THE POLITICAL ECONOMY OF LIBERAL
DEMOCRACY
Dani Rodrik
December 2015
Based on: “The Political Economy of Liberal Democracy,” NBER Working Paper No. 21540,
September 2015 (with Sharun Mukand)
Democracy has won – but what kind of
democracy?
Figure 1: Numbers of democracies and non-democracies since 1800
Data are from Polity IV (http://www.systemicpeace.org/inscrdata.html). “Democracies” are countries that receive a
score of 7 or higher in the Polity’s democ indicator (which takes values between 0 and 10), while “non-democracies”
are countries with a score below 7.
.4
.6
.8
1
1.2
1.4
Distribution of electoral and civil rights
0
.2
.4
.6
.8
1
x
electoral_rights
civil_rights
Source: Based on Freedom House raw scores for 196 countries. “Electoral rights” refer to free and fair elections
(A1+A2); “civil rights” combine measures of independent judiciary (F1), rule of law (F2), and equal treatment (F4).
Distribution of electoral and civil rights
.4
.6
.8
1
1.2
1.4
modal electoral
rights are quite
high, while modal
civil rights are on
the low side
0
.2
.4
.6
.8
1
x
electoral_rights
civil_rights
Source: Based on Freedom House raw scores for 196 countries. “Electoral rights” refer to free and fair elections
(A1+A2); “civil rights” combine measures of independent judiciary (F1), rule of law (F2), and equal treatment (F4).
What we do
• provide an analytical taxonomy of political regimes
• distinguishing, in particular, between “liberal” and “electoral
democracy”
• propose a particular approach to modeling liberal politics
• focusing on absence of discrimination in provision of public goods,
broadly construed
• examine, formally, circumstances under which LDs might
emerge
• emphasizing the role of income and identity cleavages
• relate theoretical findings to empirical cases
Three sets of rights
• Property rights protect asset holders and investors
against expropriation by the state or other groups.
• Political rights guarantee free and fair electoral contests
and allow the winners of such contests to determine
policy subject to the constraints established by other
rights (when provided).
• Civil rights ensure equality before the law – i.e. nondiscrimination in the provision of public goods such as
justice, security, education and health.
Electoral versus Liberal Democracy
• ED = property rights + political rights
• LD = ED + civil rights
A taxonomy of political regimes
property rights
no
no
civil rights
yes
no
yes
political rights
political rights
yes
no
Yes
A taxonomy of political regimes
property rights
no
no
yes
political rights
political rights
yes
no
Yes
no
(5)
right-wing autocracy
(6)
electoral/illiberal
democracy
yes
(7)
liberal autocracy
(8)
liberal democracy
civil rights
A taxonomy of political regimes:
illustrations
property rights
no
no
yes
political rights
political rights
yes
no
Yes
no
(5)
right-wing autocracy
(6)
electoral/illiberal
democracy
Argentina, Croatia,
Turkey, Ukraine,…
(n=45)
yes
(7)
liberal autocracy
Monaco
(n=1)
(8)
liberal democracy
Canada, Chile, S.
Korea, Uruguay,…
(n=46)
civil rights
Source: Based on Freedom House raw scores for 196 countries. The cutoff for electoral and liberal democracies
is 0.8 on a [0,1] scale of electoral and civil rights as defined earlier. “Electoral rights” refer to free and fair elections
(A1+A2); “civil rights” combine measures of independent judiciary (F1), rule of law (F2), and equal treatment (F4).
A taxonomy of political regimes
property rights
no
yes
political rights
political rights
no
yes
no
Yes
no
(1)
personal
dictatorship or
anarchy
(2)
dictatorship of
the proletariat
(5)
right-wing autocracy
(6)
electoral/illiberal
democracy
yes
(3)
n.a.
