POLITICAL COMPETITION AND
ECONOMIC GROWTH
Bruce Bueno de Mesquita, James D. Morrow,
Randolph Siverson, and Alastair Smith
Economists long ago established that competition fosters efficiency
and growth. This is no less true in the nonmarket world of politics.
Political competition promotes economic growth and prosperity when
the focus of such competition is ideas for improving the welfare of a
broad base of citizens through the provision of market-enhancing public
policies. When leaders hold power for long periods, this typically reflects
institutional arrangements that stifle just such competition, rewarding
cronyism and corruption instead. Competition for national leadership is
a hallmark of economic growth, but it also creates problems for leaders
who aspire to maximize their time in office. This leads to a central paradox
of politics: Under many political systems, good policy is bad politics,
and bad policy helps leaders stay in office. In those (mostly democratic)
systems in which good policy is also good politics, leaders face greater
obstacles to maintaining incumbency than their counterparts in nondemocratic
systems.
We outline here a theory that ties political incentives to the interest
that leaders have in staying in office. Specifically, we examine how political
institutions shape decisions about the allocation of government
resources in ways that promote public welfare or that support cronyism,
corruption, and kleptocracy.
Every political system depends, in part, on two institutions. One is
the selectorate and the other is the winning coalition, the subset of the
selectorate that keeps a ruler in power.1
The selectorate is the set of
Bruce Bueno de Mesquita is senior fellow at the Hoover Institution
at Stanford University. James D. Morrow is professor of political
science at the University of Michigan. Randolph Siverson is professor
of political science at the University of California–Davis. Alastair
Smith is assistant professor of political science at Yale University. This
essay draws from their forthcoming book, tentatively titled Staying
Alive: Political Institutions and Leadership Incentives.
Journal of Democracy Volume 12, Number 1 January 2001
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 59
citizens who have a chance of becoming members of the ruler’s winning
coalition. In a political system with universal adult suffrage, such as
prevails in most democracies and in rigged-election autocracies, any
citizen has some chance of rising to political influence. In monarchies,
military juntas, and many autocracies, by contrast, only a small portion
of the population has a chance of becoming influential in politics or of
gaining access to the benefits that involvement in politics can bring.
It is easy to identify the selectorate in democracies. All enfranchised
citizens fall within it. It is less obvious in other political systems. A few
examples may help clarify the concept. Following the death of England’s
King Richard I on 6 April 1199, a successor had to be chosen from among
three candidates. The responsibility of appointing a new monarch fell to
197 barons and the pope. This tiny group—and no one else—constituted
the selectorate. In the former Soviet Union, the selectorate consisted of
something close to the entire electorate. That is, just about any Soviet
citizen had a chance (albeit a tiny one), to rise to political prominence.
This was important, for it allowed the ruling elite to pick ready substitutes
for any disloyal member of the winning coalition.
The winning coalition, made up of a portion of the selectorate, has
two distinct qualities. First, its support is essential for the incumbent to
stay in office. If members of the winning coalition defect to a rival and
new members cannot quickly be added to replace them, then the incumbent
is deposed and the rival comes to power. Second, members of the
winning coalition are accorded privileges unavailable to nonmembers.
Specifically, members share in any of the private benefits that the leadership
distributes. These benefits include opportunities for corruption,
black marketeering, and so on.
In most democracies, the winning coalition consists of the set of voters
whose support is essential to selecting and keeping a leader in office. In
a plurality-voting system like the United States, for example, Bill
Clinton’s 1992 winning coalition included about 43 percent of the
electorate (or, perhaps, all of his supporters up to just beyond 38 percent,
the share of the vote earned by George Bush). Though these people
received few and small private benefits, we can still think of Clinton’s
legislative agenda as partially motivated to shift resources disproportionately
to these constituents through taxation, welfare programs, and
other redistributive policies. In an autocracy like Saddam Hussein’s Iraq,
by contrast, the winning coalition is tiny. Though Iraq has universal
adult suffrage, keeping Hussein in power depends on the support of the
Republican Guard and the Takriti Clan—that is, his close personal kinsmen,
a small group indeed. In King Leopold’s Congo during the latter
third of the nineteenth century, the winning coalition was made up of
the Force Publique, Leopold’s small group of henchmen who maintained
the extractive and exploitative regime that he put in place to steal the
wealth of the Congo.2
Journal of Democracy60
Maintaining the loyalty of the winning coalition is crucial for any
political leader. How best to do so depends on the size both of the
selectorate and of the winning coalition. We will show when loyalty is
best achieved by giving coalition members special privileges (“private
goods” in the vocabulary of economics) and when it is best achieved by
implementing public policies that raise the welfare of the whole society.
