the $800 per country Imagine capita year?about a level of Chad, Mozambique, and Yemen?with of twenty population an oil field that million and no oil exports. find prospectors Suppose of petroleum is exported. each year, all of which $10 billion produces The
new
oil would
boost per capita income (a prode simultaneously raise the Oil variable antidemocratic (an effect). The model that after five years the government would become predicts on the 0-10 less democratic, scale. A com losing about .93 democracy a state whose in income was initial per capita parable discovery if mean?would lose .54 the $1,720?the per capita in sample points; mocratic
effect)
and
come were $8,000?about same the level of Mexico and Malaysia?the oil field would be associated with a drop of just .16 in This pat Regime. tern is consistent with the observation that large oil discoveries appear
344
WORLD
to have no discernible
POLITICS
antidemocratic
effects
and the U.S., states, such as Norway, Britain, in countries. lize democracy poorer
in advanced but may
industrialized or destabi
harm
how general and robust these effects are, I carry out tests. First, to see whether the results are sensitive to the duration of the lag on the right-hand-side I run the same variables, on model all the explanatory variables (Table 3, col using one-year lags To
determine
five additional
umn 2). All of the variables remain significant, although the absolute on the lagged
of the coefficient
value
the absolute
regime type variable grows, and on the other of the coefficients
values and significance are reduced, perhaps artificially.60 to see whether other types of commodity Next, to the model: I add two variables democratization,
variables
sures the value
of all food
measures
the value
which
exports also inhibit Food, which mea of GDP, and Agriculture,
exports as a fraction of all nonfood agricultural
exports
as a frac
tion of GDP.As columns 3 and 4 of Table 3 show, the coefficients on Food andAgriculture are both positive?unlike Oil andMinerals, which are are consistent negative. These findings oil and other minerals impede democracy,
with
the rentier
but other primary less export produce
few or no rents, a larger fraction of the labor force?do
ties?which
generate the state, and employ
state thesis: commodi income
for
not.
The third test is designed to seewhether the model is heavily influ
enced most
of small states in the sample. Some by the inclusion on oil have small including populations, dependent
of the states Brunei
and
the Persian Gulf states of Bahrain, Kuwait, Qatar, and theUnited Arab Emirates; itwould not be surprising if they had a large influence on the magnitude and significance of the Oil variable. To determine this, I placed a dummy variable, Large States, in the model; itwas coded 0 if a states
was
population
below
one million
and
1 otherwise.
The
results
are displayed inTable 4, column 1.The coefficient on the population dummy is positive and significant at the 0.05 level, indicating that small
states do tend
sion has only
a
tiny
to be less democratic influence
on the Oil
than large ones; yet its inclu and Minerals coefficients and
leaves them highly significant. The Minerals tization
at whether the apparent effects of Oil and or historical to democra cultural by impediments are to that the Middle East and sub-Saharan Africa, specific
fourth
test
two
regions where variables dummy 60
looks
are caused
See Achen
(fn. 56).
these
states are most
to the regression,
I add two heavily concentrated. Mideast and SSAfrica, which were
DOES
OIL HINDER
345
DEMOCRACY? 4
Table
and Democracy51 Wealth variable is regime) (dependent
Resource
.255***
.209***
.227***
(.0203)
(.0205)
(.0203)
Regime
Oil
-.0333***
Minerals
(.00512)
-.0439**
-.0336***
(.00718)
.947***
.789***
(.105) Islam
-.0178***
OECD
1.41***
(.00761) .895***
(.117)
(.112)
-.00538
(.00209)
-.013***
(.0033)
(.00238)
1.6***
(.306) States
(.00557)
-.0265***
(.00802) Income (log)
Large
-.0138*
-.0209***
(.00511)
1.39***
(.31)
(.286)
.828*
(.406) Mideast
-3.65***
(.386) SSAfrica
-.998**
-1.62***
(.194)
(.2) Arabian
Peninsula
-3.74***
(.49) 2183 2183 2183 113 113 113 States -3133 -3086 -3100 likelihood Log * ** *** significant at the 0.05 level; significant at the 0.01 level; significant at the 0.001 level
Observations
aAll are
and
independent
in
parentheses run with Stata 6.0;
Each
control
below
corrected
is run with
regression
variables
the coefficients.
are entered Feasible
for first-order dummy
variables
with
Generalized
errors lags. Standard five-year Least regressions Squares
autocorrelation for every year
using
process. panel-specific covered by the data.
(but one)
coded 1 if the states were classified by theWorld Bank as residing in the lagged dependent variable these regions and 0 otherwise. While control
helps
for unspecified as summarized
crudely
be
SSAfrica
test for additional
effects?which country-level "the country's history"?Mideast or "the effects, regions region-level
might and history."
The results are listed in column 2 of Table 4. The coefficients for both Mideast and SSAfrica are large, negative, and highly significant. The
coefficients
on the 0*7 and Minerals
variables
are
again
reduced
but
remain highly significant. The Islam variable loses significance, due to its high correlation with theMideast variable (=.65).