(4)
democratic
communism
(7)
liberal autocracy
(8)
liberal democracy
civil rights
Three groups in society
• A propertied elite, whose primary objective is to keep and
accumulate their assets (property rights);
• A majority, who want electoral power so they can choose
policies that improve their economic conditions (political
rights);
• A minority (ethnic, linguistic, regional, ideological), who
desire equality under the law and the right not to be
discriminated against in jobs, education, etc. (civil rights).
Note the two cleavages;
• the income/class cleavage
• the “identity” cleavage
(elite versus non-elite)
(majority versus minority)
An immediate result
• LD is an unlikely outcome of any “democratic settlement”
between the elite (who have the resources) and the
majority (who have the numbers)
• so what’s surprising is not how rare LD is, but that it exists at all
• Why this has been overlooked:
• in (formal) political economy: tendency to focus on elite-non-elite
cleavage
• in history of liberalism: failure to ask why propertied elite should
want to protect minority rights in addition to theirs (i.e., property
rights), so conflation of property with civil rights
When might LD emerge? The exceptions that
prove the rule
• Weak or non-existent identity cleavages
• Japan or South Korea (after late 1980s)
• Elite shares identity with minority
• South Africa (after 1994)
• Multiple identity cleavages
• Lebanon (until 1975)
• More can be said using explicit formal structure
• Main result: LD requires both moderate income inequality and weak
identity cleavage
More formal structure (1)
• members of each group 𝑖 ∈ 𝑒, 𝑎, 𝑏 derive utility from their (after-tax)
income 𝑦𝑖 and from consuming a public good 𝜋𝑖 .
𝑢𝑖 = 𝑦𝑖 + 𝜋𝑖 .
• pre-tax/transfer shares of the elite and non-elite given by 𝛼 and (1-𝛼),
respectively.
• elite constitute a negligible share of the population but control more
1
than half of pre-tax/transfer output (𝛼 > ).
2
• non-elite are split between a majority and a minority, with population
1
2
shares n and (1-n), respectively (n > ).
• gap between 𝛼 and
1
2
is a measure of the class (income) cleavage.
More formal structure (2)
• we model the identity cleavage by assuming groups exhibit
differences in the type of public good they prefer.
• type of public goods is indexed by 𝜃 ∈ 0,1
• three groups’ ideal types are given by 𝜃𝑖 , 𝑖 ∈ 𝑒, 𝑎, 𝑏
• the deadweight loss associated with the provision of public goods
increases with the level of expenditures and the gap (from the
perspective of each group) between the type that is provided and the
preferred type.
• denoting total expenditure on the public good by 𝑟, the utility derived
from the public good is thus expressed as
𝜋𝑖 = 𝑟 − 1 + 𝜃𝑖 − 𝜃
𝛾 2
𝑟
2
• we normalize the majority’s preferred public good by taking 𝜃𝑎 = 1.
More formal structure (3)
• the political regime determines
(i) how the public good is financed (whether through general taxation or the
extraction of a surplus from the non-elite),
(ii) the level of expenditures on the public good, and
(iii) the type of public good provided.
• in right-wing autocracy (RA), the elite make all these decisions
• in liberal autocracy (LA), the elite remain in the driving seat, but they
cannot discriminate against any particular group in taxation or the
nature of public good
• in electoral democracy (ED), the majority selects an economy-wide
tax rate and chooses the type of public good, disregarding the
minority’s wishes completely.
• in liberal democracy (LD), the majority retains control over the tax
rate, but they cannot discriminate against the minority.