Before turning to the development of our theory, we pause to help
clarify the distinction between nominal regime types (like democracy,
autocracy, and monarchy) and our institutional variables (the size of the
selectorate and the size of the winning coalition). Most large-coalition
systems are democracies, just as most small-coalition, large-selectorate
polities are autocracies. Systems with a small coalition and a small selectorate
tend to be monarchies or military juntas. Because it is simple
enough to relate coalition size and selectorate size to well-known regime
types, one might wonder whether we are just adding complexity to wellestablished
distinctions.
The correlation between the size of a coalition or selectorate and
regime type is useful as a heuristic device, but no two democracies are
alike, nor are any two autocracies, monarchies, or juntas. Inevitably,
categorical discussions of regime type lead to the construction of
arbitrarily drawn boundaries. That is, of course, why there are so many
different ways in which people define democracy. Our two institutional
variables are far more fine-grained. Though difficult to estimate in
practice because little effort has been put into their measurement to date,
they are essentially continuous—rather than categorical or ordinal—
variables. This means that they have both conceptual and mathematical
properties that facilitate drawing generalizations about the marginal
impact of even small changes in their values on important political factors.
This is not as true of the more common regime typologies. Therefore,
as we learn more about estimating the values of the winning coalition
and the selectorate, we will learn more about politics than is possible by
focusing on categorical schemes for defining regimes.
There is an additional, perhaps more important, reason to distinguish
between (continuous) institutional variables, like the selectorate and the
winning coalition, and (categorical) regime types. While most systems
with large winning coalitions and large selectorates are democratic, a
large selectorate and a large coalition do not define democracy per se.
Democracy is generally associated with a variety of other characteristics,
such as an independent judiciary, a free press, civil liberties, legal constraints
on leaders, and reliance on law, which are not part of what defines
the winning coalition or the selectorate. Rather, many of these features
are themselves expected policy consequences of a large winning coalition
and a large selectorate. Equivalently, although autocracy is not defined
by the presence of corrupt politicians whose actions imply indifference
to public welfare, we will show that such behavior is an expected
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 61
consequence of having a small winning coalition and a large selectorate.
Ours is thus not a theory of democracy or autocracy dressed up in new
clothes. It is a theory of political institutions that considers the size of
the selectorate and that of the winning coalition as a means of accounting
for many of the characteristics associated with different types of
regimes.
Before returning to the development of the selectorate theory, we
offer a few examples that highlight differences between the measurement
of coalition size and the assessment of democraticness drawn from the
widely used Polity data set (see the Appendix on p. 71 for further details).3
In doing so, we emphasize a general principle: At the extremes of democracy
or autocracy, there is an almost perfect correlation (0.9995) between
our indicator of coalition size and the type of regime.4
In the interior of
both indicators, however, the correlation is only 0.63. It is in the subtleties
of nearly democratic (or only partly autocratic) systems that the indicators
and concepts of coalition size and democracy diverge.
Gorbachev’s Soviet Union earned a score of 0.75 on our coalitionsize
indicator in 1990. On our measure of democracy (see note 4) his
regime scored only 0.50 in 1990, but 0.90 a year later. Apparently, the
indicator of coalition size captured the change in Soviet political arrangements
more quickly than did the standard democracy index. Prior to the
presidency of Ulysses S. Grant, the United States generally scored 0.95
on democracy and 0.75 on coalition size. The coalition indicator does a
better job of capturing the restricted American franchise before the Civil
War than does the democracy indicator. Lee Kwan Yew’s Singapore
(1965–90) scored 0.75 on the coalition index, but only 0.40 on democracy.
The higher score for coalition size can be read as either indicating
a more broadly based government than Lee and his party actually had
or, alternatively, as capturing a breadth of popular support for his government
and its programs that is missed by the democracy score. Taiwan
also had a larger coalition score (0.25) than democracy score (0.10)
during Chiang Kai-shek’s rule. Both scores are very low (indicating very
little democracy and a small coalition), but still the relatively larger
coalition score favors somewhat more public-spirited, less corrupt policy
than does the democracy-autocracy indicator. Later, we return to these
differences in indicators and discuss the example of India to illustrate
how coalition size and democracy can differ in practice.
Resource Allocations
Governments raise resources through taxation. These resources are
presumably allocated by political leaders in a way that helps them stay
in office. Leaders can invest resources in the production of public goods
or in providing private benefits for their supporters. Whereas all members
of society benefit from public goods, only members of the winning
Journal of Democracy62
coalition enjoy the fruits of private goods. Public goods include national
security, property rights, individual liberties, the rule of law, public
health, education, and other social services. Private goods extend special
privileges to selected citizens, such as idiosyncratic, personal protection
from the law, advantages in business transactions, nepotism, cronyism,
rights to corrupt practices, and the like.
