Resolving the problem of cycles

Although some political scientists like Martha Crenshaw (1991) and Audrey Cronin (2006) have written about the life cycle and ultimate decline of terrorist groups, the technical analysis of terrorism cycles has been dominated by defence economists. There is a long list of studies, including that of Im, Cauley & Sandler (1987), Enders, Parise and Sandler (1992), Enders & Sandler (1999, 2000, 2002) and Faria (2003), which are all very similar in nature.They all use econometric or statistical techniques that were cutting-edge at the time the studies were undertaken and they all determine that there are indeed cycles in terrorism. For example, Im, Cauley & Sandler (1987) applied spectral analysis, which is a technique for detecting periodicities in data, and found a cycle with a periodicity of 28 months in duration in the time series of all events. There is no doubt about it. Terrorist activity oscillates. Within an overarching cycle, there are oscillations for particular attack types, particular groups, particular types of groups and particular outcomes such as brutality or media attention.The problem lies in explaining why this oscillation is observed.

While we were able to list several plausible explanations at the outset of this chapter, none of these explanations is grounded in the microeconomic theory of group decision-making. Such grounding is the standard in modern economics,13 and certainly in psychology. From a perspective of evolutionary economics and biology, the explanations fail to consider the dual interaction of micro-level decision-making and group dynamics in producing macro-level cycles. In mainstream economics (vis-a-vis defence economics), the ‘proper’ approach that is taught to graduate students is that econometric analysis must formally test hypotheses derived from economic theory (Summers 1991, p. 129). However, as Summers

(1991, p.130) notes, successful econometric research has not been constrained by this approach and many important contributions have been made by more pragmatic applications. In the case of terrorism cycles, the detection of periodicities is an important innovation because it tells us that there is a structure to the multitude of decisions that produce the outcomes we observe. The task of searching for and presenting a theory of terrorism cycles will not be a fruitless one.

We have presented a theory of terrorism cycles, or at least the foundations for a theory, that is based on the micro foundations of terrorist decision-making and the group dynamics that emerge from selection, survival and fitness in a context characterised by a struggle with an external environment. First, we recognise that all of the oscillations that we observe are the result of decisions made in an environment characterised by risk and uncertainty. Without people making decisions, there would be no terrorism. Without risk and uncertainty, there would be no variability over time (and no surprises for either the terrorist group or law enforcement agencies). Second, we noted that ‘share’ rather than ‘absolute’ outcomes is the relevant payoff variable in an evolutionary system. There would be no need for a terrorist group to use resources to maximise absolute outcomes in a context where no other groups or an adversarial government existed. Third, we tried to find a micro foundation for maximising ‘share’ and discovered that evolutionary game theory provided the answer. A terrorist group that attempts to maximise a logarithmic utility function will display the evolutionarily fit characteristics consistent with maximising share. Fourth, we noted that not all groups would be characterised by logarithmic utility and for reasons that include risk aversion, will not engage in the amount or types of terrorist actions consistent with survival.

Putting all of these things together, cycles punctuated by spirals up and down in terrorist activity is a pattern that is to be expected when terrorist groups make decisions in a context where ‘share’ of brutality is the relevant payoff. More terrorism is necessary to maintain or grow a share during an up-spiral while less is necessary during a down-spiral. We also expect terrorist groups to exhibit life cycles, with groups more closely represented by logarithmic utility' maximisation surviving the longest while those that are too risk averse or represented by some other unfit utility function fail to survive. Even groups that make perfectly rational and optimal decisions consistent with the maximisation of their utility functions may not survive if those utility' functions are not fit. In a context populated by these different groups, all competing for share, we will see a pattern of cycles in both terrorist activity' and terrorist group lifespans. Initial surges in terrorism will prompt additional surges, not because of any general contagion-type effect, but because each group as a decision-making unit responds to the implications for the group’s survival in such changing circumstances.

That leaves one last question that we have only vaguely answered. That is, where does an initial surge in terrorism come from? We mentioned that this might come from something as simple as an unexpectedly successful attack. However, we can provide a deeper answer based on Phillips’s (2011) work on terrorist group life cycles. Terrorist groups compete for grassroots supporters. Because those grassroots supporters face a real or psychological switching cost in abandoning the group to which they initially give their support, competition for grassroots support is most intense when the terrorist group first emerges and when supporters are not yet committed. When a new group enters the context, there is intense competition and terrorist activity. This can subside and re-emerge periodically as the contest continues. In a situation where there are already terrorist groups operating in a more or less stable cyclical pattern competing for share, the entry of an entirely new group or the establishment of a splinter group with its need to garner grassroots support through intense action leads to an up-spiral that initiates the processes that we have described.

< Prev   CONTENTS   Source   Next >