• so they must provide a public good which lies somewhere in between the majority
and minority’s ideal types
Taxes and public goods in different
political regimes
political rights
no
civil
rights
yes
no
yes
(5)
right-wing autocracy (RA)
(6)
electoral democracy (ED)
𝜎 𝑅𝐴 =
1
𝛾(1 − 𝛼)
𝜏 𝐸𝐷 =
𝛼
𝛾
𝜃 = 𝜃𝑒
𝜃 = 𝜃𝑎 = 1
(7)
liberal autocracy (LA)
(8)
liberal democracy (LD)
𝑡 𝐿𝐴
1−𝛼
=
𝛾
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
𝜏 𝐿𝐷
𝛼
=
𝛾(2 − 𝜃)
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
ED versus LD: nature of public goods
political rights
no
civil
rights
yes
no
yes
(5)
right-wing autocracy (RA)
(6)
electoral democracy (ED)
𝜎 𝑅𝐴 =
1
𝛾(1 − 𝛼)
𝜏 𝐸𝐷 =
𝛼
𝛾
𝜃 = 𝜃𝑒
𝜃 = 𝜃𝑎 = 1
(7)
liberal autocracy (LA)
(8)
liberal democracy (LD)
𝑡 𝐿𝐴
1−𝛼
=
𝛾
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
𝜏 𝐿𝐷
𝛼
=
𝛾(2 − 𝜃)
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
ED versus LD: taxes
political rights
no
civil
rights
yes
no
yes
(5)
right-wing autocracy (RA)
(6)
electoral democracy (ED)
𝜎 𝑅𝐴 =
1
𝛾(1 − 𝛼)
𝜏 𝐸𝐷 =
𝛼
𝛾
𝜃 = 𝜃𝑒
𝜃 = 𝜃𝑎 = 1
(7)
liberal autocracy (LA)
(8)
liberal democracy (LD)
𝑡 𝐿𝐴
1−𝛼
=
𝛾
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
𝜏 𝐸𝐷 ≥ 𝜏 𝐿𝐷 since 𝜃 ≤ 1
𝜏 𝐿𝐷
𝛼
=
𝛾(2 − 𝜃)
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
Payoffs to groups in different political
regimes
political rights
no
yes
(5)
right-wing autocracy (RA)
(6)
electoral democracy (ED)
𝑢𝑒 = 𝛼 +
no
civil
rights
1
2𝛾
𝑢𝑎 = 1 − 𝛼 −
1
𝛾
𝑢𝑎 = 1 − 𝛼
𝑢𝑏 = 1 − 𝛼 −
1
𝛾
𝑢𝑏 = 1 − 𝛼 +
(7)
liberal autocracy (LA)
𝑢𝑒 = 𝛼 +
yes
𝛼
𝜃
+ 𝑒 𝛼2
𝛾
2𝛾
1
+ 𝛼2
2𝛾
𝑢𝑒 = 𝛼 − 2𝛼 − 1
𝜃𝑏 2
𝛼
2𝛾
(8)
liberal democracy (LD)
1−𝛼 2
2𝛾
𝑢𝑒 = 𝛼 −
1−𝛼
2𝛾
1−𝛼
1−𝛼
2𝛾
𝑢𝑎 = 1 − 𝛼 +
3𝛼 − 1 − 1 − 𝜃𝑒 1 − 𝛼
𝑢𝑏 = 1 − 𝛼 +
3𝛼 − 1 − 𝜃𝑏 − 𝜃𝑒
2𝛼−1 𝛼
(2−𝜃) 𝛾
+
𝑢𝑎 = 1 − 𝛼 +
𝑢𝑏 = 1 − 𝛼 +
3−2𝜃− 𝜃𝑒 −𝜃
2𝛾(2−𝜃)2
1
𝛼2
2𝛾(2−𝜃)
𝛼2
3 − 3𝜃 + 𝜃𝑏
1
𝛼2
2−𝜃
2𝛾(2 − 𝜃)
Payoffs to groups in LD versus ED
• Majority:
• Minority:
• Elite:
𝑢𝑎𝐿𝐷 < 𝑢𝑎𝐸𝐷
𝑢𝑏𝐿𝐷 > 𝑢𝑏𝐸𝐷
𝑢𝑒𝐿𝐷 > 𝑢𝑒𝐸𝐷
𝑢𝑒𝐿𝐷 < 𝑢𝑒𝐸𝐷
(unambiguous)
(unambiguous)
if elites share identity with
minority (e.g., 𝜃𝑒 ≈ 0)
if elites share identity with
majority (𝜃𝑒 ≈ 1) and
income/class cleavage not too
deep (12 < 𝛼 < 23)
A specific game
• status quo: Right-wing Autocracy (RA)
• shock makes RA unsustainable, in the sense that
expected utility to majority from revolution exceeds utility
under RA, i.e. 𝑢𝑎𝑅𝐴 = 𝑢𝑏𝑅𝐴 < 𝜌𝑢𝑎𝐷𝑃 = 𝜌𝑢𝑏𝐷𝑃
[𝜌 is probability revolution succeeds; otherwise majority gets payoff of 0]
• elite move first and offer a regime in the set LA, ED, LD
• majority move next, and they either accept the regime
offered, or they mount a revolution
• finally, minority move last, and they decide either to join
the revolution or to stay put
Results
• The parameter space under which ED emerges is larger
than (and encompasses) the parameter space under
which LD emerges.