With an eye toward staying in office, the incumbent ruler decides
how much public policy to purchase and how many private goods to
generate, given the available resources. Members of the winning coalition
continue to support the incumbent, provided the benefits that they receive
outweigh those they would expect from a successful challenger. That
expectation depends on two factors. One is how much the challenger
can credibly offer. (This, of course, is limited by the available budget of
resources.) The second is how readily any member of the winning coalition
believes he will be essential to a new government and so continue
to get private benefits if he defects. As we will see, this risk of exclusion
from benefits is closely linked to the size of the selectorate and the size
of the winning coalition.
The larger the selectorate, the larger the pool from which a leader
can draw supporters. Thus, given a certain size of the winning coalition,
any individual’s chance of becoming an essential member of the winning
coalition is lower when the selectorate is big than when it is small. By
contrast, when the winning coalition is large, as in an electoral democracy,
many supporters are needed, and so each individual in the selectorate
has a better chance of joining the winning coalition. This means
that a defector in a small-coalition, large-selectorate system, such as a
rigged-election autocracy, runs a high risk of losing access to the private
benefits provided by the leadership, not to mention, in extreme cases,
the risk of losing her or his life. In a democracy, the risk of losing such
access is comparatively small (equaling about one-half), while there is
virtually no risk of losing one’s life.
The allocation of resources that maximizes incumbents’ chances of
political survival determines the degree to which they emphasize good
public policy versus cronyism and corruption. When the winning coalition
is small, a leader only needs the support of a few individuals to stay
in office. Suppose that, under such circumstances, a leader decides to
allocate the majority of resources to private goods. Since the winning
coalition is small, each supporter receives a relatively large proportion
of the available resources. As the size of the winning coalition increases,
however, each member’s share of the private goods shrinks. Once the
size of the winning coalition increases beyond a certain point, the provision
of private goods is no longer an efficient way to enrich supporters.
Leaders of large winning coalitions are thus better able to enrich their
supporters by investing a greater proportion of available resources in
public goods. But when public goods are provided, everyone in the
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 63
society benefits. The logic is straightforward: When the winning coalition
is small, leaders are more inclined to provide private benefits, rather
than public goods, because such an allocation better serves their desire
to stay in office. When the winning coalition is large, leaders must be
more concerned with the provision of public goods, not out of civicmindedness,
but because providing such goods is compatible with their
desire to stay in office. Thus large winning coalitions, such as exist in
democracies, encourage attention to the quality of public policy.
Leaders of small winning coalitions who emphasize good public policy
over bribing supporters are likely to lose to a challenger as members of
their winning coalition defect to get a better deal. Small winning
coalitions discourage emphasis on good public policy because such a
focus diverts attention from them, raising the risk of defection and thus
of the incumbent’s being removed from office. Suppose that one or more
members of a small winning coalition defect and the incumbent is
removed from office. A challenger now has the opportunity to form a
new government, for which he or she must draw enough supporters from
the selectorate. Since the winning coalition is always smaller than the
selectorate, defectors cannot be certain of making it into the new winning
coalition. Many of them may be weeded out as they prove irrelevant to
the new coalition. There is thus a risk and a cost associated with political
defection. The risk involves the chance of exclusion and the cost entails
being cut off from the future stream of private goods that flows to members
of the new winning coalition. As the size of the winning coalition
becomes smaller, or the size of the selectorate becomes larger, challengers
are less likely to rely on any particular individual in forming
their winning coalition. Hence, as the size of the winning coalition shrinks
or as the size of the selectorate grows, defecting becomes riskier.
Political systems with large selectorates and small winning coalitions
(like many autocracies) thus induce a norm of loyalty toward the incumbent
leadership by making defection from the current winning coalition
unattractive. Loyalty is a powerful force in autocracy. It insulates leaders
from being turned out of office just because they do a poor job on public
policy. They are protected as long as they do not “squander” resources
on economic development, using those resources instead to bribe their
few essential supporters. Loyalty to the incumbent declines as the winning
coalition grows relative to the selectorate because the cost and risk
of defection declines.
In democracy, the value of individual private goods is small because
they must be distributed among so many people. Lacking the ability to
secure the loyalty of their supporters through cronyism and corruption,
democratic leaders must compete over the provision of successful
policies. Or, to put it more precisely, leaders of systems dependent on a
large coalition work harder at producing effective policies, spend more
public resources to produce public goods, enjoy less loyalty, and get
Journal of Democracy64
turned out of office more frequently than their small-coalition, autocratic
counterparts. For autocrats, bad policy is often good politics because
their focus on cronyism and corruption ensures their enduring leadership.
