public administration and development

Public Admin. Dev. 21, 77±87 (2001)

DOI: 10.1002/pad.157

RISK, RULE AND REASON: LEADERSHIP IN AFRICA

ARTHUR A. GOLDSMITH*
University of Massachusetts, Boston, and Harvard Institute for International Development, Cambridge, MA, USA

SUMMARY
Why are most African countries poorly led? This article looks at the question from the perspective of political economy, con-
sidering leaders to be individuals who respond rationally to incentives created by their environment. Critical is the level of risk,
which can encourage leaders to pursue short-term, economically destructive policies. Using a new database of national leader-
ship transitions in Africa since 1960, the article demonstrates that the risks of of®ce holding are very high in the region. It also
®nds that African countries where leaders face less risk tend to have more open economic regimes and to have lower levels of
perceived political corruption, as predicted by political economy theory. Because it makes the political environment less peri-
lous, democratization in the region holds out some hope for encouraging leaders to govern with an eye to longer-term results.
Copyright # 2001 John Wiley & Sons, Ltd.

INTRODUCTION
Sub-Saharan Africa is poorly led. The region has far too many tyrants and `tropical gangsters', far too few states-
men, let alone merely competent of®ceholders. Too often, these leaders reject sound policy advice and refuse to
take a long and broad view of their job. They persecute suspected political rivals and bleed their economies for
personal bene®t. With a handful of exceptions, notably South Africa under Nobel laureate Nelson Mandela, coun-
tries in the sub-Saharan area are set back by a personalist, neopatrimonial style of national leadership (Aka, 1997).1
Better leadership is not the cure-all for Africa's lack of development, but it would be an important step in the
right direction. A few years back some observers saw hope in a new generation of supposedly benevolent dictators,
such as Isaias Afwerki in Eritrea, Meles Zenawi in Ethiopia, or Yoweri Museveni in Uganda (Madavo and Sarbib,
1997; Connell and Smyth, 1998). Subsequent events (war between Eritrea and Ethiopia, invasion of the Congo
Republic by Uganda) chilled the optimism (McPherson and Goldsmith, 1998; Barkan and Gordon, 1998; Ottaway,
1998). In most countries, it seems progressive leadership soon reverts to the more familiar form of autocratic one-
man rule.2
There is no shortage of macro-level explanations for this pattern. Authoritarian political traditions, lack of national
identity, underdeveloped middle classes and widespread economic distress are among the sweeping, impersonal forces
cited as factors that produce poor leader after poor leader. Foreign aid may have enabled some of these leaders to hang
on longer than they would have otherwise, especially during the Cold War. This article instead takes a micro-level view
of leadership. Without denying that macro-level social and economic factors bear on leaders' behaviour, I ®nd it also
useful to look at these people as individuals and to speculate about the incentives created by their environment.
In the tradition of political economy, we can begin with the assumption that African leaders are usually trying to
do what they think is best for themselves. We can posit that they choose actions that appear to them to produce the
greatest bene®t at least cost, after making allowances for the degree of risk involved. Such a leader also is capable

*Correspondence to: Arthur A. Goldsmith, College of Management, University of Massachusetts Boston, Boston Mass. 02125, USA. E-mail:
[email protected]
1
For simplicity's sake, I shall refer interchangeably to Africa and sub-Saharan Africa in this article. The four Arab states of North Africa are
speci®cally excluded from consideration.
2
In Africa, the national rulers are always men, with only two exceptions in the past 40 years ± Rwanda's Sylvie Kinigi and Liberia's Ruth Perry.
In this article, I will often refer to rulers by using the male pronoun, which is historically accurate but does not imply any endorsement of the lack
of female participation in national leadership roles.

Copyright # 2001 John Wiley & Sons, Ltd.

78 A. A. GOLDSMITH

of learning, and takes cues from what is happening to other leaders in neighbouring countries. He can improve his
behaviour if he has to.
While no African leader fully exempli®es this rational actor model, all these individuals' behaviour can be illu-
minated by it. After all, even the best leaders have mixed, sometimes egoistic motives. To the extent that it repre-
sents reality, the rational actor model also may suggest how changing the political incentive system might induce
African leaders to behave less autocratically.
I start this article by speculating about how these leaders might react to perceived levels of risk in their political
environment. Next, I investigate the actual level of risk, guided by a new inventory that covers every major leader-
ship transition in Africa since 1960. Then, I assess how risk appears to have distorted the way African leaders act in
of®ce. Finally, I consider the ways in which democratization may be changing political incentives for the better.

