Modernization: Theories and Facts

Author(s): Adam Przeworski and Fernando Limongi

Source: World Politics, Vol. 49, No. 2 (Jan., 1997), pp. 155-183
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/25053996
Accessed: 30/07/2009 01:38
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=jhup.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to
World Politics.

http://www.jstor.org

MODERNIZATION
Theories

and Facts

By ADAM PRZEWORSKI andFERNANDO LIMONGI*

Introduction
T

JL

THAT makes political regimes rise, endure, and fall?Do democ

V V

racies

as a consequence
of economic
development?
destabilize
democracies?
Is there some
growth

emerge

rapid economic

Does
level of

development beyond which democracies aremore likely to fall? Is Eu

ropean
veloped
Our

history unique
countries?

or is it
repeating

itself

in contemporary

less de

two theories
is to distinguish
that relate economic
de
purpose
some facts in
to examine
and
and
these
of
velopment
democracy
light
concern
theories. While
the interesting
the mech
ultimately
questions
anisms that mediate
between
economic
and
the
dynamics
development
we must nevertheless
of political
the facts to be ex
regimes,
identify
we
before
into
stick as close as
Hence,
plained
plunging
explanations.
nar
to
pose the question
patterns. We
possible
elementary
descriptive
the impact of development,
rather than
rowly, examining
exclusively
seeking

to

broadly

deliberately
the world

explain
ignore factors
income
system,

to influence

by others
that our question
answers,
vergent
well understood.
found

In Section

we
the dynamic
of political
regimes. Hence,
such as religion,
in
colonial
legacy, position
or diffusion,
which
have been
distribution,

the incidence
of democracy
We
believe
to di
in its own
is important
it
that
lends
itself
right,
are
not
and that it raises methodological
issues that

I, we

reconstruct

two alternative

views

of the relation

be

tween development and democracy, both put forth by Lipset,1 and we

count

the cases

that fit them.

In Section

II we

examine

the vulnerabil

*
We
Cheibub, Fernando Cort?s, Larry Dia
appreciate comments by Mike Alvarez, Jos? Antonio
mond, John H. Kautsky, Seymour Martin
Lipset, Alejandro
Lopez, Jos? Maria Maravall, Guillermo
and Susan Stokes. This work was supported in part by a grant from the National
Science
O'Donnell,
no. SES-9022605.
Foundation
1
of Democracy:
Economic Development
and Po
Seymour Martin
Lipset, "Some Social Requisites
litical Legitimacy,'' American Political Science Review 53 (March 1959); and idem, Political Man:
Social Bases of Politics (Baltimore: Johns
Press, 1981).
Hopkins University

WorldPolitics 49 (January1997), 155-83

The

WORLD POLITICS

156

to economic
crises. In Section
the
III we consider
ity of democracies
most
in
criticisms
and
Section
substantive
of
's
views,
important
Lipset

IV we study methodological
reflections

close

criticisms. Methodological

the paper.
Appendix

1 explains

and political

our classification

of

regimes, while Appendix 2 spells out the analytics of regime dynamics.

I. Economic

Development

and Democracy

observation
that democracy
is related to economic
develop
Lipset's
in 1959, has generated
the largest body of research
ment, first advanced
on any
and con
It has been supported
topic in comparative
politics.
while
several
revised
and
buried
and
And
resuscitated.
tested,
extended,
to
in the recent Festschrift
neither
conclusions,
Lipset proclaim
the theory nor the facts are clear.2
Even a glance at the aggregate
such as Figure
1, shows that
patterns,
the relation between
levels of development
and the incidence of demo

articles

a
of regimes
is strong.3 Indeed,
condi
regimes
probit analysis
as
on the per
we
to
refer
which
throughout
only
capita income,
77
of
annual
the level of development,
4,126
percent
correctly classifies

cratic

tional

observations.4

The probability
is
chance
greater than 0.99.
by
Yet there are two distinct
democracies
nomically,

may
or

that

this classification

is not generated

reasons

this relation may hold: either

eco
as countries
to emerge
develop
likely
de
be established
of economic
independently

be more

they may

2
in Gary Marks
and
"Economic Development
and Democracy
Reconsidered,"
Larry Diamond,
eds., Reexamining
Larry Diamond,
Lipset (Newbury
Democracy: Essays inHonor of Seymour Martin
Park, Calif: Sage Publications,
1992).
3
lead to somewhat divergent results, the
While
different data sets and different estimation methods
most

finds that
careful statistical study of the aggregate patterns thus far, by Burkhart and Lewis-Beck,
causes democracy. Ross E. Burkhart and Michael
economic
S. Lewis-Beck,
Granger
development
Thesis," American Political Science Review 88
"Comparative Democracy: The Economic Development
(December
1994), 903-10.
4
and democ
the relation between development
A fair amount of ink has been spilled over whether
to Democratic
of Economic Development
racy is linear. See Robert W. Jackman, "On the Relation
and Zehra F. Arat,
Science 17 (August 1973), 611-21;
American Journal of Political
Performance,"
Modernization
Theory Revisited," Comparative Politics 21
"Democracy and Economic Development:
is a qualitative or a lim
however measured,
(October 1988), 21-36. We now know better. Democracy,
it ranges from 2 to 14 on the Freedom
ited variable: it assumes values of 0 or 1 under our measurement;
House Scale created by R. D. Gastil, Freedom in theWorld: Political Rights and Civil Liberties, 1987-88
(New York: Freedom House,
1988); from 0 to 100 on the scale of Kenneth A. Bollen, "Issues in the
American Sociological Review 45 (June 1980),
of Political Democracy,"
Measurement
Comparative
can become negative as the level of de
no
and so on. Hence,
index of democracy
370-90,
predicted
can exceed whatever
is the maximum
index of democracy
tends to zero, and no predicted
velopment
value of a particular scale as the level gets very large. Only a nonlinear function, such as the normal or
as
can satisfy these constraints. See Dahl, Polyarchy (New Haven:
logistic,
suggested by Robert A. Dahl
iswhy we use probit or logit models
Yale University
Press, 1971). This
throughout.

MODERNIZATION: THEORIES & FACTS

157

1.00000
0.94737
0.88421
0.82105
0.75789
0.69474
0.63158
0.56842
0.50526
0.44211

*
*

0.37895
0.31579
0.25263
0.18947
0.12632
0.06315

2,0000

4,000
Level:

Probability

that

a Regime

a1985 PPP USD-purchasing-power

parities

6,000 8,000

GNP/cap

in 1985

1
Figure
is Democratic,
inU.S.

PPP USD*

by Per Capita

Income

dollars.

to survive in
but may be more
countries.
velopment
likely
developed
We
call the first explanation
and
the
second
"endogenous"
"exogenous."
two
Since we are dealing with
democracies
emerge
only
regimes,
to assert that democracies
whenever
die.5 Hence,
emerge
dictatorships
as a result of economic
is the same as to say that dictator
development
as countries
ruled by them become
ships die
economically
developed.
out of
is
then
secreted
Democracy
by economic
develop
dictatorships
ment. A story told about country after country
is that as they develop,
to
social structure becomes
labor processes
require the
begin
complex,
of employees,
and new groups emerge and organize.
cooperation
a result, the
can
no
run
system
longer be effectively
by command:
too
the society is
the direct pro
complex,
technological
change endows
some autonomy
ducers with
and private
civil society
information,
and
control
dictatorial
of
forms
lose
their
effectiveness.
Vari
emerges,
ous groups, whether
or
the bourgeoisie,
workers,
just the amorphous
"civil society," rise against the dictatorial
and
it falls.
regime,
a "modernization"
The
is
endogenous
theory. The basic
explanation
of
in
this
its
is
of
that
there is one gen
versions,
any
assumption
theory,
active

As

5
is not quite true of our data set, since different countries enter and exit the sample at differ
This
ent moments.
For now, we consider the population of countries as fixed, but see Section IV.

WORLD POLITICS

158

eral process of which democratization

is but the final stage.Modern

ization consists of a gradual differentiation

structures
in a separation
that culminates
other

structures
consist

chains

and makes

a
society
ready
Modernization
lated

impute

paraphrases

urbanization,

educa

among
incorporation,
that
of social changes
accumulation
progressive
to
to its culmination,
democratization.
proceed
is re
of democracy
may be one reason the incidence

to economic

tators

democracy
of industrialization,
and political
mobilization,

of sequences

tion, communication,
others:
innumerable

of social
and specialization
structures
from
of political
causal
The
specific
possible.

to

commen
and this is the reading most
most
influential
critic, O'Donnell,
as
as
countries
become
other
that
"if
saying

development,
His
Lipset.6
s thesis

Lipset

rich as the economically advanced nations, it is highly probable that

democracies."7
then, is endoge
Democracy,
they will become
political
Ac
since it results from development
under authoritarianism.
nous,
one
one
to this
events
is
the
of
would
sequence
expect
theory,
cording
democratic
countries
and
of poor authoritarian
becoming
developing
a "threshold."
some level of
once
they reach
development,
are
to die and democra
that dictatorships
equally likely
so many
cies to emerge at any level of development.
They may die for
conse
reasons
with
all its modernizing
that development,
different
as
no
Therborn
role.
After
all,
quences,
emphasized,
plays
privileged
not
because of
of
countries
because
democratized
wars,
many European
a
in
the Malv
defeat
the
"modernization,"
story repeated by
Argentine
of the
fell in the aftermath
inas and elsewhere.8
Some dictatorships
Yet

suppose

for instance?who
had been
dictator?a
of a founding
Franco,
order.
Some
dictatorial
of
the
collapsed
maintaining
capable
uniquely
crises. Some because of foreign pressures.
because of economic
to
die and democracies
If dictatorships
emerge randomly with
regard
more
be
democracies
is it still possible
that there would
development,
one
is to take
ones?
If
countries
than
among
poor
among wealthy
a nation,
more well-to-do
at his own word?"The
the greater
Lipset
even if the emergence
of
the chances
itwill sustain democracy"9?then
death

democracy is independent of the level of development, the chance that

6
and John D. Stephens,
Diamond
(fn. 2), 45; as well as Evelyne Huber, Dietrich
Rueschemeyer,
on
"The Impact of Economic Development
Democracy," Journal ofEconomic Perspectives 7 (Summer
1993), 71-86.
7
Studies in South American
and Bureaucratic Authoritarianism:
Guillermo
O'Donnell, Modernization
of California,
Politics (Berkeley: Institute of International
1973), 3.
Studies, University
8
Goran Therborn,
"The Rule of Capital and the Rise of Democracy," New Left Review, no. 103
(May-June
9
Lipset

1977).
(fh. 1,1959),

56.