This is almost never true for democrats.
Several implications follow from the theory sketched above. Large
coalitions induce leaders to emphasize public welfare, while small coalitions
encourage leaders to focus on private goods. The few autocrats
who do focus on public goods tend to abbreviate their tenure in office as
compared to those who emphasize corruption and cronyism. Because of
a weaker loyalty norm, leaders of large coalitions spend more, on average,
than those who head a small coalition. The opportunity for kleptocracy
grows when the gap between government revenue and expenditures
grows. Because small-coalition leaders, including autocrats and monarchs,
are predicted to spend less per capita than large-coalition leaders,
such as democrats, theft of society’s resources is greater in smallcoalition
systems than in large-coalition systems.
If given the chance to select the political institutions, leaders who
want only to stay in office would erect polities with small coalitions and
large selectorates, emphasizing rigged elections with universal suffrage
rather than truly competitive democratic polities. In contrast, if coalition
members get to choose institutions, they would favor systems with a
weak loyalty norm, meaning either small coalitions and small selectorates
(monarchy, military junta) or large coalitions and large selectorates
(democracy, and not autocracy). Those excluded from the winning
coalition, including selectorate members and the disenfranchised, prefer
institutional arrangements that encourage the production of public goods.
Thus they favor democracy and, if they could vote with their feet, would
be expected to emigrate from autocratic systems to democratic polities
with easy rules for citizenship.
To maximize his tenure in office, a leader must allocate resources in
a manner compatible with the institutional arrangements of the type of
regime. Civic-mindedness, therefore, is neither necessary nor sufficient
to produce successful public policies; what is essential is a system that
makes leaders rely on a large winning coalition and a large selectorate.
As mentioned earlier, although these characteristics do not define
democracy, they are most frequently found in democracies.
Unfortunately, other aspects of democracy—such as bloc or ethnic
voting, or subtly restrictive rules of enfranchisement—may detract from
the benefits of a relatively large-coalition, large-selectorate political
system. For example, bloc-voting systems can give the impression of
being highly democratic, while leaders actually rely on a more limited
coalition. India, for instance, briefly scored as low as 0.85 on our
democracy index (in the late 1970s, when Indira Gandhi attempted to
alter the Indian constitution to strengthen the executive), while generally
scoring between 0.90 and 0.95. In our data set, India’s democracy score
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 65
never reaches the upper bound of 1.00. On our indicator of coalition
size, India consistently earns a score of 0.75. That is, the country’s
coalition size is consistently estimated to be smaller than its free, competitive
electoral system or degree of democracy suggests. In many (if
not most) Indian constituencies, a local patron can deliver a bloc of votes,
thereby requiring political leaders to satisfy the needs of the few patrons
rather than the many voters. Thus, while the system appears to require
the prime minister’s party to muster a huge coalition, in reality, a relatively
small group of key voters (bloc leaders) is all that is needed to elect
and maintain the leadership. The same was true in Boss Tweed’s New
York, Mayor Daley’s Chicago, and the PRI’s Mexico (until the 2000
election). In such “democratic” systems, pork, patronage, and cronyism
play an important role that reduces the size of the winning coalition and
thus diminishes the incentives that leaders have to enhance public
welfare. Yet such systems frequently earn a high “democracy” rating by
standard measures.
Political Survival and Policy Performance
In democracies and autocracies alike, economic growth increases with
turnover in power. The uncertainty provoked by instability might be
thought to scare investors, but it is offset by the advantages inherent in
political competition, even when that competition occurs in an autocratic
environment. Although leadership turnover is generally beneficial for
growth, the benefits are smaller in autocracies than in democracies. Just
how growth rates vary according to coalition size is the subject to which
we now turn our attention.
It is natural to consider successful those leaders who foster economic
growth and prosperity for their citizens. By contrast, leaders who produce
famine, poverty, and misery seem like dismal failures who ought to be
removed from office as quickly as possible. Yet the irony is that leaders
who produce poverty and misery keep their jobs much longer than those
who make their country richer.
We divide the world into two categories—largest-coalition, most
democratic governments and smallest-coalition, most autocratic governments—and
compare each group’s track record in producing prosperity.5
To evaluate prosperity, we examine the average annual real growth rate
in per-capita income delivered by the 176 most democratic leaders and
the 179 most autocratic leaders since 1952. The average democratic
leader produces a real growth rate of over 3.04 percent per annum,
compared to 1.78 percent for the average autocrat.