WORLD POLITICS
346
For the final test, I use a new dummy, Arabian Peninsula, in place of theMideast dummy; itwas coded 1 for the seven states of theArabian Peninsula (Bahrain, Kuwait, Oman, Saudi Arabia, Qatar, the United Arab Emirates, andYemen) and 0 otherwise. Conceivably theMideast to capture the effects of is too broad, since it attempts residing dummy in a region that is socially and geologically diverse. The antidemocratic to the Arabian Penin effects of oil might be somewhat more restricted is dominated
sula, which
by monarchies,
sparsely
and en
populated,
dowed with spectacular oil wealth. Using Arabian Peninsula instead of Mideast
reduces
the problem
of collinearity
with
Islam,
although
Ara
bian Peninsula and Oil remain highly collinear (simple correlation =.74). Still, while including theArabian Peninsula dummy reduces the magnitude of the Oil coefficient by about 60 percent, Oil remains sig at the 0.05
nificant
These
level.
tests support both the validity and the generality of the oil
that a states re suggest the following: impedes-democracy claim.They on or to it make less democratic; either oil mineral liance exports tends
that this effect is not caused by other types of primary exports; that it is not
to the Arabian
limited
Peninsula,
to the Middle
East,
or to sub-Sa
haran Africa; and that it is not limited to small states.These findings are
consistent
generally
with
the theory
of the rentier
Area specialists might also feel vindicated
state.
in noting that in these
to include the vari democracy impediments and Arabian Peninsula, which represent the ac and cultural factors in each country, and in the that are not captured by in Arabian Peninsula and Mideast regions, or resource status. This underscores non-Western come, Islam, wealth, of case studies in explaining the critical importance regime types. tests
the most
powerful ables Mideast, Regimet? cumulation of historical
Causal Mechanisms To
test the three causal mechanisms variables,
intervening
lagged
a series of
I add to the basic model
one year. by Adding
new variables
reduces
the sample size from 2,183 observations to between 2,183 and 426 ob servations. ward
it becomes skewed to the sample shrinks, increasingly are and democratic, Western, wealthy, relatively a this problem, after sample bias. To minimize pronounced
As
states
that
introducing each running
the same
of the following reduced sample,
regressions, but without
I run a second
the intervening using If the intervening variable the two regressions. then compare and?if the Oil andMinerals should be statistically significant,
regression I variable. is valid, variables
it
DOES OIL HINDER DEMOCRACY?
347
are significant in the reduced sample?its inclusion should reduce the absolute values of the Oil andMinerals coefficients. This provides at least a crude
Rentier
test of some
of the causal mechanisms.
Effect
I use three indicators. For hypothesis, is the percentage the variable Taxes, which of revenue collected on taxes services, income, through goods, The effect taxation that states that capital gains. implies taxes selves through these assorted personal and corporate on more have higher values the Taxes variable) should be test the rentier
To
I use
effect
states
that
conversely, trade taxes, parastatals, be more authoritarian.
fund
grants, is constructed variable
by the International Monetary
government profits, and fund
them
(and hence
democratic; means other (such as through and right-of-way fees) should
themselves
external The
the taxation
from
data
collected
Fund and covers 104 of the 113 states
in the basic model. test
To measures cludes
the
I use Government
effect
which Consumption, as a of GDP; this in government percentage consumption all current spending for purchases of goods and services (includ spending
ingwages and salaries) by all levels of government. If the spending effect is valid, higher levels of government spending should result in less de mocracy. The data cover 104 states and are compiled by theWorld sta in turn collects from the OECD, national information Bank, which tistical organizations, central banks, andWorld Bank missions. is Government/GDP, third variable The which measures the share of GDP accounted in for by government 1985 international prices; activity, the data way
are from
to look that
Summers
and Heston.61
for a group-formation increase governments
as
This
final
effect.
Proponents (relative the formation
in size
imply are more to prevent they likely economy) are tions and social groups that independent that the absence of these groups will hinder
from
indicator of this
is one effect
to the domestic of civic
institu
the government, and to democracy.62
a transition
good indicators for civic institutions or social groups, this hy pothesis cannot be tested directly with regression analysis. Still, the
Without
an indirect
Government/GDPvariable
offers
(as a fraction cial groups will form.
of GDP),
ment's
size
test: the greater the govern so that independent
the less likely
61 "Penn World Robert and Alan Heston, Summers 5.6," http://cansim.epas. Tables, Version 1999 (consulted March 1,2000). utoronto.ca;5680/pwt/pwt.htm/, 62 Of course, a larger budget may not be the only cause of such government actions, but it is the only cause that can be linked to resource wealth in an obvious way.