MODERNIZATION: THEORIES & FACTS

159

such a regime will survive is greater if it has been established in an af

fluent

country. We

thus expect

would

to observe

democracies

to appear

randomly with regard to levels of development, but to die in the poorer

ac
ones. Thus,
history gradually
a
since every time
cumulates
democracies,
dictatorship
happens
wealthy
to die in an affluent country,
is there to stay (see
democracy
Appendix
emer
is therefore no longer a modernization
2). This
theory, since the
not
is
about by development.
of democracy
Rather,
gence
brought
as a deus ex machina.
survives
if a
It
appears exogenously
democracy
a
not
it
is
is
of
"modernization."
but
"modern,"
country
product
countries

in the wealthier

and survive

Are we splitting hairs?

Examine

first some descriptive

patterns.

The

facts we

report

concern

135 countries between roughly 1950 and 1990. "Entry" year refers to
or to the first
country became
independent,
are
and "exit" year refers to
data
available,
year
the data are available. All the regimes that
1990 or to the last year when
or dictator
as democracies
occurred during
this period were classified
use
term
with
"authoritarian
latter
the
(we
interchangeably
ships
we observed
101 democratic
224 regimes,
and
regimes").10 Altogether,
1950,

or to the year when

economic
for which

123 authoritarian.
rates are
power

The

expressed
and expressed

parities

to levels of
and growth
development
at
dollars computed
purchasing
in 1985 prices.
refer
(Thus all $ numbers

references

in constant

U.S.

to 1985 PPPUSD.) The lowest level we observed in the entire sample

is $226 (Burma in 1950), the highest is $18,095 (United States in
1989).11
as a result of economic
devel
If the theory that democracy
emerges
more
to
be
is
transitions
would
true,
opment
democracy
likely when
In
levels
of
authoritarian
reach
fact, tran
development.
regimes
higher
as per
rises
income
of
sitions are increasingly
likely
capita
dictatorships
but only until it reaches a level of about $6,000. Above
that, dictator
more
as countries become more affluent. Dictator
become
stable
ships
10
1 and in
in Appendix
and the resulting list of regimes are described
Our regime classification
Alvarex et al., "Classifying Political Regimes," Studies in International Comparative Development
reason for
selecting this period and the sample is the availability of internationally
(forthcoming). The
we describe
we took from the Penn World Tables 5.6. The
sample
comparable economic data, which
here and use throughout does not include six countries that derive at least half of their income from oil
are available for 4,730 country years, data for economic growth are avail
revenues. While
political data
in most analyses.
is the number of observations
able for only 4,126 country years, which
11
Readers used to the UN or theWorld Bank GNP figures should be aware that counting incomes at
to increase significantly
the levels for poor countries and to decrease
purchasing-power
parities tends
to know what different
slighdy the numbers for rich countries. It may be useful for future reference
numbers describe: by 1990, Nigeria had a per capita income of $995, Indonesia had $1,973, Czecho
slovakia $4,094, Spain $9,576, and the United States $18,073.

Mike

WORLD POLITICS

160
survive,

ships

or at least succeed

one another,12

in the
invariably
are somewhat
less stable
almost

those under $1,000. They

very poor countries,
in countries with
incomes between
and $4,000
$1,001

and even

less so

above $4,000. But if they reach the level of $6,000, transitions to

democracy become less likely.As the lower panel of Table 1 (PADcol
umn 4) shows, the probability of any dictatorship dying during any year
is 0.0206; for those dictatorships with incomes over $1,000, this prob
over
ability is 0.0294, over $5,000 it is 0.0641, over $6,000 it is 0.0484,
$7,000

it is 0.0333.

it seems, was

Huntington,

correct with

regard

to

dictatorships: they exhibit a "bell shaped pattern of instability"13

To
we

these patterns can be predicted

by per capita income,
on level and,
the transition
conditional
given
probabilities

test whether

estimate

the nonlinearity
The results are

of the observed

its square (see Appendix

2).
patterns,
of dic
2. As we see, the probabilities

inTable

presented

tatorships falling, p^, predicted by the level of development correspond

to those observed.
closely
and then decline.
Indeed,

dictatorships

wealthy. Whatever

They

increase

survived

until

for years

the $5,001-$6,000
in countries

range
that were

the threshold at which development is supposed to

it is clear that many dictator

regimes,
that
those countries
Even
disregarding
ships passed
good
revenues
more
from
than one-half
derive
of their
oil, dictatorships
in
East
flourished
USSR,
Taiwan,
Germany,
Singapore,
Spain, Bulgaria,
in
and Mexico
for many years after these countries
enjoyed
Argentina,
comes above $5,000, which Austria,
Ice
France, Germany,
Belgium,
dig

the grave

for authoritarian

it in

health.

land, Italy,Netherlands, and Norway did not have by 1950. Table 3 lists
the dictatorships that survived even though the probability that the

was above
regime is democratic
predicted
by the level of development
to
income
of
which
$4,115.
0.50,
per capita
corresponds
Yet this may not be a fair test of modernization
theory. The hypoth
over a
esis implied by this theory is that ifa country develops
longer pe
so that all the
have
riod under dictatorship,
consequences
modernizing
most
dic
But for
time to accumulate,
then itwill embrace democracy.
this premise
is vacuous:
only 19 dictatorships?to
tatorships
over
out of 123?did
of time and
longer periods
develop
us
more
these
examine
thus
countries,
closely
"modernity." Let
that developed

under

authoritarianism

and became

"modern,"

remind,
reached
the ones
which

12
If President Viola succeeds President
successive dictatorships.
that we do not distinguish
Note
Videla or even if ayatollahs succeed a shah, we treat it as one continuous
spell of dictatorship.
13
Political Order in Changing Societies (New Haven: Yale University
Samuel P. Huntington,
Press,
1968), 43.

MODERNIZATION: THEORIES & FACTS

Regime

Table
by Lagged

Transitions

Per Capita

(annual

Income

(Level)"

data)
TRD

TA

PDA

TRA

TD

0.0066
0.0248
0.0276
0.0161
0.0492
0.0641
0.0625
0.0333

6
18
8
3
6
5
2
1

915
727
290
186

0.1250
0.0571
0.0380
0.0333
0.0187
0.0088
0.0083
0.0000

9
14
7
5
2
1
1
0

72
245
184
150
107
113
121
619

3991

0.0206

49

2380

0.0242

39

1611

3004
2032
1558
1222
993
802
649

0.0294
0.0339
0.0379
0.0534
0.0571
0.0484
0.0333

43
25
17
14
8
3
1

1465
738
448
262
140
62
30

0.0195
0.0124
0.0081
0.0042
0.0023
0.0014
0.0000

30
16
9

1539
1294
1110
960
853
740
619

PJK

TTR

TOT

PAD

-1000
1001-2000
2001-3000
3001-4000
4001-5000
5001-6000
6001-7000
7001

0.0152
0.0329
0.0316
0.0238
0.0349
0.0314
0.0196
0.0015

15
32
15

987
972
474
336
229
191
153
649

AU

0.0221

88

0.0243
0.0202
0.0167
0.0147
0.0101
0.0050
0.0015

73
41
26
18
10
4
1

Low-High

161

122
78
32
30

Above

1000
2000
3000
4000
5000
6000
7000

is lagged, we

"Since per capita income

are used:
abbreviations

lose 135 observations,

that either regime dies during

PJK is the probability
TTR is the number of transitions
TOT is the total number

of regime years at a particular

of transition to democracy

PAD is the
probability
TRD is their number
TA is the total number
PDA is the
probability
TRA is their number
TD is the total number

we will

for the total of 3,991. The

particular

following

year

level

of years under authoritarianism

of transition to authoritarianism
of years under democracy

take arbitrarily

to mean

that at some

time

a per
they had
capita

income of $4,115. (SeeTable 4.)