LEADERSHIP AND INDIVIDUAL MOTIVATION

Perhaps the most troubling thing about African leaders is their tendency to reject (or simply not follow through on)
conventional economic advice (Scott, 1998). Africa is the graveyard of many well-intended reforms. The vacillat-
ing public attitude of Kenya's President Daniel Moi is emblematic. In March 1993, he rejected an International
Monetary Fund (IMF) plan for being cruel and unrealistic. One month later, he reversed himself, and agreed to the
plan. In June 1997, the IMF cut off lending to Kenya after Moi refused to take aggressive steps to combat corrup-
tion. Again, his initial reaction was de®ance, swiftly followed by a more accommodating line.
Why are African leaders apt to resist advice to carry out market-friendly reforms that could boost national rates
of economic growth? If one accepts the premise that, with suf®cient time, open market policies will work in Africa,
such a choice can look senseless. Certainly, no African leader would prefer to perpetuate mass poverty and eco-
nomic stagnation in his country, which can only make governing more dif®cult. More to the point, perhaps, coop-
erating with the international ®nancial institutions is the best way to assure continued diplomatic support and
®nancial credit. Yet, many African leaders apparently see political rationality in choosing policies that are econom-
ically damaging or irrational. Miles Kahler (1990) refers to this as the `orthodox paradox'.
Political economy offers a theory of micro-level behaviour that may explain the paradox. Mancur Olson (1993)
argues that time is the key. According to this theory, the predicament facing any individual national leader is that
the pay-offs to most economic reforms lie in the future, but he also has to hold on to power now. An insecure power
base is likely to encourage either reckless gambling for immediate returns or highly cautious strategies to preserve
political capital; it is unlikely to promote measured actions to obtain long-range returns. Whether a leader acts for
the short or the long term, therefore, is in¯uenced by his sense of the level of threat to his career.
A more technical way to understand a leader's intertemporal choices is to think of a `political discount rate'.
One of political economy's core ideas is that future events have a present value, which one can calculate by using a
rate of discount. That rate of discount rises with risk and uncertainty. When an outcome is doubtful over time, it
makes sense to mark down its present value. The more doubtful the outcome, the more valuable are alternative
activities that yield immediate dividends, even if the expected return of those activities is low. Thus, under con-
ditions of political uncertainty, the narrowly `rational' leader will systematically forgo promising political `invest-
ments' ± ventures whose bene®ts he may not survive to reap. Whenever he is given a choice, according to this
argument, such a leader will usually prefer current political `consumption'. It follows that free-market reforms
look like a poor bargain, requiring immediate political pain in exchange for distant (and therefore questionable)
gain.
High political discount rates are also a possible explanation for the extensive and destructive political corruption
seen in Africa. The Democratic Republic of Congo's Mobutu Sese Sekou is the archetype.3 The late dictator erased
the line between public and private property, accumulating a vast personal fortune and bankrupting his country. His
is an extreme case, yet every national leader has opportunities to pro®t individually from his of®ce. According
to the premises of political economy, it is the leader with the least certainty about his fate who has the strongest

3
I will use the current name to refer to this country, though this is not always historically accurate.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
 RISK, RULE AND REASON 79

incentives to take his rewards now ± and to take as much as possible. A more self-assured leader may calculate that
it is safe to defer most personal ®nancial gain until after he has left of®ce. Some of the misuse of public of®ce also
may be due to the need to buy support from friends and extended family members. Olson (1993), for example,
postulates that leaders with an insecure grip on power have an incentive to take steps to patronize favoured ethnic
groups, often at the expense of national economic health. This sort of pork-barrelling is well known in Africa.
Political economy thus presents a cogent theory for why African rulers act they way they do. Short-term policy
making and political corruption are `rational' ways of trying to manage the risks associated with governing in an
unsettled political system, as we typically ®nd in Africa. According to this thesis, overly cautious or corrupt leaders
may simply be attempting to maximize utility under conditions of personal and political uncertainty. Their assess-
ment of risk is affected by their personal experiences and by their perceptions about larger trends in their country
and region. Unfortunately for the social welfare, their effort to protect their individual interests has spillovers that
hurt everyone else.
The issue for this article is whether the facts support this theory. First, is it true that African leaders face a high
degree of risk? We can reasonably assume these people are tolerant of risk, or they would not have chosen political
careers. Thus, we need to look for evidence of extraordinary occupational hazards for leaders. The second question
is whether political risk in Africa is associated with `bad' (anti-market) economic policy choices or with corrup-
tion. As we will see below, the answers to both questions seem to be af®rmative: there is signi®cant physical risk
for leaders, and that risk correlates with anti-market policies and with corruption. Those two ®ndings, in turn, sug-
gest scope for enhancing the area's national leadership by reducing the risks of governing, a goal that may be
abetted by democratization.