• Proposition 1. There exist parameter combinations under which ED
will emerge and LD will not. The reverse is not true.
• Under the specific parameterization in the paper:
• Proposition 2. The equilibrium configuration of the political regimes
is as follows:
when 𝛼 ≤ 𝛼𝑎𝑚𝑖𝑛 , the elite offer LA and the majority accepts it;
ii. when 𝛼𝑎𝑚𝑖𝑛 < 𝛼 ≤ 𝛼𝑎𝑖𝑛𝑡 , the elite offer LD and the majority accepts it;
iii. when 𝛼𝑎𝑖𝑛𝑡 < 𝛼 ≤ 𝛼𝑎𝑚𝑎𝑥 , the elite offer ED and the majority accepts it;
iv. when 𝛼 > 𝛼𝑎𝑚𝑎𝑥 , the majority mount a revolution regardless of what
the elite offer.
i.
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑎𝑥
𝛼𝑎𝑚𝑖𝑛
Note: the greater the identity cleavage (𝜃𝑎 − 𝜃𝑏 ), the larger the difference
between 𝛼𝑎𝑖𝑛𝑡 and 𝛼𝑎𝑚𝑎𝑥
Equilibrium regimes
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑖𝑛
𝛼𝑎𝑚𝑎𝑥
Equilibrium regimes
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑖𝑛
LA
𝛼𝑎𝑚𝑎𝑥
Equilibrium regimes
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑎𝑥
𝛼𝑎𝑚𝑖𝑛
LA
Revolution
Equilibrium regimes
𝛼𝑎𝑚𝑎𝑥
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑖𝑛
LA
ED
Revolution
Equilibrium regimes
𝛼𝑎𝑚𝑎𝑥
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑖𝑛
LA
LD
ED
Revolution
Equilibrium regimes
𝛼𝑎𝑚𝑎𝑥
𝛼𝑎𝑖𝑛𝑡
𝛼𝑎𝑚𝑖𝑛
LA
LD
ED
Revolution
Note: the emergence of liberal democracy requires both mild inequality (low 𝛼)
and the absence of large identity cleavages (proximity between 𝜃𝑎 and 𝜃𝑏 ).
Discussion
• region in which LD emerges as equilibrium is squeezed from below by
the availability of LA (which satisfies the elite’s incentive constraint)
and from above by ED (which satisfies the majority’s participation
constraint)
• Size of LD zone depends on the nature of identity cleavages.
i.
ii.
when the elite share an identity with the minority they would prefer LD
to ED for a larger share of the parameter space
when the identity cleavage between majority and minority gets smaller,
the majority’s preference for ED over LD becomes weaker.