Gabon,
Syria,
enced a sustained
teen,
more

and eighteen

are the three countries

and Yugoslavia
that experi
increase in income over, respectively,
twelve, seven
was the
at
which
years, reached the level
democracy

remained
under
likely regime, and, having
experi
dictatorships,
a series of economic
are the two
crises.
and Malaysia
Singapore
over a
countries
that
became wealthy,
and re
long period,
developed
now.
mained
In East
until
USSR, Spain,
Taiwan,
dictatorships
Germany,
and
but
fell,
Bulgaria,
dictatorships
eventually
only many
Hungary
enced

WORLD POLITICS

162

Table
Regime

Transition
_Per

Level

Probabilities

Capita

Income

by Lagged

Predicted
and

Its

Square3_
PM N
PD*

PDA
0.098

PAD
0.010

PDD
0.902

0.990

0.092

987

(0.125)
0.072

(0.007)
0.017

(0.875)
0.928

(0.993)
0.983

(0.053)
0.191

972

(0.057)

(0.025)

(0.943)

(0.975)

(0.304)

2001-3000

0.044

0.028

0.956

0.972

0.388

474

3001-4000

(0.038)
0.026

(0.028)
0.039

(0.962)
0.974

(0.972)
0.961

(0.424)
0.600

336

4001-5000

(0.033)
0.015

(0.016)
0.047

(0.967)
0.985

(0.984)
0.953

(0.326)
0.758

229

5001-6000

(0.019)
0.008

(0.049)
0.050

(0.981)
0.992

(0.951)
0.950

(0.720)
0.862

191

6001-7000

(0.009)
0.004

(0.064)
0.047

(0.991)
0.996

(0.936)
0.953

(0.876)
0.921

153

7001-

(0.008)
0.0006

(0.063)
0.017

(0.992)
0.999

(0.937)
0.983

(0.887)
0.965

649

(0.000)

(0.033)

(1.000)

(0.967)

(1.000)

<1000
1001-2000

All

0.051

0.028

0.959

0.977

0.354

(0.024)

(0.021)

(0.976)

(0.979)

(0.466)

3991

*
Based on a dynamic probit model. See Appendix
2. The probabilities pv j=A,D, k=A,D
sitions and survival.
is the equilibrium proportion of democracies. Observed
transition
p*D
Table 1) are in parentheses.

are of tran
rates (from

years after they had reached the critical level of income. Given its 1974
income

level, Uruguay

should

never

have been

dictatorship.

The

eco

nomic history of the Chilean dictatorship is convoluted: its income in

1974 was $3,561, it climbed with downs and ups to $4,130 by 1981,
collapsed to $3,199 by 1983, recovered to surpass the 1974 level only
by 1986, and passed the threshold of $4,155 in 1989, exactly the year of
transition.

The

the threshold

is similar: by our criteria,

it reached
history of Poland
an economic
in 1974; it experienced
of democracy
crisis

in 1979 and a mass movement for democracy in 1980, passed the

threshold again in 1985, and became a democracy in 1989. In turn,
Brazil,
Greece

Czechoslovakia,
are the dream

and perhaps
Portugal,
cases of a modernization

that developed
under a dictatorship,
off more or less at the same
threw dictatorships
are few.
countries

This

is not

to say that democracies

did not

even

South

Korea

and

theorist.

These

are

wealthy,
levels. But

and

became
income
sometimes

emerge

they
be

MODERNIZATION:

Highest

163

3
Table
of Per Capita
Income
(Level) under Which
in different
dictatorships
survived
countries
Highest Level
11698
10433
8598
8067
7744
7390
6969
6939
6866
6505
6463
6434
5815
5674

Year

1990
1988
1979
1990
1989
1976
1976
1957
1988
1980
1981
1976
1972
1979
1987
1973
1981
1990
1978
1987
1981
1974
1962
1957
1981

Singapore
East Germany
Iraq
Taiwan
USSR

Spain
Gabon
Venezuela
Bulgaria
Argentina
Mexico
Iran
Argentina
Yugoslavia
Hungary
Greece
Uruguay
Malaysia

Poland
Korea

Syria
Portugal
Argentina
Argentina
Suriname
*
The

& FACTS

Levels

Country

South

THEORIES

prob(reg=dem)?

0.992
0.977
0.923

0.895
0.875
0.851
0.818
0.815
0.809
0.776
0.772
0.769
0.705
0.690
0.687
0.637
0.630
0.625
0.623
0.620
0.569
0.568
0.553
0.530
0.513

5650
5218
5162
5117
5102
5080
4668
4657
4541

4355
4220

is the probability that a regime is democratic given the level. It is calculated as

PROB(reg=DEM)
where the parameters are estimated by the probit model and F(.) is the cdf of the nor

l-F(a+?*LEVEL),
mal distribution.

cause

do not
modern;
put otherwise,
dictatorships
fall for the same reasons in all countries. Thus moderniza

countries

necessarily
tion may

became

"explain" why

democracy

was

established

in countries

that de

even it these countries had waited for its

veloped over a long period
But ifmoderniza
of time that cannot be predicted.
periods
must
be some level of
there
is to have any predictive
power,
theory
sure
can
one
at
the
that
be relatively
income
which
country will throw

advent

for

tion

off the dictatorship. One is hard put to find this level, however: among
the countries

that

satisfy

the premise

of the modernization

theory,

the

range of levels atwhich dictatorships survived is very wide (see the list
inTable 4).

WORLD POLITICS

164

Countries

That

4
Table
over Long

Developed
and Reached

Incomes

Entry
Country
Gabon
Brazil

Chile
Uruguay
South Korea
Malaysia
Singapore

Syria
Taiwan
Bulgaria
Czechoslovakia
East

Germany

Greece
Hungary

Poland
Portugal

Spain
USSR
Yugoslavia

under
Periods
above
$4,115a
Peak

Passes

Year

Level

1961
1965
1974

1969
1864
3561

1974
1961
1957
1965
1961
1952
1981
1964
1971
1967
1971
1971

4148
911
1282
1845
1607
968
4216
1654
4995
3308
3657
3109

1951
1951
1961
1961

1314
2205
2536
2073

Dictatorship

PROB=0.50

1973
1980
1981
1989
1974
1985
1982
1972
1978
1979
?
1989
?
1970
1974
1974
1985
1973
1964
1971
1974

Transition

Year

PROB

1976
1980
1981
1989
1981
1988
1990
1990
1981
1990
1989
1989
1988
1974
1987
1978
1988
1974
1976
1989
1979

0.82

never

0.52

0.47

0.53

1978
no
1989

0.63

1985

0.48
0.68

0.50

at PROB

Year

0.68

1988

0.63

never

0.99

never

0.53

0.57

never

0.90

post 1990
1989
1989
1990
1974
1989
no
1989

0.80
0.51
0.98
0.61
0.69
0.62
0.55

0.57
0.85
0.88
0.69

1975

0.80
0.51
}
0.61
0.68

0.55

0.52
0.85

1976
collapsed

collapsed

that grew over the period of at least seven years and at some time reached
income
of
$4,115.
per capita
Entry is 1951 or the year after the country became independent or the
year after economic data became available. Passes PROB=0.50 is the year when the country reached per
capita income of $4,115. Peak gives the time when the country reached the highest income level under
aThis table lists countries

and the probability, as predicted

dictatorship
transition
gives the year the dictatorship
democracy. Finally,
racy at that time.
the particular

Moreover,

even

if to

predict

is not

by per capita income, that itwould be a

fell, if ever, and the probability of democ

the same

as to

explain,

"explain

ing" can easily entail an ex post fallacy.Consider Taiwan, which in 1961

had a per capita income of $968, which developed rapidly, passing by
1979 our threshold of $4,115, which on the basis of its income level
had a probability of 0.10 of being a dictatorship in 1990, andwhich in
1995

elected

its president

in contested

elections

for the first

time.

Sup

pose that every year during all this time, the Taiwanese dictatorship
faced a probability of 0.02 of dying for reasons not related to develop
ment.

It thus had

about

a 50 percent

chance

of not

being

around

by

1995 even if it had not developed at all.We may therefore attribute to

development

what

may

have

been

just

a culmination

of random

haz

MODERNIZATION: THEORIES & FACTS

165

ards.And, indeed, theTaiwanese dictatorship most likely democratized

ones.14
reasons, not for economic
geopolitical
of
economic
the
causal
Thus,
power
development
down
appears paltry. Few authoritarian
torships

for

of modernization

premise

theory;

in bringing

dicta

regimes
satisfy the
over a
that is, few developed
long pe

riod. And even ifmost of those that did develop eventually became

no level of income
occur.
that would
democracies,
predicts when
of the level of development,
In turn, per capita income, our measure
is
The
has a strong impact on the survival of democracies.
simple fact
that during the period under our scrutiny or ever before, no democracy
ever fell,
else, in a country with a per capita in
regardless of everything

come higher than that of Argentina

democracies
collapsed,
countries

in 1975: $6,055.15 Thirty-two

and not one

incomes
above $6,055
years with
spent
out
of
democracies
while
sixty-nine
thirty-nine
that were poorer.
736

did

fall

in

survives increases
1 shows, the probability
that democracy
In countries with per capita
with per capita income.16

As Table
monotonically

income under $1,000, the probability that a democracy would die dur

ing
was

life
year was 0.125, which
implies that their expected
was
this
Between
and
$2,000,
$1,001
probability
years.17

particular

eight
of about
duration
for an expected
0.0571,
years. Above
eighteen
could expect to last forever. Statistical
$6,055, democracies
analysis, the

results of which

are shown inTable 2 (column 1), confirms that per

is a good predictor
of the stability of democracies.
capita
an
out
himself
for
These
cry
thought
Lipset
explanation.
findings
is that wealth
survive in affluent countries
that the reason democracies
income

moderates
is a plausible

in various ways
explanation

the intensity
but not easy

of distributional
to prove

conflicts.