348
WORLD
POLITICS
Table
5 Rentier Effect variable is regime) (dependent The
.259***
Regime
.243***
(.021)
Oil
(.0211)
-.0223***
(.0113)
(.00675)
.889***
.857***
(.104)
(.112)
(.106)
-.0165***
-.0191***
-.0161***
(.00218)
1.19***
(.00212)
1.57***
1.53***
(.314)
(.272) Taxes
-.0369***
(.00677)
(.00205) OECD
(.00511)
-.0463***
1.005***
Islam
-.0351***
(.00544)
-.0157
Income (log)
(.0203)
-.0323***
(.00647) Mineral
.251***
(.303)
.02***
(.00373) Government
-.0305***
(.00866)
Consumption
-.0332***
Government/GDP
1698 104 -2320
Observations States
Log likelihood *
(.00739) 2168 111 -3107
2121 110 -3036
*** at the 0.01 level; at the 0.001 level significant significant are entered with and variables variables control lags; intervening independent five-year are entered with (Taxes, Government Government/GDP) one-year lags. Stan Consumption, below the coefficients. Feasible Generalized Least dard errors are in parentheses Squares run with for first-order autocorrelation Stata 6.0; corrected using panel-specific regressions at the 0.05
significant
Each process. the data.
**
level;
is run with
regression
dummy
variables
for every
(but one)
year
covered
by
As Table 5 shows, the coefficient on Taxes is highly significant and positive: taxes are
as the rentier
effect
associated
implies, with more
and corporate higher personal More democratic government. a 17 in Oil the coef percent drop
strongly of Taxes produces over, the inclusion ficient, which implies that the taxation
effect may
the antidemocratic influence of Oil.63While also
runs
should
the other
be
way?that
in the opposite
63 The Minerals the mineral-exporting
account
type influences regime more direction: democratic
variable is not significant states.
in this sample, making
for part of
it is possible that causality
it difficult
taxation?it governments
to draw inferences
about
OIL HINDER
DOES
349
DEMOCRACY?
should be less disposed to fund themselves through personal and cor taxes, given their unpopularity. effect of taxes on regime types
porate The
term: when
Taxes
is introduced
into
out
turns
the model
to be
with
short strictly a two- or three
year lag, its coefficient quickly drops in size and loses significance. This that
implies
tend
people
tax increases to
respond
have only short-term effects on democracy: or to tax hikes not at all.64 right away
The Government Consumption variable is also highly significant in the hypothesized direction (Table 5, column 2).When Government Consumption is included in the model, Oil andMinerals drop slighdy, by 7 and 6 percent, respectively. The spending effect appears to last effect: the Government longer than the taxation has much the same effect on regime type after one. after These idence
results
variable
Consumption three years
as it does
are not
that regime
there is ev likely caused by endogeneity. While influences levels of government type consumption,
it is in the opposite direction found here: democratic governments tend to favor higher
levels of social
than
spending
their authoritarian
coun
terparts.65
Finally, Government/GDP is also highly significant with the hypothe sign: the larger the government, over the five years. mocracy following sized
the
less movement
Its inclusion
toward
has no effect
Oil variable but produces a 12 percent drop in theMinerals (Table 5, column 3). In short,
the results
are consistent
with
all three aspects
de
on the
variable
of the rentier
effect.
Repression I use
Effect
two variables
governments
causes to test the that resource wealth hypothesis more heavily against popular pressures.
to arm themselves
The first isMilitary/GNP, which measures the size of the military bud get as a fraction
of GNP. The
data were
Control andDisarmament Agency
originally
collected
by the Arms
(ACDA)of theU.S. government and
64 Note that other studies have found that a government's reliance on personal and corporate tax revenues is influenced by per capita income: poor states tend to rely on trade strongly and negatively taxes, rich ones on personal and corporate taxes. SeeWilliam Easterly and Sergio Rebelo, "Fiscal Pol icy and Economic Growth," fournal of 'Monetary Economics 32 (December 1993); Howell H. Zee, "Em World Development 24 (October 1996). Since per Tax Revenue Comparisons," pirics of Cross-Country the actual effect of Taxes on regime types is probably larger capita income is included in the model, than the coefficient in this regression suggests. 65 in Latin America, David and Wendy and Social Spending S. Brown Hunter, "Democracy 1999). 1980-92," American Political Science Review 93 (December
350 cover
WORLD POLITICS states between
101
1985
and
1995.66
Since
states
resource-rich
tend to have government budgets that are atypically large relative to the size of their economies, this is a better indicator than military spend as a fraction of government ing spending. is The second variable the size of Personnel, which measures Military a some as the military fraction of the labor force; it includes paramili tary forces
"if those
forces
training,
equipment,
resemble
or mission."
The
regular units in their organization, data are also from ACDA and are
available from 1985 to 1995 for 105 of the states in the database. Un like the this indicator helps measure, Military/GNP in military wages and the presence of conscription are Income When and Oil, Minerals, regressed
control across on
for variations states.
Military/GNP
di
recdy (with a five-year lag), the behavior of oil exporters and mineral exporters diverges. Oil exports are indeed positively and significantly correlated
with military
but mineral
exports
as the suggests; repression hypothesis associated with and significantly negatively
spending, are
military spending. Neither variable is significantly linkedwith Military Personnel.