The difference in economic growth rates tells a dramatic story. Two
countries starting out with the same average per-capita income in 1950
would look radically different in 2000 if they had different forms of
government. Suppose the average resident in a democracy and an
Journal of Democracy66
autocracy each had an income of $500 in 1950 and suppose real incomes
grew at the rate typical for the type of government. In 2000, the “democratic”
citizen would have a real per-capita income of about $4,480,
whereas the “autocratic” one would have a real income of only $1,200.
Assuming that each kept its form of government and that the average
performance of these government types did not change, by 2050 the
“democratic” citizen’s real per-capita income would be expected to grow
to over $10,035, whereas the “autocratic” citizen’s would rise to just
$2,910—a 3.5 to 1 advantage. Apparently, prosperity for the average
citizen seems closely tied to democracy, and poverty to autocracy.
The advantages for citizens from large-coalition, democratic systems
are much broader than what an evaluation of growth in per-capita income
alone might suggest. Polities dominated by a small winning coalition,
such as autocracies, do markedly worse on average than large-coalition
systems on a host of indicators of public welfare. For example, more
democratic, large-coalition systems generally do better than more
autocratic systems at producing safe drinking water, offering access to
medical care, encouraging free trade, avoiding corruption and black marketeering,
attracting investors, and so forth. Each of these effects remains
strong across time and geographic locale, even after controlling for percapita
income. That is, even poorer large-coalition societies offer more
of these advantages than do their autocratic counterparts.6
With individual prosperity and general social welfare closely associated
with political institutions, we might well ask why all leaders do not
adopt the governance structures that encourage the creation of healthy,
educated, prosperous citizens. Standard economic accounts conceive of
leaders as the benign agents of their citizens. They explain failures to
adopt effective policies by appeals to ignorance or ideological blinders.
We offer a different account.
What does political survival look like from the viewpoint of largecoalition
and small-coalition leaders? With rare exceptions, only autocrats
hold on to power for a really long time. The Figure on the following page
looks at political survival by assessing the percentage of leaders in democracies
and autocracies who have lasted for particular periods in office.
The results are striking. For instance, 100 percent of those lasting 40 or
more years are autocrats (although only a little more than 2 percent of
autocrats actually hold on to power that long). Yet two facts are clear:
First, nearly half of all democratic leaders are out of office within about
one year of coming to power. Such a short tenure is true of only about
one third of autocrats. This is a remarkable difference in survivability.
Second, virtually no democrats, but one quarter of autocrats, stay in office
for more than eight years. Autocratic systems seem to favor long-term
political survival.
A look back at the twentieth century confirms this. Over the past 40
years, the world’s most autocratic governments have averaged only seven
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 67
leaders, some countries managing only one or two leaders in that time
span. Fidel Castro, for instance, has led the Cuban government for four
decades, running its economy into the ground the whole time. Since its
revolution, China has had only five leaders (and strong growth for two
of the five decades). In the more than 75 years since Lenin’s death, Russia
has had only nine leaders (and poor growth for most of its history). By
contrast, established democracies like the United States and Britain have
had many more leaders since the end of World War II. The average
number of leaders in the most democratic countries since 1961 is 12.
As troubling as the differences in prospects for political survival are
across regime types, even more disturbing is the relation of policy
performance to political survival. Consider the survival rates of leaders
who perform exceptionally well or exceptionally poorly when it comes
to promoting public welfare. The selectorate theory suggests that those
who head a large-coalition government improve their prospects for
political survival by promulgating effective public policies. Those who
head a small-coalition government, however, enhance their survival by
distributing private goods at the expense of effective public policy.
We define as exceptionally effective those leaders whose countries
both avoid involvement in military conflicts and enjoy growth in percapita
income in the upper quartile of all countries. We define as
ineffective those leaders whose countries have black-market exchangerate
premiums in the upper quartile. Table 1 on p. 69 compares survival
rates in office for leaders of large-coalition (at least as large as 0.75)
and small-coalition systems, conditioned by whether they offer exceptionally
effective or exceptionally ineffective policies. That is, we ask
the following question: For how many years do the leaders of largecoalition
systems and those of small-coalition systems have a 50 percent,
FIGURE—SURVIVAL BY REGIME TYPE
PERCENTAGEOFGOVERNMENTTYPE
AUTOCRAT
DEMOCRAT
1 2 4 6 8 9 10 15 20 30 40 50 60
YEARS IN OFFICE
50
40
30
20
10
0
Journal of Democracy68
25 percent, and 10 percent chance of remaining in office if they produce
public or private goods at exceptional levels? Table 2 displays the same
assessment, but is based on whether a leader heads a democratic or
nondemocratic government.7
The lessons portrayed in Tables 1 and 2 are depressing. Truly bad
policy turns out to be good politics for small-coalition leaders and for
autocrats. Small-coalition autocratic leaders who promote or protect the
black market have a 10 percent chance of lasting in office for at least 19
years. If they follow good policies instead, their 10 percent survival time
drops to 13.5 years for small-coalition leaders and to 16.5 years for
nondemocrats. Good policy is bad politics for small-coalition leaders and
almost as bad for nondemocrats. Early in their terms, small-coalition
leaders and autocrats benefit a little from good policy, but once they
secure their hold on power, they do much better at keeping their jobs by
emphasizing a vibrant black market rather than income growth. The effects
are more pronounced for small coalition size than for nondemocracy.
