Do democracies experience an increased risk of terrorist attacks when an election draws near? Deniz Aksoy presents results from an analysis of terrorist attacks in western European democracies between 1950 and 2004. She writes that while there is evidence that terrorist attacks increase in the lead up to elections, this is only the case in countries where marginalised groups find it difficult to influence the political process, such as in states with high electoral thresholds for representation in parliament.
On 11 April 2015 in Agri’s Diyadin district near the Turkish-Iranian border, a clash between Turkish security forces and the Kurdistan Workers’ Party (PKK), which is defined as a terrorist group by NATO and several other western countries, left five militants dead and four soldiers wounded. It was two months before the upcoming legislative elections in Turkey. Following the event, many commentators and officials accused the PKK of trying to derail the elections.
Deputy Prime Minister Yalçın Akdoğan underlined the connections between the attack and the upcoming elections and argued that the PKK was trying to pressure the electorate. Similarly, Prime Minister Ahmet Davutoğlu noted that an earlier attack by the terrorist group DHKP-C on the Istanbul courthouse that killed a prosecutor on 31 March and the attack in Diyadin were not randomly timed; they both aimed to influence the elections.
Policy-makers in many other democracies often raise similar concerns over the connections between terrorism and elections. For example, following a deadly attack by Basque group ETA only a month before the 2008 Spanish general elections, then Spanish Prime Minister Zapatero said: “The terrorists have today tried to interfere in the peaceful manifestation of the will of the people at the polls”.
The above examples raise an interesting question: Do approaching elections lead to an increase in the volume of terrorist activity in democracies? In a recent study, I illustrate that approaching elections are not always associated with an increase in the volume of terrorist activity. Electoral permissiveness, a central feature of democratic electoral systems, determines the extent to which approaching elections lead to an increase in the volume of violence.
Elections and terrorism in democracies
Even though all democracies provide their citizens with peaceful opportunities to influence politics, some provide marginalised groups easier access to the political system than others. Some democracies have very permissive electoral institutions, which rely on highly proportional representation electoral formulas and low electoral thresholds. High levels of electoral permissiveness make it easier for a wide range of political actors, mainstream or marginal, to enter electoral competition. Thus, permissive electoral institutions can help channel discontented actors’ activities to the peaceful arena of electoral competition.
However when electoral permissiveness is low, it is harder for marginalised discontented actors to meaningfully participate in elections. For example, if the electoral system imposes a high electoral threshold (such as the 10 per cent threshold in Turkey), a small political party formed by marginal groups does not have much chance of gaining legislative representation. Consequently, while approaching elections and increasing competition motivate discontented actors to increase their activities, obstacles they face to enter electoral competition pushes these actors to violent means.
Therefore in democracies with low electoral permissiveness, increased competition around elections combined with low levels of electoral permissiveness increases incentives to use violent tactics. Accordingly, approaching elections lead to an increase in the volume of terrorist attacks only in democracies with low levels of electoral permissiveness.
Electoral permissiveness and terrorist attacks
Using data from Western European democracies between 1950 and 2004 and statistical models that include country fixed effects, I show that approaching elections are linked with an increase the volume of attacks in democracies with the least permissive electoral systems, but not in countries with permissive electoral systems.
In Figure 1, I illustrate the relationship between the number of months to an election, the effective threshold, which is a measure of electoral permissiveness, and the predicted number of terrorist attacks. A move from 60 to 0 on the x-axis indicates that elections are approaching. A move from 10 to 30 on the y-axis indicates an increase in effective threshold, thus a decrease in electoral permissiveness. Figure 1 shows that when the threshold is at low levels, approaching elections are not associated with an increase in the expected number of attacks. However, at high levels of threshold, approaching elections are positively associated with an increase in the volume of attacks.
Figure 1: Relationship between terrorist attacks, time to election, and electoral permissiveness
Note: For more information see the author’s article in The Journal of Politics.
Figure 2 illustrates the relationship between approaching elections and the predicted number of attacks in three different scenarios: when threshold is at its minimum value (.7), maximum value (35), and set to 31. The vertical dashed line indicates when the differences in predicted number of attacks across the three scenarios are statistically significant. Whenever there are less than 28 months to elections, the differences in predicted number of attacks across the three scenarios are statistically significant. Negative slopes of the black dashed lines suggest that when threshold is high, decreasing the number of months to elections is associated with an increase in the number of attacks.
Figure 2: Relationship between upcoming elections and predicted number of terrorist attacks in three different scenarios
Note: For more information see the author’s article in The Journal of Politics.
Accordingly, in the least permissive systems, approaching elections are likely to be associated with an outburst of attacks. When the effective threshold is 35, the predicted number of attacks 6 months before elections is 4.05 and it increases to 4.39 in the month of elections, holding all the other variables at their mean values. This is around an 8.3 per cent increase in the volume of attacks. However, if the effective threshold is 31 then the predicted number of attacks 6 months before elections is 2.99 and it increases to 3.17 in the month of elections. This is around a 6 per cent increase in the volume of attacks.
Thus, as the threshold increases the influence of approaching elections becomes more dramatic. This is evident from the slope of the short-dashed line being steeper than the slope of the long-dashed line. The slope of the solid line is positive, even though it is not steep. This suggests that in countries with very low thresholds, approaching elections lead to a slight decrease in the predicted number of attacks. However, based on the calculations of the marginal effect of approaching elections, this effect is not statistically significant. In fact, only when the effective threshold is above 30 are approaching elections significantly related to an increase in the volume of attacks.
Overall, this empirical analysis shows that approaching elections are not consistently associated with an increase in the volume of domestic terrorist attacks. Among the democratic countries studied, only in those with the least permissive electoral institutions are approaching elections associated with an increase in the volume of attacks.
Please read our comments policy before commenting.
Note: This article gives the views of the author, and not the position of EUROPP – European Politics and Policy, nor of the London School of Economics.
Shortened URL for this post: http://bit.ly/1ecfpZp
Deniz Aksoy – Princeton University
Deniz Aksoy is an Associate Research Scholar in the Department of Politics at Princeton University.
Very nice article. I am analysing the same issue in the Indian states where the insurgency is in place. Can you please share your empirical data.