When Military/GNP is placed in the basic model of regime types, its coefficient is negative andmarginally significant at the 0.10 level; its in clusion produces a 6 percent drop in the Oil coefficient (Table 6). The Military Personnel coefficient is negative and highly significant, al though it paradoxically induces a 7 percent rise in Oil. In both samples theMinerals coefficient is not significant and cannot be interpreted. Overall, it appears that oil wealth may be linked to higher levels of as the re in turn tends to spending, which impede democracy, of a similar pattern for effect suggests. But there is no evidence or nor to is evidence there wealth; support the claim that oil
military pression mineral
mineral wealth leads to higher levels of military personnel. do oil-rich governments Why Is it to repress popular militaries?
on their they do or is it a response to higher
as much
invest pressures,
as
levels of instability?To address this question I use data from the Polit
measures a uses to subjective Group, private firm that a 0-6 measure o? Eth risks for its clients. It produces gauge investment a nic Tensions, which measures "the degree of tension within country at ical Risk
tributable
Services
to racial,
nationality,
or
language
divisions."
Scores
are
available for 102 states between 1982 and 1997. Higher values indicate less ethnic 66 Since gressions
tension. When
added
to the model?first
the data cover only eleven years, the maximum to 1,642. drops from 3,752
number
of possible
separately, observations
then
for these re
DOES OIL HINDER DEMOCRACY? 6
Table The
Repression variable (dependent
Effect is regime)
.414***
Regime
.334***
(.032)
Oil
-.0679***
(.00566) .0169
Islam
(.00609)
-.00344
(.0179)
.848***
.822***
-.000964
(.0201) .824***
(.132)
(.145)
(.117)
-.0173***
-.0158***
-.0263***
(.00266) OECD
-.0517***
(.00632)
(.0272) Income (log)
.34***
(.0262)
(.0314)
-0591***
Minerals
351
(.00251)
(.00235) -.00168
-.071
(.332)
-.0957
(.355)
(.3)
-.0366
Military/GNP
(.0197) Military Personnel
-.09**
(.0304) Ethnic Tensions
-.0254
(.0485) 841 1167 874 101 States 105 102 -1293 -1642 -1228 likelihood Log * ** *** significant at the 0.05 level; significant at the 0.01 level; significant at the 0.001 level
Observations
aAll independent
and
control
ables
Military (Military/GNP, errors are in Standard parentheses regressions process. Each
Squares
specific ered by the data.
run with
are entered
variables Personnel,
Stata
regression
Ethnic
below
the
with
vari
five-year lags; intervening are entered with one-year coefficients. Feasible Generalized
Tensions)
for first-order 6.0; corrected is run with variables dummy
lags. Least
autocorrelation
using
for every year
(but one)
panel cov
together with Military/GNP, and finally controlling for ethnolinguistic is not statistically Ethnic Tensions variable signif In column other tensions caused by racial, na words, (Table 6, 3). so states do not explain why oil-rich tional, or language divisions spend on heavily repression. fractionalization?the icant
Modernization To mine
Effect
to deter I use eleven indicators hypothesis low levels of ed specialization, abnormally occupational can and urbanization health services, media participation, help
test the modernization whether
ucation,
352
WORLD POLITICS of democracy of indicators allows me
ernization
large of mod
s version
to test both versions
and earlier
theory
states. The
in the resource-rich
the dearth
explain number
Inglehart described by Lerner,
Deutsch,
and Lipset. to
and education Inglehart, occupational specialization measure links between economic growth and democracy. To at the number of men and women I in look occupational specialization as a sectors the economy's and (industrial) secondary tertiary (services) men women in the economically fraction of the and active population. According are the key
These
are drawn
data
from
the International
Labor
Organization
and
cover 76 of the 113 states used in the basic model. For
levels, I use figures on as a fraction in secondary school
educational
women
the enrollment
of men
of
the corresponding on college enrollment
at group in the population large and figures fraction of the population. Both data sets are collected
by national
and age as a gov
ernments and assembled by theUnited Nations Educational, Scientific, and Cultural available
(UNESCO). Figures countries; figures countries.
Organization for forty-eight
are available
for ninety-six
on on
enrollment secondary are enrollment college
of modernization that improve Early proponents theory suggested can also lead to democratiza ments in a populations health physical has argued that as a tion.67 More basic recently Inglehart populations turn nutritional and health needs are satisfied, will they increasingly a to for desire and values, "postmaterialist" including self-expression shift, in turn, will facilitate more dem ocratic government.68 Earlier scholars measured the quality of a popu I use lations health by using the number of doctors per capita. Here at birth, a measure life expectancy that also accounts for nutrition lev across of health services els and the distribution the population. The individual
freedom;
underlying
data
are
this value
compiled
by several UN
and cover ninety
agencies
states.
In Lipset's
classic
analysis,
the greater
a
society's
level of "media par
ticipation/' the more likely it is to be democratic.69 Lipset measured media
radios, and newspaper using telephones, participation copies per measure To these indicators both the number capita. slightly, I update on of telephone and televisions mainlines per capita. Data telephone are collected mainlines and televisions the International Telecom by 67 Daniel Lerner, The Passing 68 Inglehart (fn. 1). 69 Lipset (fn. 38).
of Traditional
Society (New York: Free Press,
1958); Deutsch
(fn. 38).