This

The

intu

rigorously.18

itive story is this: Suppose that the political forces competing over the

14
An analogy may be useful. Suppose that someone runs the risk of 0.01 of dying from accidental
at the age of seventy-eight
causes
she gets hit by a falling brick. To
during each year of her life and that
is to conclude that she died of old age.
attribute this death to development
15
The claim about the prewar period is based on rather heroic backward extrapolation of 1950 in
lower: we guess
fell in Europe were an order of magnitude
comes, but the levels at which democracies
in 1933, and
in 1930, $1,474 inGermany
it to have been $1,825 inAustria in 1934, $1,974 in Finland
$1,814 in Italy in 1922.
16
and Keith T. Poole, "Poverty, the Coup Trap, and the Seizure of Executive
John B. Londregan
and Poole found a similar pattern with regard to
Power," World Politics 42 (January 1990). Londregan
times more
1950 and 1982 coups were twenty-one
coups. In their sample of 121 countries between
to
occur
wealthiest
countries.
than
the
the
among
among
poorest
likely
17
life in any state is the inverse of the probability of transition away from this state.
18Expected
at the
Survive in Affluent Countries?"
Adam Przeworski,
(Paper presented
"Why Democracies
annual meeting

1996).

of the American

Political

Science Association,

San Francisco, August

28-September

1,

WORLD POLITICS

166
of income

distribution
democratic
share

of

choose

in which

competition,
total income,

or

the verdicts of
with
complying
some
case each can
to
expect
get
a
over
is
which
fight
dictatorship,

between

risking

costly but which gives the victor all of the income. Now

suppose that

the marginal
is lower at higher
levels of con
utility of consumption
is
the
Thus
the
from
for
sumption.
gain
winning
struggle
dictatorship

smaller. In turn, if the production function has diminishing marginal

a part of it dur
in capital stock, the "catch-up" from destroying
war for
at lower levels of wealth. Hence,
the
in
is
faster
ing
dictatorship
a
accu
countries
is
the
value
of
dictator
and
the
poor
greater
becoming
returns

cost of
countries,
destroying
capital stock is lower. In wealthy
a
income
the
from
rather
than
of
total
all
contrast,
part
by
gain
getting
is smaller
is slower. Hence,
and the recuperation
from destruction
is more attractive
in poorer countries.
struggle for dictatorship

mulated

for ex
there are always alternative
One,
interpretations.
Obviously,
a proxy for education
more educated
is
is
income
that
and
just
ample,
are more
to embrace democratic
the ac
values. But while
likely
people
an
cumulated
of
of the labor
years of education
average member
measure
stocks we have?does
increase the
force?the
of educational
probability
of income
stronger.
These

of survival

of democracies

survives when

education

of level, the effect

independently
is controlled,
and indeed it ismuch

the
confirm
strongly
is
Once
established,
theory.
democracy
the more
likely that itwill survive.
reason we observe
the relation between
observations

Lipset
nation,
The

and the incidence

to survive

once

democracy

of
a

well-to-do

levels of development
are almost certain
is that democracies
of democracy
are established
in
countries.
rich
True, dictator
they

are less stable when

ships
But what

version

exogenous
the more

they

reach

the pattern we
generates
is terribly fragile in poor

the per capita income of $4,000.

1 is that while
observe
in Figure
countries,

it is impregnable

in the

rich ones. The probability that a democracy will die during any partic
an income above
ular year in a country with
two in a thousand
since at such
years. And

zero:
is practically
$4,000
die at
levels dictatorships
that independently
of the

one would
percent,
expect
in the long run democracies
initial distribution,
in such wealthy
countries.19
percent of regimes
the rate of 5.7

torships

died

at a double,

triple,

or whatever

would
Even
times

constitute
if wealthy

higher

96.1
dicta

rate, that

is,

*
19
where p stands for tran
In the long run the proportion of democracies
equals p^/
PDA)>
(p^
2. The
sition probabilities, A for dictatorship
and D for democracy. See Appendix
("authoritarianism"),
numbers in the text are derived from Table 1.

MODERNIZATION: THEORIES & FACTS

even

to
much more
democracy
likely,
is 3.9 percent.
could make
to believe
that economic
develop

if development
made
transitions
all the difference
endogenous
theory
there are no grounds
To conclude,

ment

breeds

democracies:

Lipset's

167

"optimistic

equation,"

as O'Donnell

dubs it,20 the "benign line" in the language of Huntington

and Nel

it and those that do scatter

along
are
democracies
established,
likely
to survive in
ones.
wealthy

son,21 has few countries

running
In turn, once
random directions.
and certain
die in poor countries

in
to

II.Ups or Downs?
to
irony
Lipset
why democracies

cited several
theory. While
Lipset
and
factors
Olson22
survive, anticipating
explain
were more
to be
he
that
democracies
likely
thought
Huntington,23
s view, this threat to
countries grew rapidly. In
when
destabilized
Lipset
and com
democracy
originated with "extremist movements"?fascism
was a
as he saw it, extremism
of rapid de
munism?because,
product
occurred
"Wherever
industrialization
velopment.
rapidly, introducing

There

is yet another
to

between
the pre-industrial
sharp discontinuities
more
rather than less extremist working-class
Here

and company

Lipset

could

not have

and industrial
movements
been more

situation,

emerged."24
wrong.

Rapid

growth is not destabilizing for democracy (and neither is it for dicta

at
in incomes,
face a decline
they die
can
to
when
but
be expected
last nineteen
and
years,
an
comes are
at the rate of 0.0160, with
die
expected
they
growing,
democracies
that
of sixty-four
grow slowly, at
years.25 Moreover,
torship). When
rate of 0.0523

democracies

the
in
life
the

rate of less than 5 percent per annum, die at the rate of 0.0173, while
those that grow at a rate faster than 5 percent die at the rate of 0.0132.

(SeeTable 5.)
ismost

What

striking is how fragile

crises. In poor countries,

of economic

poor democracies
those with
per

are in the face

capita

income

under $2,000, of the 107 years during which a decline of incomes oc

curred,

twelve

democracy

democracies

under

such

fell

the following
is about

conditions

life of
year: the expected
nine years. Even
among

20O,Donnell(fn.7),4.
21
inDeveloping
and Joan M. Nelson, No Easy Choice: Political Participation
Samuel P. Huntington
Countries (Cambridge: Harvard University Press, 1976) 19.
22
as aDestabilizing
Mancur Olson Jr., "Rapid Growth
Voice? Journal ofEconomic History 23 (De
cember 1963).
23
(fn. 13).
24Huntington
54.
(fn. 1,1981),
25Lipset
and Poole (fn. 16) with regard to coups,
This finding parallels again the results of Londregan
which
they found to be less likely when the economy grows.

Observed

Rates

Table
5
of Transitions,
by Lagged
Rate of Economic
Lagged
All

Growth

Level

0-1000

1001-2000

2001-3000

3001-4000

4001-5000

5001-6000

6001-7000

7001

Total

PJK

TOT

PAD

15

987
420
567

0.66
1.01
0.39

972
299
673

Total

1.52
2.14

G> 0

1.06

3.29

32

G<=0

Total

4.68

14

G> 0

2.67

18
15

TRD

6
4
2

915
397
518

12.50
21.74

23

8.16

49

2.48
3.26
2.15

18
7
11

727
215
512

5.71

8
3
5

290
80
210

186

3.33

52

4.88

134

4.55

245
84
161

7
3
4

184
41
143

2.75

5
2
3

150
41
109

1.87

107
22
85

4.35

3.16

474
121

2.76

G> 0

2.55

353

2.38

Total

2.38

8
3
5

336
93
243

1.61
1.92
1.49

229
54
175

4.92
6.25
4.44

122
32
90

191
35
156

6.41
10.53
5.08

78
19
59

0.88

0.83

2.06

Total

3.49

G<=0

5.56

G> 0

2.86

Total

3.14

G<=0

5.71

G> 0

2.56

Total

1.96

G<=0

8.82

G> 0

0.00

Total

0.15

G<=0

0.00

G> 0

0.19

Total

2.21

G<=0

3.43

G>0

1.70

a
Since per capita income
are used:
abbreviations

88
40
48

is lagged, we

3.75

PDA is the
probability
TRA is their number
TD is the total number

7.32
2.80

1.18

113
16
97

0.00
1.03

32

27

0.00

121
29
92

649
110
539

3.33
0.00
3.70

1
0
1

30
3
27

0.00
0.00

0
0

619
107

0.00

512

3991
1166
2825

2.06
2.61
1.78

49
21
28

2380
803
1577

2.42

39
19
20

lose 135 observations,

a

of regime years at a particular

of transition to democracy

TOT is the total number

TA is the total number

3.80

153
6.25
34 40.00
119
0.00

that either regime dies during

PJK is the probability
TTR is the number of transitions
PAD is the
probability
TRD is their number