For those dependent on a large coalition and for democrats, the opposite
is true, although again, the benefits of good policy are more pronounced
as a function of coalition size than as a function of democracy.
Though at a survival disadvantage compared to small-coalition leaders
or autocrats, those who head large-coalition governments or democracies
improve their chances of staying in office by promoting growth in real
per-capita income and avoiding an emphasis on private-goods production
such as typifies economies with vibrant black markets. Large-coalition
leaders with exceptionally good policies have a 10 percent chance of
surviving in office for 12.5 years. If they foolishly follow the bad-policy
approach that is good politics for small-coalition leaders, the largecoalition
cohort reduces its outer (10 percent) survival prospect to only
eight years. The comparable figures for democratic leaders are 9.5 years
for exceptionally good policy and 7.5 years for exceptionally bad policy.
A Tale of Two Leopolds
Politicians, like everyone else, are concerned with their personal welfare.
Peace and prosperity may be good for most people, but they are
not good for many leaders. Systems that encourage poor policy performance
also provide a strong incumbency advantage. This derives from
the strength of the loyalty norm in autocratic systems, which have a
small coalition and a large selectorate. Leaders undoubtedly understand
this. That is why, given the chance, most would prefer to establish a
small-coalition, large-selectorate autocracy rather than a democracy.
A cursory look at postcolonial Africa supports this inference. Although
many anticolonial independence movements established their newly won
states as democracies, almost all switched quickly to rigged-election
autocratic systems. The few exceptions, like South Africa and Zimbabwe,
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 69
involved places where freedom fighters had to negotiate a power-sharing
deal with their foes. The selectorate theory shows that when broad coalitions
form institutions, they do not opt for rigged-election autocracy
because such a system, possessing a strong loyalty norm, requires incumbents
to spend less on the public than more democratic systems. But
when a leader rather than a coalition chooses, the preference is for the
strong loyalty norm generated by a small-coalition autocracy. To illustrate
this point, we consider briefly the institutional constraints and policy
choices of Leopold II of Belgium.
Leopold II was king of Belgium from 1865 to 1909. He was also the
personal owner and ruler of the Congo Free State from 1885 to 1908.
When he acceded to the throne, Belgium was already a constitutional
monarchy, but the king still wielded considerable authority. Leopold
proved skillful in its use. Belgian governments initially depended on a
fairly large selectorate for the time. During his reign, Leopold helped
promote universal adult male suffrage in free, competitive elections,
thereby greatly expanding the selectorate and the winning coalition. This
suggests an emphasis on providing public goods at home.
By most accounts, Leopold was an excellent leader. Belgium experienced
exceptional economic growth and rapid industrialization under
his leadership. Fueled by a free-trade economy, imports and exports
expanded at a remarkable pace. As early as 1866, Leopold declared his
intention to improve the lot of the working class, and by 1873, in response
to a downturn in the business cycle, workers were given the right to
strike. In 1886, Leopold strongly supported a huge public-works program
and in 1889 he supported legislation to protect women and children.
Leopold was instrumental in promoting numerous additional reforms
within Belgium, including proportional-representation voting and improvements
in education and defense.