353
DOES OIL HINDER DEMOCRACY? 7
Table
Effect3 is regime)
The Modernization variable (dependent
.529***
.462***
.513***
.604***
(.0316) O? -.0182
(.0408)
(.0336)
(.0324)
Regime
-.116
(.0221) .146*
Minerals
Income (log)
(.0666) -.251
Islam
-.0121
(.305) (.0082) OECD
in
Women
?
(.271)
(.343)
(.344)
-.0232***
-.000534
(.372)
(.419)
?
?
(.0166) ? .0685***
?
(.0155)
Services in
Women
.391
1.13**
?
in
(.0104)
(.00652)
(.00545)
.0814***
Industry Men
3.8
-.0154**
(.432)
.0733*
(.0714)
(.0657) -.408
.565*
(.419) (.0143)
(.0234) .115
(.0207) .0952
(.0635)
.652
in
-.0315
-.0187
(.0202) .112
.752*
Industry
Men
4
3
12
?
?
?
-.0185***
(.00512)
Services
States 75
***
**
*
-921
-835
-772
Log likelihood_-878
629 76
622 76
615 75
626
Observations
at the 0.01 level; at the 0.001 at the 0.05 level; level significant significant significant are entered with vari aAll and control variables intervening five-year lags; independent are entered in Services) Men in in Services, Women ables {Men in Industry, Women Industry, are errors in Feasible Gen below the coefficients. Standard with one-year parentheses lags. run with autocorrela for first-order Stata 6.0; corrected eralized Least Squares regressions for every is run with variables Each tion using panel-specific process. regression dummy year (but one) covered by the data.
munications tively,
Union
and are available
and cover virtually
all country
for years
113
and
110
in the data
states, set.
respec
Finally, Lipset also suggested that higher levels of urbanization will I use the To measure urbanization levels of democracy. lead to higher in urban areas. The data, of a state's population currently living
fraction
collected by theUnited Nations, are available for all 113 states. The
results
from
these
regressions
are
reported
in Tables
7, 8, and 9.
All of the variables measuring occupational specialization are highly
WORLD POLITICS
354
Table
8
The Modernization variable (dependent
.378***
.378***
(.0449)
(.0451)
Regime
Oil
Effect is regime)
-.0158
Minerals
-.0168
-.033***
(.00966) .0251
(.00952) .0255
(.00991) .0517
(.0431)
(.0433)
(.0325)
Income (log)
.258
Islam
.34***
(.0334)
.364
.678***
(.296)
(.29)
(.19)
-.0393***
-.0385***
-.0348***
(.00507) OECD
(.00479)
.159
.187
(.345)
(.336)
Male Secondary
(.00407) -.0759
(.436)
.004
Enrollment
(.00856) .000812
Female
Secondary Enrollment
(.00882)
College Enrollment
-.00289
(.0105) 426 48
Observations States
688 96
426 48
-566 -563 -1109 Log likelihood * ** *** significant at the 0.05 level; significant at the 0.01 level; significant at the 0.001 level aAll independent ables
(Male
tered with
Secondary one-year Least
Generalized
and
control
variables
Enrollment,
Female
lags. Standard
errors
are entered
using
every year
significant
and positively of modernization
proponents wealth influence
five-year
lags;
vari
intervening
are en
Secondary College Enrollment) are in below the coefficients. Feasible parentheses run with auto for first-order Stata 6.0; corrected
Squares regressions Each process. panel-specific (but one) covered by the data.
correlation
with
Enrollment,
associated
regression
with
theory. The
is run with
dummy
variables
for
as democracy, by predicted evidence that oil and mineral
is somewhat however, occupational specialization, variables measuring Ufe expectancy, urbaniza education, are measure not and while televisions the of tion, per capita significant,
weak.70
The
70 is significantly in these reduced samples, Oil nor Minerals Neither correlated with democracy are on each which makes it hard to be confident about these results.When Oil anaMinerals regressed of the four variables for occupational (with Income and Islam included as control vari specialization in Industry but positively correlated ables), the results are mixed: Oil is negatively correlated with Men with Women in Industry; Minerals is not significantly correlated with Men in Industry and is negatively, but weakly, linked to Women in Industry.