72

14
7
7

8.33

4.96

G> 0

TD

PDA

Total

3.23

TRA

TA

G<=0

G<=0

and

Democracies

Dictatorships

TTR

G<=0

Income

Per Capita
Growth3

of years under authoritarianism

of transition to authoritarianism
of years under democracy

5.23
1.60

for the total of 3991. The

particular
level

3.44

year

1611
363
1248

following

MODERNIZATION: THEORIES & FACTS

169

countries with incomes between $2,001 and $6,000, a decline of in

comes resulted in the fall of six democracies in 120 years during which
to last 20 years. And
could expect
this happened:
these democracies
a
occurs:
in
miracle
above
the
252
$6,055
then,
years during which
economic
democracies
crises, none ever fell.
wealthy
experienced
is that the political
effects
Another
striking feature of these patterns
crises are immediate:

occur one year later.We

tried to
they
over a
account
into
5
longer period and
taking
growth
reproduce
more
we did statistical
than one year.
(survival) analyses
lagging growth
one year of
Both
show that past
does not matter:
procedures
growth
to
economic
crisis is enough
effects.
the
produce
political
of economic

Table

Thus the hypothesis that rapid growth destabilizes regimes is simply

it is true that "economic
In turn, to cite Diamond
and Linz,
crisis
common
one of the most
to democratic
threats
represents
stability."26
are economic
What
destabilizes
crises, and democracies,
par
regimes
false.

ticularly

poor

are

democracies,

vulnerable

extremely

to bad

economic

performance.

III. Kinks: Modernization

While
ences

there are important

between

theoretical

Huntington

Theory
and even

and O'Donnell,

both

Revisited
sharper political differ
a
argued that there is

level beyond which further development decreases the probability that

that both regimes be
will survive. Huntington
contended
democracy
occurs
a country
come unstable when
which
modernization,
undergoes
at some
in
levels of development.
intermediate
turn,
O'Donnell,
a country exhausts "the easy
tend to die when
claimed
that democracies
stage of import

substitution,"
was concerned

Huntington
were
whether
they
distinction
political
form

of government

democratic

at some intermediate
level.
again
with stability of regimes and did not care
or authoritarian.
"The most
important

countries,"
among
but their degree

he

not

their

Hence,

the

told us, "concerns

of government."27

United States, theUnited Kingdom, and the Soviet Union were all sys
tems in which
it is the
"the government
governs." Whether
politburo,
he insisted,
little. "The problem,"
the cabinet, or the president matters
"was not to hold elections but to create organizations."
Indeed, we were

told, "The primary problem is not liberty but the creation of a legiti
26
in Latin Amer
and Juan J. Linz. "Introduction: Politics, Society, and Democracy
Larry Diamond
Countries: Latin Amer
ica," in L. Diamond,
J. J. Linz, and S.M. Lipset, eds., Democracy inDeveloping
ica (Boulder, Colo.: Lynne Rienner, 1989), 17.
27
(fn. 13), 1.
Huntington

WORLD POLITICS

170

never
to
While
public order."28
explicitly
referring
Lipset, Hunt
some of the tendencies
encom
"in
that
observed
ington
actuality, only
in the concept
of
the
modernization'
characterized
political
passed
areas. Instead
a trend toward
of
and
modernizing'
competitiveness
was an erosion of
a
to auto
there
and
democracy,
democracy'
tendency
cratic military
and
of
Instead
,
one-party
regimes
regimes.
stability
there were repeated coups and revolts."29

mate

raked Lipset through the coals

Anticipating Huntington, O'Donnell

for various
in retrospect,

methodological
he observed

nant?testimony

of a debate that in 1971 had recently begun and today

it is no

is finished:
series

on his criticisms
transgressions.
Reflecting
rem
that "Chapter I is now an archeological

of data

longer necessary
to demonstrate
that

not foster

to lead the reader

'socio-economic

and/or

through

tedious

development'
"30
What
the data

does

show,
'democracy
stability.'
political
is that in contemporary
South America,
the higher
asserted,
are associated with non-demo
and the lower levels of modernization
are found at inter
democracies
cratic political
while
systems,
political
at
least
within
the range
mediate
levels of modernization."
Hence,
we should observe
fall as
that democracies
observed
by O'Donnell,
O'Donnell

economies
Is there

develop.
some level of
development

more likely to die than before? Note

that the function

relating

beyond

which

democracies

are

(returning toTable 2, column 5)

the equilibrium

proportion

to

of democracies

per capita income has a kink at levels between $3,001 and $4,000: the
observed

values

are 42.4

percent

between

$2,001

and $3,000,32.6

per

cent between $3,001 and $4,000, and 72.0 percent between $4,001 and
$5,000. But this kink is due to the fact that dictatorships
ally stable in this range, rather than that democracies

are
exception
are less stable.

The probability of a democracy dying declines monotonically with per

did find a countercase
O'Donnell
capita income. While
his account of the rise of bureaucratic
authoritarianism

against Lipset,
is not a com

studied a country that turns out to be a dis

peting
theory.31 O'Donnell
a
tant outlier:
is the only country where
fell at an
democracy
Argentina
income above $6,000; Argentina
is also the only country where one col
at an income between
two democra
and $6,000. Only
$5,000
lapsed
28
29

Ibid., 7.
Ibid., 35-36.

30
Studies in South American Politics, 2d
and Bureaucratic Authoritarianism:
O'Donnell, Modernization
ed. (Berkeley: Institute of International
of California,
1979), 204.
Studies, University
31
was careful about not
was to
O'Donnell
making general claims: his purpose
explain the downfall
of democracies
in the Southern Cone. But his theory of "bureaucratic authoritarianism"
captured the
imagination of scholars around the world, who treated it as applicable almost everywhere.

MODERNIZATION: THEORIES & FACTS

171

cies fell in countries with incomes between $4,000 and $5,000: again
one of them inArgentina, and the other inUruguay. Five democracies
fell between $3,000 and $4,000: one of them inArgentina. Indeed,
only five democracies

outside Argentina,

fell in countries

with

incomes

above $3,000: in Uruguay in 1973 at $4,034, Suriname in 1980 at

$3,923, Chile in 1973 at $3,957, Fiji in 1987 at $3,398, and Greece in
1967 at $3,176. Thus, Lipset was right in thinking that the richer the
the more

country

likely

it is to sustain

IV.Does History

democracy,

Repeat

except

inArgentina.

Itself?

came
in 1950, the regimes we observed
or the en
of two effects: their dynamic
into being
or at least into our
trance of new countries
into the world,
sample.
inde
in our sample that were
countries
the seventy-three
Consider
of them had democratic
in 1950, when
regimes. By
thirty-five
pendent
to
increased
these countries
of democracies
1960 the number
among
was still
in
to fall to
1968.
It
thirty-one
by
only
thirty-one
thirty-nine,
Since

our observations

begin
as a result of either

1978, afterwhich

it climbed, back to thirty-nine in 1984 and to forty

our count
to the "old" countries,
with
regard
by 1990. Hence,
to which
oceanic
with
analysis, according
Huntingtons32
roughly agrees

eight

(1) the "secondwave" of democratization began in 1943 and ended in

1962, (2) the "second reversewave" started in 1958 and ended in 1975,
and (3) the "thirdwave" of democratization began in 1974. But the
story of the countries

that became

independent

after

1950

is entirely

different. Three out of twenty-five (12.0 percent) newly independent

countries
seven

were

democracies

out of forty-two

(14.3

in 1960;
percent)

the numbers were

subsequently,
in 1968, eleven out of fifty-five

(16.6 percent) in 1978, and twelve out of sixty-eight in 1990 (17.6 per

of democracies
the proportion
among these "new" coun
cent). Hence,
or up. In turn, the
no waves
tries grew slightly with
rolling down
in the world during
of democracies
of the aggregate
decline
proportion
new
rather than
countries
of
the 1960s is largely due to the emergence
to transformations
of old ones.

Since observations of any limited period of time combine dynamic

contro
and entry effects, the question whether
repeats itself is
history
tradition assume it does: they infer the his
in the
versial. Studies
Lipset
observations.
from cross-sectional
of "modernization"
torical process
32
in the Late Twentieth Century (Norman:
The Third Wave: Democratization
Samuel P. Huntington,
considered 74 countries while our sample cov
of Oklahoma
Press, 1991), 16. Huntington
University
ers 135 countries; hence, the data are not exacdy comparable.

