Elections are supposed to provide both accountability and legitimacy. Yet recent African elections have ranged from the grotesque (Zimbabwe), through the dirty (Nigeria and Kenya), to a clean change of regime (Ghana). Does the conduct of elections matter and, if so, what is going on?
Kudamatsu (2006) measures government performance by infant mortality and shows that, in Africa, elections produce no improvement except in the (rare) instances in which the incumbent is defeated. Chauvet and Collier (2009) measure government performance in developing countries by the World Bank’s economic policy rating, the Country Policy and Institutional Assessment, and by the commercial rating, the International Country Risk Guide. They find that elections that are properly conducted lead to improvement in economic policies, but that elections that are badly conducted have at best no effect.
African elections are rife with illicit tactics. The key illicit tactics are vote buying, voter intimidation, and ballot fraud. Vote buying has been shown to be effective by Vicente (2007), through a field experiment conducted in Sao Tome and Principe. Our recent study was more ambitious (Collier and Vicente 2009). Its main focus was voter intimidation.
A field experiment
To the best of our knowledge, our project is the first systematic empirical account of how voters react to intimidation. We partnered with ActionAid International to conduct a field experiment during the April 2007 elections in Nigeria. This prominent NGO, specialising in community improvement and capacity building, ran a randomised campaign against electoral violence in six states of Nigeria (Delta, Kaduna, Lagos, Oyo, Plateau, and Rivers), covering the main socio-economic regions of the country. The campaign held town meetings and popular theatres and distributed informational materials (see our fieldwork website for full details of the fieldwork). The objective was to increase citizen confidence in local security so that they would vote and punish violent politicians at the polls (see poster below).
We devoted most of our effort to measurement. We collected information from all locations covered by ActionAid and a strictly comparable set of control locations (this task was made possible by the agreed randomisation of ActionAid’s intervention). We ran representative household surveys, both using a panel (interviewed before the campaign and after the elections) and a post-election oversample. And we contracted local journalists in each observed location to keep diaries of local violent events.
We found that all three of the illicit electoral tactics were common and used strategically. Fraud and vote buying were associated with the incumbent, while intimidation was used by the weakest challengers. We interpret these patterns in the context of a model where the incumbent is the dominant national player. In constituencies where the incumbent faces strong opposition, he resorts to the tactics of fraud and vote buying, which are less damaging to incumbent interests than violence. Where opposition is weaker the incumbent can afford not to use illicit tactics. But in some of these situations – where opponents are very marginal – it is the opponents who resort to intimidation. Analogous to terrorism, violence in these situations is a last-resort strategy of the marginalised.
Our experimental results point to a high degree of effectiveness of ActionAid’s campaign in terms of increasing confidence in security in treated locations. Perceptions of violence originated by politicians and implemented by violent gangs decreased by 5% to 12%. Various measures of empowerment, including survey questions and directly-measured behaviour of respondents after the campaign, also point to an increased sense that “something could be done to counteract electoral violence”. Consistent with these chances in perceptions, the diaries of violent events also reveal a consistent statistically significant reduction in actual violence.
The campaign increased voter turnout by 10%, implying that, where politicians threaten violence, fear of intimidation has large effects on turnout. We are also able to show that non-incumbent groups were blamed by voters for the violence. In the presidential race, AC’s candidate, former vice president Atiku Abubakar, suffered a significant loss of support as a consequence of ActionAid’s campaign. At the gubernatorial level, the incumbent was the main beneficiary from the intervention. Both findings point to the violence being mainly attributable to weak opposition groups. We show that there were two distinct changes in voting resulting from the campaign. People who prior to the campaign had intended to vote for violent candidates became more likely to abstain, and supporters of non-violent candidates who had intended to stay away from the polls became more likely to vote. The campaign was thus successful both in increasing a sense of safety and in changing opinions.
Finally, we find that the campaign was especially effective with those people who were less locally integrated (those not owning land, with smaller households, and working outside the district). These voters are less likely to be embedded in local client-patron political deals. Hence, the strategic use of violence may be primarily targeting these least committed voters.
Our perception results are robust to conformity of the treated subjects to the message of the campaign. We encountered clear heterogeneity of effects across different variables within the same type of measurement, and clear homogeneity of results across different types of measurement within the same variable. In addition, an oversample of treated locations, not directly treated by campaigners (therefore arguably less prone to conformity), reports very similar observations when compared to the directly-treated respondents.
We are now ready to return to the limitations of elections in many African countries. Nigeria, the most populous of them, features a powerful incumbent party (nurtured by oil revenues) and virtually no restraints on illicit electoral conduct. With the incumbent empowered by vote buying (financed by oil revenues) and ballot fraud (enabled by patronage within the bureaucracy), violence was the only feasible strategy left for the very weak. In Kenya, another powerful incumbent faced a tight election. Far from the election conferring legitimacy on the winner, the perception of ballot fraud was sufficient to provoke opposition voters to post-election violence. In Zimbabwe, the incumbent lacked both the finances to win through bribery and the necessary local bureaucratic control to win through ballot fraud. Having lost the first round, Mugabe therefore resorted to violence as the only way to retain power. Such violence is, however, a last-resort strategy for incumbents because it comes at such a high reputational cost.
The conclusion is that in the absence of effective restraints, elections are unlikely to achieve legitimacy and accountability. The international community already monitors the conduct of elections, but these assessments now need to be linked to credible and powerful incentives that make illicit tactics unattractive. Collier (2009) suggests one way of making this linkage.
Chauvet, Lisa and Paul Collier (2009), “Elections and Economic Policy in Developing Countries”, Economic Policy, forthcoming;
Collier, Paul, (2009), Wars, Guns and Votes: Democracy in Dangerous Places, HarperCollins, New York.
Collier, Paul and Pedro C. Vicente (2009), “Votes and Violence: Evidence from a Field Experiment in Nigeria”.
Kudamatsu, Masayuki (2006), “Has Democratization Reduced Infant Mortality in Sub-Saharan Africa? Evidence from Micro Data”, University of Stockholm, IIES, Working Paper;
Vicente, Pedro C. (2007), “Is Vote Buying Effective? Evidence from a Field Experiment in West Africa”, Oxford University and BREAD, Working Paper.