DOES OIL HINDER DEMOCRACY? Table
355
9
The Modernization variable (dependent
Effect is regime) 4
Regime
.194***
.196***
.413***
.253***
(.0232)
(.0225)
(.0516) .0247
(.0203)
Oil
-.0463***
Minerals
-.00929
-.04***
(.00609)
Income (log)
(.00551)
(.016)
(.0152) .882***
(.119)
(.134)
Islam
-.0194***
-.023***
OECD
2.96***
Telephones
-.00543**
(.00214)
(.00509) -.0441***
(.0605)
(.008)
1.07***
.983***
(.315)
(.149)
-.0104
(.00231)
-.0174***
(.0168)
(.00213)
-.041
1.75***
(.351)
(.482)
(.039) -.0376
-.0085
1.24***
-.0346***
1.51***
(.31)
(.412)
(.00118) TVs
-.00096
(.00079) Life
.00378 (.0616)
Expectancy Urban
-.00278
1830 113 -2830
Observations
1831 110 -2676
(.005) 2183 113
777
103 -3133 -857 likelihood Log * ** *** significant at the 0.001 level significant at the 0.05 level; significant at the 0.01 level; States
are entered with vari and control variables five-year lags; intervening er are entered with Urban) one-year lags. Standard {Telephones, TVs, Life Expectancy, rors are in Least below the coefficients. Feasible Generalized parentheses Squares run with for first-order S tata 6.0; corrected autocorrelation regressions using panel-specific is run with for every year (but one) covered by variables Each process. regression dummy the data. aAll independent
ables
telephones per capita is highly significant but negatively correlatedwith democracy. are at least two ways to There interpret these results. One valid but that occupational modernization effect is essentially is the only real causal mechanism behind it, with A second relates of modernization being epiphenomenal. countries both the modernization is that in resource-rich ization
spending into the
effect industry
occur and
simultaneously: service sectors;
is that the
special the other cor interpretation effect and the
are drawn few people relatively to its thanks yet large revenues,
356
WORLD POLITICS
the government other services. The ices yet
can
is politically
subsidize health care, and education, generously result is that the public enjoys generous social serv two antidemocratic forces: a lack of hampered by
and a government that uses its fiscal powers occupational specialization to dissent. dampen consistent The results of these tests are at least weakly with each of the three causal mechanisms. Collectively, they provide quantitative
backing for the rentier effects described by a generation ofMideast in the case
for the repression effects observed specialists, and for a modified form of the modernization on data that are rely incomplete sults should be treated as suggestive,
tests
thesis.
and potentially not conclusive.
studies
above, the causality so the re biased,
Still,
Conclusion This
article
has four main
findings.
First,
the oil-impedes-democracy
claim is both valid and statistically robust; in other words, oil does hurt to in poor oil does greater damage democracy. Moreover, democracy states than in rich ones, and a given rise in oil exports will do more harm ones. Hence, states than in oil-rich in oil inhibits democracy oil-poor even when are in poor states. exports relatively small, particularly
Second, the harmful influence of oil is not restricted to theMiddle East.
Oil wealth has probably made democratization harder it may well like Indonesia, Malaysia, and Nigeria; Mexico, same affect on the oil-rich states of Central Asia.
The
in states have
the
third finding is that nonfiiel mineral wealth also impedes de
mocratization.
While
the major
mineral
Mideast, exporters major the Americas; this group includes
oil exporters are scattered many
are concentrated
in the
across Africa, Asia, and states where toward progress
democracy has been halting or elusive, including Angola, Chile, the Democratic Each scholars
of Congo, Republic runs of these findings that the antidemocratic
and Peru. Cambodia, counter to the of earlier assumptions effects of oil?if they existed?were
states that were East, that they influenced only on oil, and that not extend to the did wholly they dependent states. mineral-rich restricted
to the Middle
almost
is that there is at least tentative The fourth finding that link oil and authoritarianism: causal mechanisms through dampen
support for three a rentier effect,
use low tax rates and to governments high spending a for democracy; effect, by which govern pressures repression
which
ments build up their internal security forces to ward off democratic
357
DOES OIL HINDER DEMOCRACY? and a modernization pressures; move into industrial lation to to likely push thoritarianism from
the failure of the popu effect, in which and service sector jobs renders them less The links between mineral wealth and au
for democracy. are more elusive:
a rentier
effect
a
but not
that they are afflicted
evidence
appear to suffer exporters effect, and there is only weak
the mineral
repression a by modernization
effect.
Collectively, these findings should help vindicate two very different theories
of comparative politics: modernization out of favor in the 1970s and 1980s made
falling
after theory, which a strong comeback in
the 1990s; and the theory of the rentier state, which has long been championed byMiddle East area specialists but overlooked by scholars of democratization.
They also highlight the value of bringing cross-national quantitative contact
into closer
studies
with
area studies. Global
studies
of democ
racy have generally overlooked theMideast, a practice that is difficult to justifymethodologically (since it arbitrarily truncates the researcher s sample of states) and one that has contributed to a belief that theMid and culture of the region is sui generis. Of course, the history on the Mideast are note the enormous coefficient exceptional:
dle East Mideast
dummy variable inTable 4. But excluding Middle
Eastern states from
area the gap between only widen science. It also deprives mainstream po area studies scholars? litical science of the many insights developed by that, like the oil-impedes-democracy claim, may turn out to insights have general applications. of democracy large-N studies and the rest of political studies
can
Finally, these findings have implications for the fate of resource-rich states
across
the developing
world.