THE DATA SET

To establish a basis for exploring these issues, I compiled a new inventory of all national leaders in independent
African countries from 1960 to 1999. My focus is on the main power holders. That usually means the president, but
in some countries it can be the prime minister or another of®cial. For simplicity, I assume that each country has
only one such leader at a time, which entails judgement calls where competing claimants vie for ultimate national
authority, as for example in The Democratic Republic of Congo before Mobutu consolidated his power in 1964.
Many studies of African leaders have looked at transitions, usually de®ned as a shift from an existing national
leader to a new one, whether by legal or extra-legal means (Hughes and May, 1988; Londregan et al., 1995;
Breytenbach, 1997). Most of these earlier studies have focused on the acquisition of power. In this article, however,
I am more interested in the means by which power is lost. The probability of leaving of®ce and the likely mode of
exit are two factors that plausibly affect African leaders' intertemporal risk estimates, which, in turn, may in¯uence
the way they exercise power.
In taking an inventory of Africa's leaders, I started by enumerating all formal heads of state (or heads of govern-
ment in parliamentary systems) for the 48 sub-Saharan countries, going back to independence (though no earlier
than 1960). To eliminate ®gureheads and deputy personnel from the inventory, I cross-checked these names with
two earlier studies on leadership succession by Henry Bienen and Nicolas van de Walle (1991) and by Rodger
Govea and John Holm (1997).4 This procedure resulted in many re®nements of the list, though unlike that pair
of earlier studies I have chosen to leave in most rulers who presided over interregnum governments ± the exceptions
being leaders who were in of®ce for only a few days during a period of political confusion.
My goal was to single out every principal power holder in Africa over the past 40 years. I came up with 228
leaders (though only 212 individuals, because of 16 people who served non-consecutive terms.) Leaving aside the
48 leaders currently in power, this means Africa as a region has experienced 180 political transitions since 1960.
Due to the subjective element in my de®nitions, this ®gure cannot be considered de®nitive. Another coding system
might differ at the margins. The important thing is that my list contains a suf®ciently large number of transitions to
speculate about the personal and political incentives these changes in power produce for African leaders.

4
The latter study has been published without the details of all the political transitions. The authors generously shared with me the full data set.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
80 A. A. GOLDSMITH

LEADERSHIP TRANSITIONS
Table 1 summarizes how independent Africa's 180 leadership successions took place. By far the most common
means for African leaders to lose power is through a coup d'eÂtat or similar extra-constitutional event. The 1960±
1999 period's ®rst successful coup took place on 13 January 1963 in Togo; the last occurred on Christmas Eve,
1999, in CoÃte d'Ivoire.5 In between these events, another 99 regimes ended with a coup, civil war or invasion.
Three rulers ± Ahmed Abdallah of Comoros, David Dacko of Central African Republic, and Milton Obote of
Uganda ± had the misfortune of being overthrown twice.
The large proportion of coups can be construed as a sign that leaders in Africa typically employ high political
discount rates. If a national leader thinks the odds are good of his being overthrown, which is reasonable given the
region's history, it is also reasonable for him to place a premium on things that he can do now. I will look at the
implications of coups as a mode of leadership transition more fully in the next section. First, however, I want to
highlight an apparent counter trend reported in Table 1: African leaders' long terms of of®ce.
Africa is justly famous for `political dinosaurs', such as Mobutu, who clung to power for 32 years before being
driven into exile in 1997. Fourteen present national heads in the region have been in of®ce for between 10 and 20
years; nine have served more than 20 years. The mean tenure for all former African leaders is 7.2 years, and about
twice that for leaders who died in of®ce or retired. Among current African leaders, the ®gure for average length of
service is even higher.
We can gain a useful perspective by looking at the number of leaders in the economically advanced countries
during the same period. If we take all G-7 nations since 1960, they had an average of one new leader every 3.2
years (counting non-consecutive terms separately) ± twice the tempo in Africa. Expanding the comparison to the
European Union produces exactly the same result: European national leaders served an average of 3.2 years over
the past four decades, with Finland having the shortest average duration and Luxembourg the longest.
To look at the data another way, the average African country has had only between four and ®ve leaders since
1960. The corresponding number for the United States is eight presidents since 1961; for the United Kingdom, it is
eight prime ministers since 1963.6 Note also the infrequency of retirements listed in Table 1. Until the 1990s, only
eight established rulers ever retired from the top of®ce in Africa. It is not surprising, therefore, that nearly one
African leader in 10 died while in of®ce over the past 40 years. The lack of retirements and the large number
of natural deaths are both signs of entrenched leadership.
At ®rst, the relative lack of turnover would seem to contradict the hypothesis that the political environment for
leaders is unstable or insecure. As with American corporate CEOs (Ocasio, 1994), any given political ruler's odds
of retaining power improves with time in Africa (Bienen and van de Walle, 1989). Even elderly leaders in the

Table 1. How leaders leave office in Africa, 1960±1999

Number of incidents Mean time in
1960±69 1970±79 1980±89 1990±99 Total office (year)
Overthrown in coup, war or invasion 27 30 22 22 101 5.7
Die of natural or accidental causes 2 3 4 3 12 11.7
Assassination (not part of coup) 1 0 1 3 5 7.8
Retire 1 2 5 9 17 11.7
Lose election 0 0 1 12 13 14.8
Other (interim or caretaker regime, impeachment) 6 8 4 14 32 1.2
All regime transitions 37 43 37 63 180 7.2
Note: Terms of of®ce are rounded off to the nearest half-year (or third of a year when three leaders held power during the calendar year).