WORLD POLITICS

172
Followers
claiming
not to be

of Moore33
that

contest

theWestern

repeated.
Cross-sectional

the validity
of such inferences,
however,
was
route to
democracy
unique,
European

observations

can be used

to infer historical

processes
on some ex
that regimes
survive or die conditional
are the same across
(in our case per capita income)
cross sections,
so that the
a country has a par
different
that
probability
on the realized values of these
ticular regime at any time depends
only
or the time when
the
rather than the period,
the region,
variables,
if the probabilities
variables
ogenous

country became independent. And we know that (1) the probability

that a democracy is born iswidely scattered with regard to the level of
development, rising at low levels and declining at high levels; (2) the
probability that a democracy dies declines monotonically with per capita
the probability
that a country has a demo
these
level. The question,
then, iswhether
were the same in different
or
conditional
regions.
probabilities
periods
about the validity of inferences
The
based on cross-sec
controversy
can be formulated
in a number
tional observations
of alternative ways:

income; and
cratic regime

(3) as a result,
increases with

(1)Were these probabilities different beforeWorld War II inWestern

Europe and elsewhere? (2)Were they different during the postwar pe
countries
and
that existed before
1950 ("old" countries)
among
later
that became
("new" countries)?
independent
we can
a full set of data for the prewar
Without
period,
only make
to
economic
with
the
first
data for
guesses
regard
question. Although
riod

those

with
after
the prewar period are not comparable
those at our disposal
we
to guess the
made heroic assumptions
levels at
1950,
approximate
which
democracies were established
and fell in some of the present OECD
are
inTable 6.
countries. The results of these calculations
presented
are
not
that
these
the
democracies
that existed
however,
Note,
only
as
as
War
II. Southern
well
before World
several Latin
European
American
while

countries

in Eastern

experienced
relatively
long spells of democracy,
most of the democratic
regimes that emerged
Europe

in the aftermath ofWorld War I collapsed after the first election.34And

must have been
at the
and Uruguay
relatively wealthy
Argentina
in
income
Latin
Amer
of
the
the
average per capita
century,
beginning
in 1913
ica was about one-half
of that of the present OECD countries

while

and

in 1950,

while

the average

income

in Eastern

Europe

was

only

33
and Democracy
(Boston: Beacon Press, 1965).
Jr., Social Origins ofDictatorship
Barrington Moore
34
in office between parties occurred
is Czechoslovakia,
The exception
but note that no alternation
in the history of Eastern Eu
during this period. Indeed, the first alternation resulting from elections
rope occurred in Poland in 1991.

MODERNIZATION:

Approximate

THEORIES

6
Table
at
the
Per Capita
Income
Time of Democratization
in Some of the Present
oecd Countries3
First Democratization
Date

Australia (1901)
Austria

Belgium
Canada (1920)
Denmark

Finland (1917)
France
Germany

Italy
Norway (1905)
Sweden
Switzerland

United Kingdom
United States

173

& FACTS

1901
1918b
1919b
1920
1901
1919
1875c
1919
1919
1884
1918
1870d
1911e
1830f

Reversal

Level

Date

Level

3733
1545
2960
3838
2213
1184
1748
1072
1920
1228
1919
2226
3016
1119

Present

none

1934

1825

none
none
none

1930
none
1933
1922

1974
(?)
1474
1814

none
none
none
none
none

Democracy

Date
1901
1951
1919
1920
1901
1944
1875
1949
1946
1884
1918
1870
1911
1830

Level

3733

2535
2960
3838
2213
2636
1748
2567
1708
1228
1919
2226
3016
1119

"Levels are GDP/cap expressed in 1985 USD. They are calculated by extrapolating backward the 1951
for per capita GDP expressed in 1985 PPP USD, using the index numbers for GDP and the pop
in the Twentieth Century (Paris:
ulation figures provided by Angus Maddison,
The World Economy
is dated by (1) the presence of contested elections or
OECD, 1992), Appendixes
I, II. Democratization
on a partisan basis and (2) legislative
sovereignty of the house elected by broadest suffrage
ganized
came later, but not
to the crown or a nonelective
(rather than responsibility
upper chamber), whichever
numbers

or
For countries that became independent after 1871, dates in
participation.
by the extent of franchise
are for the year of
independence.
parentheses
b1920 figures were used.
in France to 1884, while John D. Stephens, "Democratic Transi
cTherborn (fn. 8) dates democracy
in Europe, 1870-1939:
A Test of theMoore Thesis," American Journal of Sociol
tion and Breakdown
as one of consolidation.
The question mark for France
ogy 94 (1989), refers to the period 1875-84
refers to the Vichy regime.
and Stephens
dBothTherborn
electoral register was established.

date democracy

in Switzerland

to circa 1880, when

the first national

uses 1911 to date democracy

in the United Kingdom,
but scholars who use universal
as the criterion date it to 1918. See Dahl,
suffrage
Democracy and Its Critics (New Haven: Yale
Press, 1989).
University
f
in the United States ranges widely, from 1828 by Huntington
The dating of democracy
(fn. 32), 16,
as a criterion, we date it early.Maddison
to 1970 by Therborn.
does
Since we do not take participation
not provide a figure for 1830; we interpolated
the numbers using 1820 and 1840.
eRobert Dahl

male

slightly higher.35All this is not much to stand on, but perhaps enough
to believe that (1) the levels at which democracies emerged before
World War IIwere highly scattered; (2) they did not differ between
and (3) once established,
and other parts of the world;
Europe
were more
to fall in the poorer countries.
democracies
likely

Western

35
World
Table 1.1.

Bank, World Bank Development

Report

1991

(Washington,

D.C.: World

Bank,

1991),

WORLD POLITICS

174
are on firmer

the second question.36 Compar

ground answering
are more
"new"
countries
shows
that democracies
the
and
the
"old"
ing
are more
to die in
brittle in the new countries while dictatorships
likely
We

the old ones. And, asTable 7 shows, the level of development again has
powerful effects. The probabilities of a democracy falling decline dra
matically with level in both groups of countries: indeed, this probabil
ity is the

same

once

countries

reach

an

income

$2,000. The
level among the

above

to
increases with
of a transition
democracy
probability
after
But among the countries
that became
old countries.
independent
are
as
are
are
as
when
when
stable
1950, dictatorships
they
they
wealthy
in new countries with
incomes above
fifteen dictatorships
poor. Among

$2,000, only one fell during their 185 years until 1990, in Suriname in

at $2,888,
in the Seychelles,
after 1990.
and only one more,
coun
the effect of levels at which
We may be confusing,
however,
tries were first observed
and the effect of development
they experienced
the
the new countries were much
under
scrutiny. And
during
period
the old ones?which
average income was $1,103?than
poorer?their
were
an
first observed. To
had
average income of $2,613?when
they
1988

of the tran
8 the derivatives
these effects, we show inTable
distinguish
to the entry levels and to the
with
sition
regard
separately
probabilities
since then.37 The
effects of the entry level are about the
development
are more
same for the two groups of countries. Democracies
stable and
more brittle in countries
either when
that were wealthier,
dictatorships
But the
in 1950 or whenever
first observed
they became
independent.
between
since the time of entry differ greatly
effects of development
increases much
the two groups of countries. The
stability of democracy
36
A

third question has also been posed: when D. A. Rustow, pointed out that the levels of develop
atwhich different countries permanendy
institutions vary widely, Lipset s
established democratic
was that the thresholds at which democracy was established were lower for the
(fn. 1,1981)
rejoinder
to Democracy,"
see Rustow,
A
"Transitions
Comparative Politics 2 (April 1970).
early democracies;
war indicates
at
the levels at which democracies
emerged before and after the
rough guess
comparing
that levels at which democracy was established before the war must have been on the average lower.
of incomes during the two periods was not the same: it is doubtful that many
But the distribution
countries enjoyed incomes above $4,000 before the war. Hence, we do not know how long the coun
democracies. At most, we can
tries that were poor at the time would have waited before becoming
occurred before the war with the distrib
of levels at which democratization
compare the distribution
inTable 6 we
ution in the postwar period truncated at $4,000. If in addition to the guesses presented
the two distribu
also assume that incomes were lower in Eastern Europe and most of Latin America,
tions will be highly similar. But that is too many guesses to take seriously.
37
to level is Pr[R?GIME(t)=DEMOC
relates regimes
that the function which
Suppose
where F stands for a normal or logistic distribution. Now sub
RACY]=REG(t)=F[a+?LEVEL(t)],
to get REG(t)=F{cc+?LEVEL(0)+
tract and add ?LEVEL(O)
within
the square brackets,

ment

?[LEVEL(t)-LEVEL(0)]}. Defining LEVEL(O) as INI andLEVEL(t)-LEVEL(0) asDEV(t), and

allowing the (cross-sectional)

ment
yields REG(t)=
F[a+?c

effect of the initial level to differ from the (dynamic) effect of develop
is the model we estimated, by dynamic probit.
INI+?DDEV(t)].This

Table
and Predicted
Observed
Regime
Lagged
Per Capita
Income
(level)

by
Transition
Probabilities,
and by Groups
of Countries51

PAD

PJK

TJK

TOT

0.0124

18

1448

0.0058

2543

(0.0058)
0.0359

Low-High

TRD

PDA

TRA

TD

1211

0.0464

11

237

1169

(0.1183)
0.0204

28

1374

TA

All
New

Old

70

0.0275

7
42

(0.0544)

(0.0340)

-2000
0.0135

New

Old

15

0.0377

32

1111

0.0058

848

(0.0058)
0.0297

6
18

1036

0.1200

75

606

(0.1480)
0.0578

14

242

(0.0279)

(0.0707)

2000
0.0086

New

Old

3
38

0.0225

347

0.0054

1695

(0.0058)
0.0427

1
24

185

0.0123

162

563

(0.0203)
0.0124

14

1132

(0.0676)
are used:
aThe following abbreviations
that either regime dies during
PJK is the probability
TTR is the number of transitions

(0.0156)
a

particular

of regime years at a particular

of transition to democracy

TOT is the total number

PAD is the
probability
TRD is their number
TA is the total number

level

of years under authoritarianism

of transition to authoritarianism

PDA is the
probability
TRA is their number
TD is the total number
New

year

of years under democracy

that did not exist in 1950

stands for countries

Old

stands for countries

Numbers

in parentheses

that existed

in 1950

are values
predicted

by the dynamic

probit model.