In ruling Belgium, Leopold worked hard to deliver public goods to
TABLE 1—POLITICAL SURVIVAL, COALITION SIZE, AND EXCEPTIONALLY
GOOD OR EXCEPTIONALLY POOR POLICY PERFORMANCE
50 PERCENT 25 PERCENT 10 PERCENT
SURVIVAL PROBABLILITY (YEARS) (YEARS) (YEARS)
Exceptional Income Growth and Peace—Large Coalition 4.5 8.0 12.5
Exceptional Black Market Premium—Large Coalition 1.5 4.0 8.0
Exceptional Income Growth and Peace—Small Coalition 5.0 8.5 13.5
Exceptional Black Market Premium—Small Coalition 4.0 10.0 19.0
TABLE 2—POLITICAL SURVIVAL, DEMOCRATICNESS, AND EXCEPTIONALLY
GOOD OR EXCEPTIONALLY POOR POLICY PERFORMANCE
50 PERCENT 25 PERCENT 10 PERCENT
SURVIVAL PROBABLILITY (YEARS) (YEARS) (YEARS)
Exceptional Income Growth and Peace—Democracy 3.5 6.0 9.5
Exceptional Black Market Premium—Democracy 1.5 4.0 7.5
Exceptional Income Growth and Peace—Non-Democracy 5.5 10.0 16.5
Exceptional Black Market Premium—Non-Democracy 4.0 10.0 19.0
Journal of Democracy70
the people. He did so because he depended on the largesse of a popularly
elected government—in other words, on a large winning coalition. In
ruling the Congo during the same period, however, he relied only on
himself and the (quite literally) hired guns who supported his objectives.
The Congo winning coalition was minuscule. The selectorate model suggests
that in the Congo Leopold would have to turn his efforts to generating
private goods for his backers and opportunities for kleptocracy for
himself. He did exactly that.
As liberal and progressive a reformer as he was in Belgium, Leopold
was a monstrously cruel and greedy leader of the Congo. There, through
surrogates, he ruled for the sole purpose of enriching himself and his
key backers. Slave labor first extracted ivory and then the rubber that
fueled his kleptocracy and the corrupt gains of his overseers. Those
Congolese who failed to meet their quotas were beaten, maimed, and
often killed. As many as 10 million may have been killed so that Leopold
and his cronies could become rich.8
While Leopold and his supporters in the Force Publique grew fabulously
wealthy off ill-gotten gains from the Congo Free State, little, if
anything, was invested in improving conditions in that hapless land.
Roads were built only where they facilitated moving rubber to market.
Leopold worried about improving the welfare of his Belgian subjects,
but he ravaged the welfare of his Congolese subjects. Laws protecting
women, children, or workers were unheard of in the Congo Free State.
Virtually the only items exported to the Congo Free State were weapons
for the Force Publique, while vast riches were imported to Belgium. It
was this extraordinary imbalance in trade that eventually led to the discovery
by outsiders that Leopold was growing rich through the use of
slave labor. By 1908, the atrocities reached such a level that they could
no longer be denied. With great reluctance, Leopold ceded his control
of the Congo to the Belgian government.
Who was the real Leopold? Was he the civic-minded king of Belgium
or the murderous ruler of the Congo Free State? We must conclude that
he was the latter. In Belgium, Leopold operated under the institutional
constraint of a large winning coalition. He inherited such a system and
acted in accordance with the incentives created by it. In the Congo, Leopold
was free to choose whatever institutional arrangements he wished.
Finding himself unconstrained, he chose to focus on providing private
goods for a small coalition and vast opportunities for kleptocracy for
himself. In each case, his actions were consistent with the institutional
incentives he faced.
The Importance of the Institutional Setting
Economic growth depends on good economic policies, but good
economic policies are not always in the best interest of political leaders.
Bruce Bueno de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith 71
Market competition is an essential step toward prosperity, but it is not
sufficient. When authoritarian leaders face a choice between promoting
economic growth at the expense of losing office and protecting themselves
in office at the risk of slowing growth, they can be expected to
choose the latter.
Authoritarianism is especially conducive to political success, at least
in the sense of long-lasting leadership. Autocrats sometimes make use
of discretionary resources to encourage growth, but more often they use
those resources for their own benefit. When politicians depend on a few
supporters and the latter’s risks from defecting to a political rival are
high, so too is the incidence of corruption and kleptocracy. Though
neither corruption nor kleptocracy is conducive to economic growth,
both are closely tied to political survival.
Public welfare is enhanced when leaders depend on a large coalition
to keep them in office. Under these conditions, those motivated to stay
in power have no choice but to promote the public’s welfare. To do
otherwise is to risk being ousted from office. Those who depend on a
small coalition, by contrast, are harmed politically if they promote good
public policy. Therefore, exhortations or instructions to such leaders
about policies that promote peace and prosperity can be expected to fall
on deaf ears. Unless the institutional setting within which they operate
changes their incentives, they can be expected to continue engendering
corruption and kleptocracy rather than peace and prosperity. Civicmindedness
is no substitute for correctly configured political institutions
designed to link longevity in office to the promotion of public welfare.
Appendix
The Polity II data and Polity III data provide detailed variables about the institutional
characteristics of almost all states since 1800. These data are routinely updated and
available on the worldwide web. We focus on four institutional variables that provide a
reasonable basis for constructing an index of the size of winning coalitions. A fifth
variable, Legislative Selection (LEGSELEC), appears to be an appropriate indicator of
selectorate size. We discuss the latter first.