5
Post-colonial Africa's ®rst coup actually occurred in Sudan on 17 November 1958, 13 months before the starting date I use in this article.
See McGowan and Johnson (1984).
6
One British prime minister ± Harold Wilson ± served two non-consecutive terms, so only seven individuals held this top of®ce.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
 RISK, RULE AND REASON 81

Table 2. Leadership turnover in Africa, 1960 (or independence if later) through 1999
Number of Country
transitions
0 Eritrea, Namibia, Zimbabwe
1 Angola, Cameroon, Cape Verde, Djibouti, Equatorial Guinea, Gambia, Kenya, Malawi, Mozambique,
Senegal, Seychelles, Zambia
2 Botswana, Congo Republic, CoÃte d'Ivoire, Guinea, Guinea-Bissau, Mauritius, Tanzania
3 Gabon, Mali, Rwanda, SaÄo Tome & PrõÂncipe, Swaziland, Togo
4 Central African Republic, Ethiopia, Mauritania
5 Burkina Faso, Somalia, Sudan
6 Chad, Lesotho, Niger, South Africa, Uganda
7 Congo, Liberia, Madagascar
8 Ghana
9 Burundi, Comoros, Sierra Leone
10 ±
11 Nigeria
12 BeÂnin
Note: Leaders serving non-consecutive terms are counted twice.

region can hold on against pressures from younger politicians with greater energy and fresher ideas. We might infer
that this type of job security would produce long time horizons, not the short-termism reported for Africa.
Yet, another interpretation is plausible. In the ®eld of ®nance, we know that risk is associated with volatility. The
same may be true in politics. For the individual ruler who wants to weigh his prospects of retaining power, the
average term of of®ce may be a less meaningful statistic than is the variation around the mean. Table 2 gives
one indication of how much volatility exists. Three recently independent countries (Eritrea, Namibia, Zimbabwe)
have had no leadership transitions. Another 11 countries have had just one transition since independence. At the far
end of the spectrum, Nigeria has had 11 transitions and BeÂnin has had 12.
For the high-variance countries like Nigeria, the average term is a poor indicator of what any given leader is
likely to experience. Transitions occur randomly, and, most worrisome, they are often in the form of a coup. Taking
history as their guide, even rulers with proven staying power must acknowledge the possibility that, someday, they
too may ®nd themselves ejected by the army. However, they cannot forecast when that day may come. Those
uncertain prospects would be grounds for safety-®rst political behaviour, and would tend to deter long-term
planning.

PERSONAL HAZARDS
The rational leader also needs to anticipate what may happen to him were he to lose power. Table 3 shows the price
has been high.7 Of the 101 past leaders who left of®ce due to a coup or similar unauthorized event, roughly two-
thirds were killed, imprisoned or banished to a foreign country.
Twenty-seven former rulers died violently, counting ®ve whose deaths appear to have been independent of a
coup or coup attempt. These last would include Mozambique's Samora Machel, killed when his plane was elec-
tronically diverted into a mountainside by South African agents in 1986, and South Africa's Hendrik Verwoerd,
murdered by a lone fanatic in 1966.8 The remaining 22 leaders in this category clearly perished as a direct result of
coups. Among them were three ex-presidents in Ghana who all died during Jerry Rawlings' ®rst takeover in 1979.

7
These data are assembled from several sources, primarily Brockman (1994), Lentz (1994), Rake (1992) and Wiseman (1991). They possibly
understate the level of personal risk for African leaders, since information is not available on all 171 former power holders.
8
Due to the unresolved circumstances surrounding the 1994 plane crash that killed Cyprien Ntariamyra of Burundi and JuveÂnal Habyarimana of
Rwanda, I classify these two deaths as non-coup assassinations. The alternate classi®cation is also plausible.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
82 A. A. GOLDSMITH

Table 3. Fate of Africa's ex-leaders, 1960±1999

All ex-leaders Overthrown Killed Arrested Exiled
Number 159 101 27 37 29
Percentage 100% 64% 17% 23% 18%
Note: African leaders (12) who died in of®ce of natural causes are excluded. Former leaders who served non-consecutive terms (9) are counted
once. Three ex-leaders were overthrown twice.

We can put these executions and assassinations in perspective by looking at the rate of fatal occupational inju-
ries in other lines of work. According to the United States Bureau of Labor Statistics, the three most dangerous
occupations for Americans are commercial ®sherman, logger and small plane operator. Each of these jobs has an
annual death rate of about 100 per 100,000 workers. By contrast, the implied occupational death rate for African
rulers is about 16 times greater.9
Of Africa's overthrown leaders who were not executed or assassinated, 37 were detained and held in jail, or if
lucky, placed under house arrest. Sentences can be stiff. For example, Hamani Diori of Niger passed 13 years
behind bars, Justin AdhmadeÂgbe of BeÂnin spent 9 years, and Jean-Hilaire Aubame of Gabon spent 8 years.
Twenty-nine other ex-leaders were forced into exile, at least temporarily. That ®gure does not include nine ex-
leaders who experienced periods of both imprisonment and banishment. Perhaps the oddest case of the latter is
Jean-BeÂdel Bokassa of Central African Republic. Ousted by his army in 1979, this self-proclaimed emperor ¯ed to
neighbouring CoÃte d'Ivoire. France, his ®rst choice for sanctuary, would not have him. Perhaps hoping for a poli-
tical comeback, Bokassa returned home in 1986, whereupon he was arrested, tried and sentenced to death. For-
tunately for him, the death sentence was commuted to life at hard labour. Bokassa did obtain release in 1993, but he
died 3 years later.
Some rulers may look for reassurance in the region's declining rate of coups. Table 4 reports how many suc-
cessful coups (including victories in civil war and foreign invasions) took place in each of the last four decades,
adjusted for the number of years of national independence in the region. The rate of military takeovers has dropped
steadily, falling from 0.087 per country year in the 1960s, to 0.046 per country year in the 1990s. So, the region-
wide probability of being overthrown is currently about half what it was in the early independence period.10
To the cautious leader, however, the improved odds still may not be good enough. Successful coups are but the
tip of the iceberg. According to the count made by Pat McGowan and Thomas Johnson (1986), for every successful
coup in Africa, there are approximately one other failed coup and two reported military plots to take over power.11
As recently as the 10 years starting in 1980, more than half the countries in the region experienced at least one
`coup event', which includes the unsuccessful incidents (Wang, 1998). Thus, it would be unlikely if many con-
temporary African rulers count themselves safe from being overthrown.