8
Table
of Transition
with
to the
Probabilities
Regard
the
and
Initial
Level
Development
Accumulated
("Entry")
at the Means
by Groups
of Countries51
Evaluated

Derivatives

with Regard to
Initial Level
Derivative

Old

of

-0.0554

Pda

aLevel ismeasured
for countries

in thousands. New

that existed

stands for countries

-0.0191

0.0112

0.0277

in 1950. Derivatives

New

-0.0966

-0.0552

0.0383

Pad

Development

Old

New

-0.0190

that did not exist in 1950. Old

are based on a dynamic

probit model;

see

stands

Appendix

2.

176
more

WORLD POLITICS
with

development

in the old

in the new

than

countries.

In turn,

while development decreases slightly the probability of survival of dic

to
in old countries,
the probability
of transitions
democracy
tatorships
as
new
countries
declines
under
authoritarian
rule.
develop
the promise
that development
would breed democracy
Hence,
proved

to be particularly futile precisely with regard to thoseThirdWorld coun

tries towhich itwas supposed to offer hope. Development during the
postwar period just did not have much of an impact on the collapse of
an increase

dictatorships:

of per capita

income

raised the probability of dictatorship

among
countries.

the old countries

and lowered

at least "modernization"

But

of one

thousand

dollars

falling by only 1.12 percent

it by 1.90 percent among the new

in the right direction
in
worked

where most
countries,
including
long-standing
dictatorships,
did in the end fall. Most
of the new countries,
those in Eastern Europe,
of them poor when
the great majority
just
they became
independent,

the old

remained

poor;

and those

few that did develop

remained

authoritarian.

V. Conclusion
Whether

in the language
theories
perspective,

couched

the historical
terministic.

In the modernization

democracy

it is secreted
about;
social
transformations.

corollary
s
Moore
theory, but they operate
class structure of the seventeenth

or
perspective
were
de
of the origins of democracy
no one does
to
bring
theory
anything
and the
by economic
development
of the modernization

Class

actors

at a distance

do move

of centuries:

in
history
the agrarian

the regimes coun

century
two or three hundred
ob
Przeworski39
later.38
As
years
this
made
both
deterministic
served,
appear
emphasis
approaches
on the
the issue of democratization
irrelevant when
political
appeared
in the struggles for democ
The protagonists
agenda in the mid-1970s.
that the fate of their countries would
racy could not and did not believe
tries

settle

determines

on

be determined either by current levels of development or by the distant

democratization
constraints,
that, albeit within
past. They maintained
was an outcome
the O'Don
of actions, not just of conditions.
Hence,
was
terms
actors
in
of
and
couched
nell-Schmitter
strategies,
project
rather than in terms of deterministic
conditions.40
38
and Stevens (fn. 6) go back just a few decades but the question remains:
Huber, Rueschemeyer,
in the 1920s cause events in the 1960s, not earlier or later?
conditions
would
found
why
39
Adam Przeworski, Democracy and theMarket: Political and Economic Reforms inEastern Europe and
Latin America (New York: Cambridge University
Press, 1991).
40
and Philippe C Schmitter, Transitions from Authoritarian Rule (Baltimore:
Guillermo O'Donnell
Press, 1986).
Johns Hopkins University

THEORIES

MODERNIZATION:
Our

& FACTS

177

this latter approach. The

emergence
strongly validate
is not a
of economic
Democ
democracy
development.
by-product
or
actors
is not established
their goals, and
racy is
pursuing
by political
once it is estab
at any level of
it can be initiated
Only
development.
findings

of

a role: the chances for the survival

play
of democracy
is richer. Yet even the cur
the country
greater when
to sur
rent wealth
ismore
of a country
is not decisive: democracy
likely
vive in a
income
with
than
less
$1,000
economy
per capita
growing
an income between
that de
than in a country with
$1,000 and $2,000
lished

do economic

constraints

are

democ
If they succeed in generating
development,
economically.
racies can survive even in the poorest nations.
de
the vision of the relation between
Viewed
from this perspective,

clines

that dominated
mood
and
the intellectual
and democracy
velopment
war years appears
the
cold
served to orient U.S.
foreign policy during
as exogenous,
treated development
strangely convoluted. While
Lipset
were
is the inevitable
his contemporaries
that dictatorship
persuaded
a
more democratic
"the
of
claimed
that
Galenson
price
development.
. . . the
resources
invest
of
from
the
diversion
is,
greater
government
ment
to
De Schweinitz
argued that if the less devel
consumption."41
must
limit democratic
"are to grow economically,
they
oped countries

participation in political affairs/'42And thiswas also the belief of Hunt

ington
The

and Dominguez:
interest

of

the voters
a

sonal

consumption
a nondemocratic

higher

system.

to
leads parties
the
give
vis-?-vis
investment
than
priority
In the Soviet Union,
for instance,
generally

expansion
itwould
the

of per
in
receive
of

percentage

GDP devoted to consumption was driven down from 65 percent in 1928 to 52

percent
tained

in 1937.
a revolution

It is unlikely
from above

that
like

competitive
this.43

party

system

would

have

sus

are needed
to generate
and
As Huntington
development.
Dictatorships
tem
at
must
least
be
held
Nelson
"Political
down,
put it,
participation
economic
in order to promote
development.,,44
porarily,
while
devel
in this view dictatorships
Since
generate
development
was
to
to
to
be a
said
the
best
leads
way
democracy
opment
democracy,
circuitous
41
Walter

one. Yet

common

sense would

indicate

that

in order

to

to Galenson,
"Introduction"
(New York:
ed., Labor and Economic Development
Galenson,
1959), 3.
Wiley,
42
Labor Controls
and Democracy," Economic Development
Karl de Schweinitzjr.,
Industrialization,
and Cultural Change 7 (July 1959).
43
in F. I. Greenstein
and
"Political Development,"
and Jorge I. Dominguez,
Samuel P. Huntington
Political Science, vol. 3 (Reading, Mass.: Addison-Wesley,
N. W. Polsby, eds., Handbook
1975), 60.
of
44
and Nelson
(fn. 21), 23.
Huntington

WORLD POLITICS

178

we should
not
democracy,
democracy
strengthen
even
that "common
if G. B. Shaw warned
And,
dictatorships.
strengthen

that

it is the best

time

this

in poor

flourish

is flat," the lesson of our analysis is

tells us that the world

that which

guide. With

can

democracy

development,

countries.

1:Classifying

Appendix

Political

Regimes45

some
is a regime inwhich
governmental
Democracy
a consequence
of contested
elections. This definition
fices"

support
sense is

offices

are filled as

has two parts: "of

and "contestation."

In no regime
elections. What

as a consequence
as democratic
a
regime

are all
offices
governmental
to
is essential
considering

filled

of
is

that two kinds of offices are filled by elections, whether directly or in

the chief
directly:
tive body.
Contestation
chance

executive

and the seats

in the effective

legisla

that has some

there exists an opposition
as a consequence
in
of elections. Whenever
as democracies
in
incum
which
only those systems
them.
did
lose
actually
occurs when

of winning
we
classify

doubt,
bent parties

office

office

a
was classified
as a
if none of the
democracy
regime
Operationally,
as a dicta
was classified
a
four rules listed below applied. Thus,
regime
at least one of these conditions
held.
torship if
1. Executive

Rule

the chief

selection:

is not

executive

elected.

Rule 2. Legislative selection: the legislature is not elected.

Rule 3. Party: there is no more than one party. Specifically, this rule applies if
(1) there were
tenure
in office

or

no

parties,
ended up

(2)

there

(4) the incumbents unconstitutionally

in their

rules

4.

Rule

Type

was

in the establishment

one
or (3) the
party,
only
or
of a nonparty
one-party

current
rule, or

closed the legislature and rewrote the

favor.
II error:

passes

regime

the

previous

three

rules,

the

incumbents

held office in the immediate past by virtue of elections for more than two terms
or without
they

have

and until
elected,
being
not lost an election.

Alternation
where

in office

today

overrides

at one

or the

time when

they

were

overthrown

the party rule. Hence,

Jamaica?
of the seats in the
100 percent

time held

party
lost an election?was
office having
yet
yielded
subsequently
legislature
as democratic
classified
during the entire period.
are the
We
code the regime that pre
rules
Our
timing
following.
came
even
to
at
if it
vailed
the end of the year,
31,
power on December
single

45
For a full explanation

and historical

details,

see Alvarez

et al. (fn. 10).