The West versus the rest
• In the West, liberalism comes before franchise
• conflation of property rights with minority rights
• extending the franchise is a compromise for liberals
• ED arrives when social mobilization is based on class (rather than
identity)
• In developing countries, bleaker prospects for LD
• social mobilization in the context of identity politics (decolonization,
wars of secession or national liberation)
• ethnic/identity cleavages comparatively strong
• delayed industrialization (and premature de-industrialization), so
class-based cleavage comparatively weak
Final words
• Puzzle is not rarity of liberal democracy, but that it exists
at all
• Endogeneity of cleavages
• determined both by structural and ideational factors
• Fundamental political difference between income/class
versus identity cleavages
• both cleavages can serve as basis for majoritarian populism, but
populism of the “left” aims to ultimately overcome income cleavage,
while populism of the “right” necessarily deepens identity cleavage
Additional slides
But what kind of democracy?
• widespread rights violations
• discrimination against minorities and opposition groups in many
OECD countries: Hungary, Croatia, Israel, Mexico, Turkey
• much worse in countries like Russia and Venezuela
• even though elections remain in principle free and
competitive
• preponderance of intermediate regimes
• Fareed Zakaria: “illiberal democracy”
• Steve Levitsky and Lucan Way (2010): “competitive
authoritarianism”
Our treatment of “civil rights” (1)
• We model civil rights as the non-discriminatory provision of public
goods.
• We interpret the relevant public goods broadly, including justice and
free-speech rights as well as education, health, and infrastructure.
• what sets LD apart from ED is that an elected government cannot discriminate
against specific individuals or groups when it administers justice, protects basic
rights such as freedom of assembly and free speech, provides for collective
security, or distributes economic and social benefits.
• Our treatment has the advantage that it provides a tractable approach
for modeling LD and distinguishing it from other political regimes.
• thinking of liberalism broadly as non-discrimination allows us to sidestep
debates about what are the “essential” characteristics of liberalism.
• our formulation is flexible enough to encapsulate individual and minority rights
• it applies to a variety of different contexts – non-discrimination in the
administration of justice, access to education, use of public infrastructure, or
right to free speech
• furthermore, our emphasis on public goods means that we focus on an
outcome that is sufficiently general that it can be applied in different
country, cultural and historical contexts.
Our treatment of “civil rights” (2)
• Our distinction between electoral and liberal democracies relies on
the presumption that free and fair elections – the hallmark of electoral
democracy – can be separated from equal treatment and nondiscrimination – the hallmarks of liberalism.
• It may be difficult at times to disentangle certain civil rights from political rights.
In particular, it can be argued that elections cannot be entirely fair when the
capacity of citizens to participate and compete in elections is constrained –
indirectly – by restrictions on their civil rights. Citizens who are deprived of, say,
adequate educational opportunities or the protections of the rule of law cannot
be effective participants in electoral contests.
• But this is a caution about the fuzziness in practice between electoral and
liberal democracies, rather than an objection that renders our distinction
between the two regimes entirely invalid.
• To require equality of access across the full range of public goods as a
precondition for free and fair elections would also set too high a threshold.
A taxonomy of political regimes
property rights
no
yes
political rights
political rights
no
yes
no
(1)
personal
dictatorship or
anarchy
(2)
dictatorship of
the proletariat
yes
(3)
n.a.
(4)
democratic
communism
civil rights
no
Yes
What about Liberal Autocracy (LA)?
political rights
no
civil
rights
yes
no
yes
(5)
right-wing autocracy (RA)
(6)
electoral democracy (ED)
𝜎 𝑅𝐴 =
1
𝛾(1 − 𝛼)
𝜏 𝐸𝐷 =
𝛼
𝛾
𝜃 = 𝜃𝑒
𝜃 = 𝜃𝑎 = 1
(7)
liberal autocracy (LA)
(8)
liberal democracy (LD)
𝑡 𝐿𝐴
1−𝛼
=
𝛾
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
𝜏 𝐿𝐷
𝛼
=
𝛾(2 − 𝜃)
𝜃 = 𝜃, with 𝜃𝑏 < 𝜃 < 1
minority could prefer LA to ED due to nature of public good
𝛼𝑏𝑖𝑛𝑡
𝛼𝑏𝑚𝑎𝑥
𝛼𝑏𝑚𝑖𝑛
𝛼𝑒∗