Table 4. Successful coups in Africa, 1960±1999

1960±69 1970±79 1980±89 1990±99 Total
No. of coups 27 30 22 22 101
Coups as % of transitions 73% 68% 59% 35% 56%
Country years 310 408 460 477 1655
Coups per country year 0.087 0.074 0.048 0.046 0.061
Note: Includes regime changes due to civil war and invasion. Country years are total years of national independence during the period.

9
It is interesting to note that, with four assassinations since 1789, the occupational death rate of American presidents is similar. The US
experience is exceptional for developed countries, however.
10
Many efforts have been made to develop models to explain when and why coups occur in Africa. See, for example, Gershoni (1996), Johnson
et al. (1984), and Wells and Pollnac (1988).
11
For the period 1956±1986, they report 60 successful coups, 71 unsuccessful coups and 126 coup plots.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
 RISK, RULE AND REASON 83

It is also true that, like Bokassa, a fair number of jailed or exiled former leaders eventually have bene®ted from
clemency. Some of these individuals were rehabilitated, and they went back into political life. A notable example is
Olusegun Obasanjo, the current president of Nigeria, who won his of®ce after serving time as a political prisoner.
Seven other current heads of African states are former power holders (none appear to be ex-convicts, however).
This phenomenon of a former leader returning to of®ce has happened 16 times in Africa since 1960. Nevertheless,
I doubt the possibilities of being let out of jail, or of re-entering politics after a period in private life, do much to
mitigate the physical dangers of losing power in Africa.
The dangers extend to include potential national leaders within government or from opposition groups. Such
individuals also die or are sent to jail or exile at an alarming rate, according to the tally made by John Wiseman
(1993). As he notes, African political competition is a self-perpetuating, zero-sum game. The stakes are very high
for both actual and would-be powerholders. Yet, none of the participants has much incentive unilaterally to change
the rules. To do so would be to cede the advantage to one's opponents, while exposing oneself to grave personal
jeopardy. The cycle of mutual suspicion and pre-emptive repression thus goes on.

RISK AND LEADERS' BEHAVIOUR

There is little doubt, therefore, that holding high of®ce in Africa poses acute risks. To what extent do those risks
affect leaders' behaviour, speci®cally their behaviour in the areas of economic reform and corruption, mentioned at
the outset of this article? That question is dif®cult to answer fully without detailed case studies of the individuals
involved. In the absence of such information, however, we can look for approximate answers in national indicators
of economic policy and corruption. To the extent we believe that country leaders control public policy or set the
tone for public honesty, aggregate data may give us clues about how these leaders conduct themselves.
To represent a country's commitment to free market economics, I use the Heritage Foundation's Index of Eco-
nomic Freedom (Johnson et al., 1999). The index is calculated by aggregating country scores on 10 policy indi-
cators and measures of the business climate. Depending on their scores, countries are categorized as free (none in
Africa), mostly free, mostly unfree, or repressed. While I do not see eye to eye with the Heritage Foundation on
many subjects, I suspect that these categories offer a good approximation for how fully countries comply with
IMF-style structural adjustment programs. My grouping of countries is based on the average economic freedom
rating for 1995±1999.
I hypothesized earlier in this article that low-risk environments would tend to produce more reform-minded
leaders, or at least leaders who would be more willing go along with economic reform in exchange for ®nancial
credit. Table 5 provides some evidence that such may be the case in Africa. As we see, a correlation exists between
the hazards of leadership and the degree of `economic freedom'. Leaders in the so-called mostly free countries
were the least likely to be overthrown, killed, arrested or exiled. Leaders in the mostly unfree and repressed coun-
tries, by contrast, experienced a greater number of negative outcomes. Low political risk and liberal economic
programmes seem to go together in Africa.