of the world
Many
smost
troubled
states have high levels of oil and mineral wealth. Earlier studies have shown
that resource wealth
tends
to reduce
economic
growth
and to in
crease the likelihood of civil war. This article suggests there is a third to "resource curse": authoritarian rule. component a "re interact in pernicious three effects may These ways, creating source may be less able to resolve do governments trap." Authoritarian to suffer from civil war. Slow likely unrest to resolve; civil wars, in domestic tougher growth may make is nothing economic havoc. There inevitable about the re turn, wreak
mestic
source
conflicts
curse:
and hence
states
more
like Malaysia,
Chile,
and Botswana
have done
rela
tively well despite their oil and mineral wealth. Yet most others have found?like pected
source
King Midas?that of grief.
their
resource wealth
can be an unex
358
WORLD
Appendix is a 0-10
Regime
variable
POLITICS
of Variables71
1:Definition indicating
a
country's
regime
type, with
0 as a
perfect autocracy and 10 a full democracy. It is taken from the Polity 98 a 0-10 indicator for by Gurr and Jaggers, who assign compiled level of autocracy and level of democracy.72 Each is a composite of measure are re the chief variables that executives way underlying office elections, cruited, whether they gain through competitive set
data
both
whether
are executive office, and whether may obtain they actors. other and accountable to, by, Following Londregan I transform into a single indicator by these two measures
nonelites
constrained and Poole,
from the democracy measure and by the autocracy measure to 10 scale as a 0 to 10 scale.73 For the six -10 the resulting rescaling states with for which Gurr than one million and greater populations subtracting
Jaggers
offer
no
indicators
(Austria,
Cameroon,
Democratic
Republic
of Congo, Libya, Sierra Leone, and Switzerland), I use data from Free dom House (1972-98) instead, summing their measures for "political and converting the results to the 0-10 scale. Income is the natural log of real per capita GDP, in current inter Log of the data come from Summers and Heston; national dollars. Most rights"
and "civil liberties"
missing values have been imputed using data from theWorld Oil
Bank.74
of GDP. fuels as a percentage natural gas, and coal, as classi 3. Following of Sachs the practice
is the export value of mineral-based Mineral-based fuels include petroleum, SITC revision
1, section I corrected andWarner, the export figures to reflect net exports, since both states are
fied
under
materials
extracted
in nearby
states.75 The
for
Singapore
and Trinidad
raw transshipment points for values for both states were
set at 0.01. as a is the export value of nonfiiel minerals of percentage ores sec SITC revision it under includes all and metals classified GDP; 1, I cor tions 27,28, and 68. Following the practice of Sachs andWarner, Minerals
rected the export figures for Singapore and Trinidad exports, tracted
since both in nearby
states are states.76 The
transshipment points values for both states were
71 Unless otherwise indicated, the data below were derived from World Bank, 1999). Indicators," CDHROM (Washington, D.C.: World 72 Gurr and Jaggers (fn. 42) 73 and Poole (fn. 43). 74Londregan Summers and Heston (fh. 61). 75 Sachs andWarner (fh. 3,1999). 76 Ibid.
to reflect net
for raw materials
Bank,
ex
set at 0.01.
"World Development
DOES OIL HINDER DEMOCRACY?
359
Islam is the percentage of the population whose professed religious affiliation in 1970 was Muslim.77 OECDis a dummy variable coded 1 for the following states and 0 for all others:
Australia,
Austria,
Canada, Denmark, Finland, Belgium, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, New Netherlands, Zealand, Sweden, Norway, Portugal, Spain,
Switzerland, United Kingdom, United States. is the export value Agriculture a as of GDP. This als, percentage
raw materi
of all nonfood includes
agricultural all commodities
as
classified
falling in SITCrevision 1, section 2 (excluding divisions 22,27, and 28). Food
is the export
value
of all edible
commodities,
agricultural
as a
percentage of GDP.This includes all commodities classified as falling in SITC sections Large
22. 0,1, and 4, and division a States is dummy variable coded
1 for states with
populations
over one million at any point between 1971 and 1997, and 0 otherwise. Mideast is a dummy variable coded 1 for the following states and 0 otherwise: Bahrain, Iran, Iraq, Israel, Jordan, Kuwait, Algeria, Egypt, Lebanon, Morocco, Oman, Qatar, Saudi Arabia, Libya, Syria, Tunisia, United Arab Emirates, and Yemen.
a SSAfrica is dummy variable coded 1 for states classified by the World Bank as residing in sub-Saharan Africa and 0 otherwise. Arabian
Peninsula
is a dummy
variable
coded
1 for the states on the
Saudi Arabian peninsula (Bahrain, Kuwait, Oman, Saudi Arabia, Qatar, theUnited Arab Emirates, andYemen) and 0 otherwise. Men in Industry and Women in Industry indicate the fraction of the totalworking population of each gender group working in activities de fined by the ILOas "industry."This includes mining and quarrying (in cluding oil production), manufacturing, to construction, major corresponding
electricity, divisions 2-5
ILO as "services." Services by the and hotels; and restaurants transport,
include wholesale
and and water, revision 2) or (iSIC
gas
tabulation categories C-F (iSIC revision 3). The data are compiled by theWorld Banks Development Data Group using an ILOdatabase cor responding to table 2a in its Yearbook ofLabour Statistics. Men in Services and Women in Services indicate the fraction of the totalworking population of each gender group working in activities de
fined
insurance, nancing, social, and personal 77 David
B Barrett,
real estate, services,
and retail trade
fi storage, and communications; and business services; and community,
corresponding
ed., World Christian Encyclopedia
to
major
(New York: Oxford
divisions
University
6-9
Press,
(ISIC 1982).