MODERNIZATION:

THEORIES

& FACTS

179

to
in 1983. Transitions
arrived in Nigeria
dictatorship
a
are
etat.
to
democ
Transitions
coup d
signaled by
are
of
the
time
elected
of
dated
the
the
gov
racy
inauguration
newly
by
a democratic
not of the election.
In the few cases where
ernment,

as, for example,

authoritarianism

regime lasted sixmonths (for example, theDominican Republic in 1963)

or where

several times (Bolivia in 1979), the infor

changed
that
the same year is lost.
regimes
began and ended within
our
is
and the alternatives
The main
difference
between
approach
rather than a polychotomous
that we use a dichotomous
classification,
are more
some
than
democratic
that while
scale. We
believe
regimes

mation

the situation

about

not be considered
de
others, unless offices are contested,
they should
mea
mocratic.
from a practical point of view, alternative
Nonetheless,
sures of
dimensions
similar results. The
generate
highly
democracy
or to what
extent a
is demo
used to assess whether
regime
particular
no
measure
seem
to make
the
is
cratic
little difference.46 Our
exception:

Coppedge-Reinicke47
the Bollen48

1965

scale for 1978 predicts 92 percent of our regimes,

scale predicts

85 percent,

the Gurr49

scales

of Au

tocracy and Democracy for 1950-86 jointly predict 91 percent. The

Gastil50 scale of political liberties, covering the period from 1972 to
his scale of civil liber
93.2 percent of our classification;
1990, predicts
ties predicts
91.5 percent;
and the two scales jointly predict 94.2 per
cent of our
there is no reason to think that our results
regimes. Hence,
are
to the
classification
of regimes.
idiosyncratic
particular

Appendix

2: Dynamics

of Regimes

in the distinction
be
is entailed
help elucidate what
mechanisms.
and
exogenous
endogenous
has an authoritarian
i= 1,...,N,
that a country,
Let the probability
a
t=
the subscript
1,...,T, be pA(it), where
regime during
particular year,
that it has a demo
and the probability
A stands for "authoritarian,"
Some

algebra may

tween

46
25 (Spring 1990),
Alex Inkeless, "Introduction," Studies in Comparative International Development
3-6. Note, however, that different measures appear to be biased in somewhat different directions. See
Factors in Cross-National
Kenneth A. Bollen, "Liberal Democracy:
Measures,"
Validity and Method
American Journal of Political Science 37 (November
1993).
47
and Wolfgang
H. Reinicke,
Michael
"Measuring Polyarchy," Studies in Comparative
Coppedge
25 (Spring 1990), 51-72.
International Development
48
American Sociologi
of Development,"
and the Timing
Kenneth A. Bollen, "Political Democracy
cal Review 44 (August 1979), 572-87.
49
Ted Robert Gurr, Keith Jaggers, andWill H. Moore,
and State Power
The Growth of Democracy,
Autocracy,
tional Development
25 (Spring
50
Gastil (fn. 4).

1990).

State:
"The Transformation
of theWestern
since 1800," Studies in Comparative Interna

WORLD POLITICS
cratic regime be pD(it) = 1
the probability that a dictator
pA(it). Let
ship dies from one year to another be pAD(it), so that the probability
that it survives is p^iit) = 1 pAD(it). Similarly, let the probability that
= 1a democracy dies be
we assume for the time
pDD(it). If
pDA(it)
180

=
are
k = A,D,
that these "transition
being
p.k, j A,D,
probabilities/'
over time and the same for all countries,
constant
then we can describe

the evolution of regimes by

pD(t+D

Pdd

Pad

Po(t)

pA(t+D

Pda

Paa

PaW

The

of regimes

proportion
therefore on the

that

are democracies

of democracies

next

that survived

year depends
from the cur

proportion
rent year,
that died, that is, be
of dictatorships
pDD, and the proportion
came democracies,
same holds for
The
pAD.
dictatorships.
of regimes that,
Given
the transition
rates, there exists a distribution
disturbances.
if reached, will remain stable in the absence of exogenous

These equilibrium probabilities are

=
Pd

Pad

+
Pda Pad

and
Pda
Pa

Pda
Moreover,
tions will

Pad

whatever

the

over

tend

time

initial

their propor
of regimes,
since the
values. And
equilibrium

distribution

to these

are likely to be
probabilities that regimes die during any particular year
fact they are low?this
convergence
of one type of regime will
the proportion
the other to decline.51
low?in

will

be monotonie;
to increase

continue

that is,
and of

of regimes depends
distribution
time passes,
then, the long-run
on
not
on the relative rates at which
their initial distrib
they die,
only
As

51
is monotonie
Convergence
around the equilibrium.

if Pad + pDA < 1; otherwise,

the proportions

of regimes will

oscillate

MODERNIZATION: THEORIES & FACTS

ution.

>

If Pad

in the long run there will

pDA, then

be more

181
democracies

than dictatorships in the world and if at the beginning the proportion

was

of democracies
crease over

lower

than

this proportion

pD*,

will

continually

in

time.

now that while

die at some constant
annual
dictatorships
Suppose
= 0. You see
never
so
democracies
that
that in
rate,
die,
immediately
pDA
run
a
time
the long
all countries will be democracies.
dictator
Every
a
it sur
is established,
and, once it is established,
ship dies,
democracy
on
rate
at
vives forever. The
of
the
which
this
process depends
speed
is inexorable.
the
of democracies
but
accumulation
die,
dictatorships
to return

Now,

to the

are not

probabilities
To keep matters

at stake,
that these transition
imagine
but depend on the level of development.
are
two levels: low (L)
suppose that there
only
issue

constant

simple,

and high (H). At the low level, both regimes have some probability of
dying

that

ismore

than zero

and less than one. Now

consider

two pos

sibilities.
< 1, once
One is that while p^L)
dictatorships pass the threshold
=
that defines the high level, they are certain to die, so that p^H)
1,
while

democracies

probabilities

die

Level = Low
Pdd
Pda

and while

at the same

rate at either

level. The

transition

are thus

Pad
Paa
the long-run

Level =High
LOO
Pdd
0.00
Pda
proportion

of democracies

at the low level will

=
+
at the high ievel itwill be
pD*(H) 1/(1
Pm/?Pad'PdaX
<
pDA), pD*(L) pD*(H). Thus, the proportion of democracies will be
be pD*(L)

are more
at the
because democracies
high level of development
as a result of
to emerge
is the
This
likely
development.
endogenous?
of the explanation.
modernization?version
higher

at
But suppose alternatively
that authoritarian
exactly the
regimes die
same rate whether
or
so
in poor countries
that
ones,
developed
p^L)
=
=
never die once
in turn democratic
while
they
regimes
pAD,
pAD(H)
are established
so that
in affluent countries,
transition
The
pDA=0.00.
are then
probabilities

Level = Low

Level =High

Pdd

Pad

LOO Pad

Pda

Paa

0.00

Paa

WORLD

182
and we

already

that while

know

POLITICS
the long-run

of democra

proportion

cies at the low levelwin be pD*(L) < 1, at the high level all countries will
we will observe
an
in the long run. Hence,
the level of development
relation
and the incidence
aggregate
are
even
to emerge at
democracies
of democracies
though
equally likely
even
if
under
that
authoritarianism
does not
is,
any level,
development
a
increase the
that
is
democratic.
This
country will become
probability
then the exogenous
version.

have

a democratic

to decide

Thus,

regime
between

which

mechanism

the relation between

generates
to
need
determine
how the respective
with
the level of development.
To esti

we

and democracy,
development
transition
change
probabilities
we
mate
the impact of level on transition
probabilities,
a first-order
data
Our
Markov
processes;
obey
myia.52
on the
present
regime depends
regimes during
only
are defined
but not beyond.
Such processes
by:

on Ame
rely
that is, the

the previous

year,

=
P(t)Rt.1,
E(R=l|R.1,Rt_2,...)
=
where R
for democracy
stands for regimes, R=D
and R=A for
D,A
the
and
is
matrix
of
transition
with
ele
P(t)
dictatorship,
probabilities,
ments
p k(t). Hence,

R = P(t) R,
Taking

u,
of both

expectations

sides yields

P(R=D

Paa

Pda

p(R,rD

p(R=0)

Pad

Pdd

p(Rt-rO)

sum of columns

where

the

k=0,l.

Hence

of the

transition

matrix,

=
+
p(R=l Rtl)
p^top?R^-l) p^WR^O)
|
=
=
+
PAAttpdVrD PDA(t)[l-p(R,rl)]
=
Now

Pda?

=
?.p.k

1, j=0,l;

[PAA^-PDA^?-rD

let X be the vector

of the exogenous

variables.

Assume

that

=
pDA(t) F(XA ?),

=
PAA(t) F[XJcc+?)],
52
Takeshi
chap. 11.

Amemyia,

Advanced

Econometrics

(Cambridge:

Harvard

University

Press,

1985),

MODERNIZATION: THEORIES & FACTS

183

where F(.) is the CDF of normal distribution. Note that p.k(t) is the
probability of transition from being in state j at time (t-1) to being in
state k at time
regime

t. Given

as the one

that whenever

a transition

installed

that became

during

occurred
this year,

we

code

the

the probabil

ity of transition between (t-1) and t depends on the conditions at (t-1).

Hence,

we

lag the X

p(R=l

|RJ

s.

Then
=

pDA(t)

[pAA(t)-pDA(t)]p(Rt.1=l)

FCX,.^
FQit_?

{FtX^a+j^-FQi^pCR^l)
=
+
FQ^aMR^l)
?Q?tJ

xt,R,?).
to estimate a and /?, from which
Hence,
and thus pDD = 1 pDA and pAD = 1 -p^,

one can calculate

all we

need

pDA and pAA,

on
to do is
probit

This is the model we used to generate results inTables 2 and 8,with

R(0) as observed. The derivatives used inTable 8 are

dPpA =
dpAD =-f
+
fQC^?)? and
[XJa ?)](a+?).
dX
dX