Table 5. Hazards of leadership, by economic policy category

Number of incidents per country (1960±99)
Policy category (1995±2000) Leaders overthrown Leaders killed Leaders arrested Leaders exiled
Mostly free countries 1.0 0.2 0.2 0.5
Mostly unfree countries 2.5 0.6 1.0 0.7
Repressed countries 1.5 0.5 0.6 0.6
Note: The economic policy categories are based on the Heritage Foundation Economic Freedom ranking for 1995±2000. Mostly free countries
are: Botswana, Mauritius, Namibia, South Africa, Swaziland and Uganda. Mostly unfree countries are: BeÂnin, Burkina Faso, Burundi,
Cameroon, Cape Verde, Chad, Congo, CoÃte d'Ivoire, Djibouti, Ethiopia, Gabon, Gambia, Ghana, Guinea, Kenya, Lesotho, Madagascar,
Malawi, Mali, Mauritania, Niger, Nigeria, Senegal, Sierra Leone, Tanzania, Uganda, Zambia, Zimbabwe. Repressed countries are: Angola,
Congo Republic, Equatorial Guinea, Guinea-Bissau, Mozambique, Rwanda, Somalia and Sudan.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
84 A. A. GOLDSMITH

Table 6. Hazards of leadership, by political corruption category

Number of incidents per country (1960±99)
Corruption category (1999) Leaders overthrown Leaders killed Leaders arrested Leaders exiled
Less corrupt countries 0 0.1 0 0
More corrupt countries 2.6 0.8 0.6 0.9
Note: The political corruption categories are based on Transparency International's Corruption Perception index for 1999. Less corrupt countries
are: Botswana, Malawi, Mauritius, Mozambique, Namibia, South Africa, Zambia and Zimbabwe. More corrupt countries are: Cameroon, CoÃte
d'Ivoire, Ghana, Kenya, Nigeria, Senegal, Tanzania and Uganda.

Correlation does not prove causation, especially in making inferences about micro-level behaviour based on
macro-level data. We cannot say whether a safer political environment encourages leaders to opt for the market,
or conversely, whether leaders who opt for the market make their political environment safer (though the latter
possibility seems less likely, at least in the short run). In either case, however, the results are consistent with poli-
tical economy theory.
What is the relationship between political risk and corruption? For a measure of the latter, I use Transparency
International's Corruption Perception Index for 1999. Transparency International is a watchdog organization
formed to help raise ethical standards of government around the world. It compiles an annual index that assesses
the degree to which public of®cials and politicians are believed to accept bribes, take illicit payment in public
procurement, embezzle public funds, and otherwise use public positions for private gains. The index is based
on several international business surveys, using different sampling frames and varying methodologies (Transpar-
ency International, 1999). While Transparency International is careful to point out that the rankings only re¯ect
perceptions about corruption, I ®nd it reasonable to assume that they correspond roughly to reality.
I have conjectured that leaders in the riskier African countries would have the greatest propensity to use their
public of®ces for personal ends. Once more, the data lend support to my hypothesis. In Table 6, I divide African
countries (for which corruption data are available) into two groups: less corrupt and more corrupt. I then tabulate
the hazards of leadership for each group. The pattern is striking. There has never been a successful coup in the less
corrupt group of countries. None of their ex-leaders has been arrested or exiled, and only one was killed while in
of®ce (South Africa's Verwoerd, cited earlier). The more corrupt countries, by contrast, have many coups and
many leaders who suffered personally upon losing power.
As with the economic freedom index, these correlations do not prove that a hazardous political environment
encourages leaders to become corrupt. The opposite is also plausible: corrupt rulers seem likely to invite coups and
to bring personal suffering on themselves. To the extent that risk and corruption are related, the relationship
between the two probably is mutually reinforcing. The important point for this article is that the observed associa-
tion of risk and corruption conforms to what you would expect, based on the assumption of `rational' behaviour
among national rulers. Without overstating the case, the correlation lends support to a political economy account of
poor leadership in Africa.

DEMOCRATIZATION AND IMPROVED LEADERSHIP

Political economy also suggests that one solution to poor leadership is to make the political environment less
hazardous. A safer environment would reduce the incentives to engage in political misbehaviour and, in principle,
encourage more responsible and forward-looking activity. In this context, Africa's recent moves toward more plur-
alistic national political systems, where people can express their political opinions and take part in public deci-
sions, are reasons for hope. It is fashionable ± and correct ± to observe that democracy has shallow roots in most
African countries (Joseph, 1997; van de Walle, 1999). Much of the impetus for reform comes from abroad, from
the region's creditors. Yet, when we observe the patterns of leadership transitions, it is hard to deny that genuine
changes are taking place.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
 RISK, RULE AND REASON 85