WORLD POLITICS
360
revision 2) or tabulation categories G-P (ISICrevision 3). The data are compiled by theWorld Banks Development Data Group using an ILO to table 2a in its Yearbook Labour Statistics. corresponding of Enrollment and Female Enrollment indicate Secondary Secondary the fraction of males in secondary and females enrolled school, relative
database Male
to their numbers
in the population.
The
data are reported
to the United
Nations Educational, Scientific, and Cultural Organization by national
(UNESCO)
authorities.
education
CollegeEnrollment indicates the fraction of the population enrolled in data are reported
college. The ties.
to UNESCO
by national
education
authori
Life Expectancy indicates the life expectancy at birth of both males and females. The underlying figures are from the United Nations Depart ment of Economic and Social Affairs, Population andVital Statistics Re and health
port; demographic
from national
surveys
sources;
and United
Nations Children's Fund (UNICEF), The State of theWorld'sChildren, 1999. Urban
is the midyear population to the United and reported
country
of areas defined Nations,
as urban
in each
as a fraction
expressed
of
the total population. The data are from from theUnited Nations, World Urbanization Telephones
The 1996 Revision. Prospects: is the number of telephone
mainlines
(that
is, separate
lines to a given household or firm) per thousand people. The data are derived from the International Telecommunication Union (ITU),World Telecommunication TVs
Development Report. is the number of televisions per
an annual
sent to member
thousand countries
questionnaire from the ITU, World Telecommunication revenue is the percentage of government
are derived Taxes on
goods,
services,
income,
lected by the IMF.
profits,
and capital
to people, according ITU.The data the by Development Report. taxes raised through
gains. The
data
are col
as a percentage of GDP, includes expressed Consumption, "all current expenditures for purchases of goods and services by all lev most government els of government, It also in excluding enterprises. on national cludes capital defense and expenditure security." The data are collected from the OECD and from national statistical organizations Government
and central banks by visiting and residentWorld are published by theWorld Bank. Government/GDP
is the
share of GDP accounted
Bank missions; they for by government
activity, in 1985 international prices. The data are from the PennWorld Tables. Military/GNP
measures
the size of the military
budget
as a fraction
of
DOES OIL HINDER DEMOCRACY?
361
GNP.The data cover 1985-95; theywere originally collected by theArms Control andDisarmament Agency (ACDA)of theU.S. government. as a percentage Personnel measures the size of the military Military some of the labor force; it includes "if those forces forces paramilitary or in units resemble their training, regular organization, equipment, mission." Ethnic gree
The
data are also from ACDA and cover
Tensions
of tension
is a 0-6
within
a
interval-level country
attributable
1985-95. that measures
variable
to racial,
"the de or
nationality,
language divisions." The data cover 97 states between 1982 and 1997; the codings are carried out by a private firm, the Political Risk Services Group, and published in their monthly International Country Risk are also available as the IRIS-3 computer The database. they mean into the data have been annual data by taking monthly changed of the twelve monthly values. Guide;
2: Appendix of Variables
Summary Variable
Log Income Minerals
Agriculture Men in Industry Women in Industry Men in Services Women in Services Male Secondary Female Secondary College Life Expectancy Telephones
Govt. Consumption Government/GDP Military/GNP Military Personnel Ethnic Tensions
Obs.
3752 Regime 3316 2322 Oil 2865 OECD 4528 4336 Islam Food 2511 2504 814 798 810 813 607 607 1272 4372 Urban 1527 3129 TVs 3040 Taxes 2325 3538 2277 1298 1440 1739
Mean
Std. Dev.
Min
0 4.48 3.79 1.2 7.45 4.53 14.1 5.5 115.60 055.1 5.8 2.25 .369 0 .163 36.6 0 25 6.23 45.9 0 5.73 1.68 031.6 2.88 29.4 12.7 8.9950.2 0 15.5 14.3 39 5 52 910025.6 327.9 57.7 1.3 5829.9 16.9 16.9 46.1 100 25 2.24 31.2 11.7 62.5 106 154 169 0 838 151 101 0 18.7 50.9 15.2 6.51 0 11.9 23.8 0102 6.64 4.36 1.84 29.6 0 2.6 0 3.791 1.633
Max
10 10.43
1 99.7
.4
66.9 69.3
.1
98.6 98.5 97.7
.1
79.8 691
.897
76.2 91.2
6