LEGSELEC measures the breadth of the selectiveness of the members of each
country’s legislature. This variable is coded as a trichotomy, where 0 means that there is
no legislature; 1 means that the legislature is chosen by heredity, ascription, or simply by
the effective executive; and a code of 2, the highest category, means that members of the
legislature are directly or indirectly selected by popular election. It is evident that the
larger the value of LEGSELEC, the more likely it is that the selectorate is large. We
divide LEGSELEC by its maximum value of two so that it varies between 0 and 1. It
should be evident that in reality the size difference between a selectorate with a score of
0 and a selectorate with a score of 1 is smaller than the size difference between a score of
1 and a score of 2. We are developing alternative indicators, partially based on
enfranchisement rules.
To estimate the size of the winning coalition, we construct a composite index based
on the variables REGTYPE, XRCOMP, XROPEN, and PARCOMP in Polity II. When
REGTYPE (regime type—the only one of these variables not updated in Polity III and
beyond) indicates the regime is not military or military-civilian, we award one point to
our index of coalition size. When XRCOMP (the competitiveness of executive recruitment)
is larger than or equal to a score of 2, another point is assigned to coalition size. An
Journal of Democracy72
XRCOMP score of 1 means that the chief executive was selected by heredity or in rigged,
unopposed elections, suggesting dependence on few people. Scores of 2 and 3 refer to
greater degrees of responsiveness to supporters, indicating a larger winning coalition.
XROPEN (the openness of executive recruitment) contributes an additional point to
coalition size if the executive is recruited in a more open setting than heredity (that is,
the variable’s value is greater than 2). Finally, one more point can be contributed to the
index if PARCOMP (competitiveness of participation) has a score of 5, meaning that
there are relatively stable and enduring political groups which regularly compete for
political influence at the national level” (Polity II, p. 18).
We divide the total value of our index of the winning coalition by its maximum
point score of 4, so that is final minimum value is 0 and its maximum is 1. And again, it
is evident that the progression from 0 to 1 to 2, up to 4 is not linear, as the indicator
suggests.
The correlation between coalition size and selectorate size is 0.46 (n=9,623), while
these two variable have correlations with our democracy scaler (that is, the difference
between the Polity democracy and autocracy scores) of 0.83 (n=11,865) and 0.36
(n=9,185), respectively. The observation unit is a leader-year; some of the data, depending
on the variable, spans the period from 1763–1992, but most variables are observed only
since 1950.
NOTES
The study on which this article is based is supported by National Science Foundation
grant SBER 9709454. A more detailed, technical description of the selectorate theory
advanced here, including formal proof of the claims made here, is found in Bruce Bueno
de Mesquita, James D. Morrow, Randolph Siverson, and Alastair Smith, “Institutions,
Outcomes and the Survival of Leaders,” Hoover Institution Working Paper, 2000.
1. Our measurements of selectorate size and winning coalition size are crude and
primitive. Still, the indicators should be adequate to evaluate whether the central tendencies
of politics are aligned with the expectations that follow from the selectorate
model. See the appendix for a detailed description of our measures.
2. At its peak, the Force Publique was under 20,000 mostly African employees and
400 European officers in a country of tens of millions.
3. See Ted Robert Gurr, “Polity II: Political Structures and Regime Change, 1800–
1986” (Ann Arbor: Inter-university Consortium for Political and Social Research, 1990);
Keith Jaggers and Ted Robert Gurr, “Tracking Democracy’s Third Wave with the Polity
III Data,” Journal of Peace Research 32 (November 1995): 469–82.
4. Our indicator for a complete regime type is a score of 0 to 1 when we subtract the
“autocracy” score in the Polity III data set from the “democracy” score and then normalize
(readjust) the scores so that they range from 0 (complete autocracy) to 1 (complete
democracy). The original scores range from +10 to –10.
5. When we subtract the Polity III data’s autocracy score from its democracy score,
we are left with 176 leaders who have the maximum democratic score of +10 (see the
Appendix). We choose a level of autocracy that yields an equivalent number of leaders
(179) and compare their performance.
6. For the details behind these analyses, see Bueno de Mesquita et al., Staying Alive:
Political Institutions and Leadership Incentives (working title of forthcoming publication).
7. The indicator we use here for a democratic government is a score of 0.60 or higher
on our normalized scale of democracy.
8. Adam Hochschild, King Leopold’s Ghost (Boston: Houghton Mifflin, 1998); and
Peter Forbath The River Congo (New York: Harper and Row, 1977).