No sitting African leader ever lost an election until 1982, when Sir Seewoosagur Ramgoolam of Mauritius was
voted out. Since then, 12 more incumbents have been turned out of of®ce by voters ± accounting for about one-
sixth of the leadership transitions in the 1990s (see Table 1). The threat of losing an election also may account for
the increasing rate of leader retirements ± nine in the 1990s versus only eight in the previous three decades.
Democratization appears to be altering the outcomes of the many coups that still occur. In the past, the new
heads of military juntas often declared themselves permanent leaders (sometimes after dof®ng their uniforms
and becoming `civilians'). Now, it is becoming the norm for coup leaders quickly to organize internationally
acceptable elections ± and, more importantly, to honour the results afterwards (Anene, 1995). Recent examples
include Niger and Guinea-Bissau. This trend is re¯ected in Table 1's `other' category ± political transitions invol-
ving short-term and provisional leaders. The fact we see more transitions of this type in the 1990s is an indirect
re¯ection of the region's growing democratization.
Table 7 reports additional reasons to think that contemporary presidential elections are not simply facËades in
many countries. The entire sub-Saharan region had only 126 elections for top national of®ce in the 30 years
through 1989. Most of those were show elections, with an average winner's share of close to 90%. Conditions have
changed signi®cantly in the 1990s. There were 73 leadership elections during that decade, or more than half as
many as in the three prior decades. All but ®ve of sub-Saharan countries were involved. Equally important, the
winner's share dropped to an average of about two-thirds of the votes cast. Such results would be considered land-
slide victories in the developed world. No president in the history of the United States has ever reached two-thirds
of the popular vote. Still, in African terms, the tendency clearly is toward greater competitiveness at the ballot box.
The classic liberal defences of free and fair elections are that they give voice to majority demands and that they
are a means for recruiting new leadership talent. Political economy and African experience suggest three additional
bene®ts, all associated with reducing the hazards leaders face.
First, elections have the virtue of softening the penalties of losing political of®ce. The defeated candidate in an
election campaign, as opposed to the victim of a coup plot, is far less likely to be executed, jailed or exiled by his
successor. By providing a low-risk avenue of exit, elections thus reduce the stakes in political competition. If the
arguments in this article are correct, that would free African leaders to take a more purposeful, pragmatic view of
their jobs.
A second bene®t occurs if elections become institutionalized, and take place according to a schedule. Countries
that hold regular elections reduce speculation about when (and how) the next political transition is likely. Again,
the probable impact in Africa would be to change the political calculations made by the region's chief power
holders, to allow them to worry less about how to hold onto power and to think more about the long term.
Predictable political transitions might also reduce anxiety among private investors, and thus mitigate the harmful
political business cycle that exists in some countries.
The third bene®t stems from the more rapid turnover among national rulers that results when elections become a
regular part of political experience. As leaders come to see their jobs less as an entitlement and more as a phase in
their careers, that actually may liberate them to `do the right thing', and not always feel forced to do what is

Table 7. Presidential elections in Africa, 1960±1999

Number of elections Number of countries Winner's share of votes cast
1960 (or independence if later) through
1989 126 37 87.9%
1990s 73 43 66.7%
Note: Includes only second round totals in cases where the constitution mandates second round voting to assure majority support for the
president (including the second round for the 1999/2000 presidential election in Guinea-Bissau, even though it occurred outside the period
covered in this table). Total also includes parliamentary elections where president is chosen indirectly or the prime minister is the main power
holder (including cases with hereditary heads of state). Winners' share in these instances refers to the largest party in parliament.
Source: Bratton and van de Walle (1997, 11.197, 208), expanded and updated by the author, principally with data from Nohlen et al. (1999).

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
86 A. A. GOLDSMITH

politically expedient. Merilee Grindle and Francisco Thoumi (1993) have remarked on this phenomenon among
lame-duck presidents in Latin America. Knowledge that their positions are transitory can, somewhat ironically,
concentrate the incumbents' attention on how best to leave a lasting legacy. Similar results are possible in Africa.

CONCLUDING OBSERVATIONS
Before multi-party competition and elections can have these positive effects on leaders, Africa's competitive poli-
tical systems must become institutionalized. This has yet to happen in most countries, according to the results of
Samuel Huntington's (1968) `two-turnover test'. Huntington notes that institutionalized democracies prove them-
selves by repeatedly carrying out peaceful transfers of power through the ballot box. The ®rst time an opposition
leader replaces an incumbent power holder does not necessarily establish a tradition of peaceful political change. It
is only after the new incumbent is defeated and leaves of®ce that one can begin to be con®dent that constitutional
procedures have taken root.
Second turnovers are almost unheard of in Africa. Botswana has not had one. The same party has ruled that
country since independence. Mauritius has had two election-based leadership turnovers, but many observers ques-
tion whether that island nation properly deserves classi®cation in the region. BeÂnin is the only other African coun-
try where incumbent power holders have twice lost elections. The dictator Mathieu KeÂreÂkou fell to NiceÂphore
Soglo in 1991, but he regained the presidency by defeating Soglo in the election 5 years later. KeÂreÂkou's continued
role raises some doubt whether BeÂnin's second transition indicates much other than the persistence of narrow,
personalistic politics in that country.
Nonetheless, the last decade does offer hope that some African societies will be able to establish more orderly
systems of political competition. That could change the incentives for African leaders, and encourage them to act
more responsibly and even-handedly. As a means of redressing decades of oppression and economic stagnation,
that cannot happen soon enough.

ACKNOWLEDGEMENTS

The author would like to acknowledge support from the US Agency for International Development, Equity and
Growth Through Economic Research project.

REFERENCES

Aka PC. 1997. Leadership in African Development. Journal of Third World Studies 14(Fall): 213±242.
Anene JN. 1995. Military coups and redemocratization in sub-Saharan Africa. Journal of Third World Studies 12(1): 181±200.
Barkan JD, Gordon DF. 1998. Democracy in Africa. Foreign Affairs 77(4): 107±111.
Bienen H, van de Walle N. 1989. Time and power in Africa. American Political Science Review 83(1): 19±34.
Bienen H, van de Walle N. 1991. Of Time and Power: Leadership Duration in the Modern World. Stanford University Press: Stanford, CA.
Bratton M, van de Walle N. 1997. Democratic Experiments in Africa. Cambridge University Press: Cambridge, UK.
Breytenbach W. 1997. Democratisation in Sub-Saharan Africa: Transitions, Elections and Prospects for Consolidation (2nd edn). Africa Insti-
tute of South Africa: Pretoria.
Brockman NC. 1994. An African Biographical Dictionary. ABC-CLIO: Santa Barbara, CA.
Connell D, Smyth F. 1998. Africa's new bloc. Foreign Affairs 77(2): 80±94.
Gershoni Y. 1996. The changing pattern of military takeovers in sub-Saharan Africa. Armed Forces and Society 23(2): 235±248.
Govea RM, Holm JD. 1997. Crisis, violence and political succession in Africa. Third World Quarterly 19(1): 129±148.
Grindle M, Thoumi E. 1993. Muddling toward adjustment: the political economy of economic policy change in Ecuador. In Political and Eco-
nomic Interactions in Economic Policy Reform, Bates RH, Krueger AO (eds). Blackwell: Oxford; 123±178.
Hughes A, May R. 1988. The politics of succession in Black Africa. Third World Quarterly 10(1): 1±22.
Huntington SP. 1968. Political Order in Changing Societies. Yale University Press: New Haven, CT.
Johnson B, Holmes KR, Kirkpatrick M. 1999. 2000 Index of Economic Freedom. Heritage Foundation and Wall Street Journal: Washington and
New York.
Johnson TH, Slater RO, McGowan P. 1984. Explaining African military coups d'eÂtat, 1960±1982. American Political Science Review 78(3):
622±640.
Joseph R. 1997. Democratization in Africa after 1989: comparative and theoretical perspectives. Comparative Politics 29(3): 363±383.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)
 RISK, RULE AND REASON 87

Kahler M. 1990. Orthodoxy and its alternatives: explaining approaches to stabilization and adjustment. In Economic Crisis and Policy Choice,
Nelson JM (ed.). Princeton University Press: Princeton, NJ; 33±61.
Lentz HM III. 1994. Heads of States and Governments. McFarland: Jefferson, NC.
Londregan J, Bienen H, van de Walle N. 1995. Ethnicity and leadership succession in Africa. International Studies Quarterly 39(1): 1±25.
Madavo C, Sarbib J-L. 1997. Africa on the move: attracting private capital to a changing continent. SAIS Review 27(2): 111±126.
McGowan P, Johnson TH. 1984. African military coups d'eÂtat: a quantitative historical analysis. Journal of Modern African Studies 22(4):
633±666.
McGowan P, Johnson TH. 1986. Sixty coups in thirty years: further evidence regarding African military coups d'eÂtat. Journal of Modern African
Studies 24(3): 539±546.
McPherson M, Goldsmith AA. 1998. Africa: on the move? SAIS Review 18(2): 153±167.
Nohlen D, Krennerich M, Thibaut B. (eds.) 1999. Elections in Africa: A Data Handbook. Oxford University Press: Oxford.
Ocasio W. 1994. Political dynamics and the circulation of power: CEO succession in U.S. industrial corporations, 1960±1990. Administrative
Science Quarterly 39(2): 285±313.
Olson M. 1993. Dictatorship, democracy, and development. American Political Science Review 87(3): 567±576.
Ottaway M. 1998. Africa's `new leaders': African solution or African problem? Current History 97: 209±213.
Rake A. 1992. Who's Who in Africa. Scarecrow Press: Metchuen, NJ.
Scott G. 1998. Who has failed Africa? IMF measures or the African leadership? Journal of Asian and African Studies, 33(3): 265±274.
Transparency International. 1999. Corruption perception index. http://www.transparency.de
van de Walle N. 1999. Economic reform in a democratizing Africa. Comparative Politics 32(1): 21±41.
Wang TY. 1998. Arms transfers and coups d'eÂtat: a study on sub-Saharan Africa. Journal of Peace Research 35(6): 659±675.
Wells A, Pollnac R. 1988. The coup d'eÂtat in sub-Saharan Africa: changing patterns from 1956±1984. Journal of Political and Military
Sociology 16(1): 43±56.
Wiseman JA. 1991. Political Leaders in Black Africa. Edward Elgar: Brook®eld, VT.
Wiseman JA. 1993. Leadership and personal danger in African politics. Journal of Modern African Studies 31(4): 357±360.

Copyright # 2001 John Wiley & Sons, Ltd. Public Admin. Dev. 21, 77±87 (2001)