An Analysis of the Effect of Ideology on the Life Expectancy of Terrorist Organisations

Acknowledgments

I would like to thank my thesis supervisors Professor John Mo and Dr. Balaram

Das for their encouragement and guidance. My sincere thanks go to Dr. Yongshun

Xiao for his valuable help and advice.

Abstract

This thesis investigates the effect of ideology on the survival times of 557 terrorist organisations. While there is broad acceptance of the view that the majority of terrorist organisations are short-lived, quantitative studies of the life spans of terrorist organisations are rare. Terrorist organisations go to great lengths to ensure their survival, with many terrorist organisations viewing their continued existence as a victory in itself. Within the broader field of organisational analysis, an organisation’s age is viewed as a key variable in explaining the development of its structures, functions, and, most importantly, its likelihood of continued survival.

The ideology through which a terrorist organisation views the world is thought by many scholars of terrorism to play an important role in its justification for violence, its strategic intentions, the targets it selects, and its survival prospects. While many typologies of terrorist ideologies have been proposed, one of the most common approaches uses four broad categories of ideological orientation: Separatist/nationalist, left-wing, religious, and right-wing. Separatist/nationalist organisations are widely believed to have much greater longevity than organisations that lack a distinct, territorially bounded, ethno-nationalist base.

The non-parametric and semi-parametric statistical analyses of the life spans of 557 terrorist groups presented in this thesis are complemented by a description of various stage-based life cycle models. The concept of an organisational life cycle is used to highlight and explain some of the frequently observed patterns and behaviours of terrorist organisations as they form, grow, peak, fragment, and fail. This provides the contextual background to the quantitative analysis. This thesis emphasises a longitudinal approach to understanding terrorist organisations. The concepts presented are intended to convey an understanding of how the simple

passing of time can radically alter a terrorist organisation’s power structure, operating environment, range of targets, and rationale for using violence.

The data analysed in this thesis were gathered from authoritative online databases, reference books, and peer-reviewed journals. A list of 557 terrorist organisations that had begun their campaigns of violence between the years 1960 and 2003 (inclusive) was compiled. Details of the organisation’s ideology, the year it first claimed responsibility for a terrorist attack, and the year it last claimed responsibility for an attack, were recorded. Organisations that had continued to engage in violence beyond the year 2003 had their life spans censored at 2003, the study’s cut-off date. For the purposes of this study, the length of time, in years, separating a terrorist organisation’s first and last claimed attacks is considered to be its life span.

Non-parametric Kaplan-Meier analysis showed that 53% of the organisations in the sample did not have life spans greater than one year. Kaplan-Meier analysis also showed that of those organisations that survived beyond their first year, approximately 50% had failed by their tenth, a figure that is close to that suggested by earlier research. Semi-parametric Cox proportional hazards regression was used to evaluate the effect of ideological orientation on the life spans of terrorist organisations. Overall, there was a significant association between ideological F (4, N = orientation and survival time (in years) among terrorist organisations, 2

557) = 42.8, p < 0.0001. While the data’s limitations made definitive conclusions difficult, the widely-held views among scholars of terrorism of the longer life expectancy of separatist/nationalist organisations, and the high rates of failure among new terrorist organisations, were supported by the results presented in this thesis.

Table of Contents

1.  INTRODUCTION  1

2.  LITERATURE REVIEW  13

2.1  The Organisational Life Cycle  13

2.2  The Organisational Life Span  18

3.  TERRORISM THROUGH AN OPEN SYSTEMS FRAMEWORK  22

3.1  The Organisation as a Machine  22

3.2  The Organisation as an Organism  24

3.3  General Systems Theory: Organisations as Open Systems  26

3.4  Terrorist Organisations as Open Systems  30

3.5  Open Systems and Organisational Longevity  33

4.  TERRORISM AS AN EXTENSION OF POLITICS  36

4.1  Terrorist Organisations as Specialised Political Parties  36

4.2  An Ideological Taxonomy of Terrorist Organisations  39

4.2.1  Separatist/nationalist Terrorist Organisations  40

4.2.2  Left-wing Terrorist Organisations  41

4.2.3  Religious Terrorist Organisations  42

4.2.4  Right-wing Terrorist Organisations  44

4.3  A Structural Taxonomy of Terrorist Organisations  46

4.3.1  Hierarchical Terrorist Organisations  46

4.3.2  Flattening the Hierarchy: Terrorist Networks  50

4.3.3  Hybrids: The Real World of Organisational Structures  54

5.  THE LIFE CYCLES OF TERRORIST ORGANISATIONS  58

5.1  Organisational Life Cycles and the Biological Metaphor  58

5.2  Empirical Evidence for Organisational Life Cycles  60

5.3  A Generic Five-Stage Model of an Organisational Life Cycle  62

5.3.1  Start-up  62

5.3.2  Growth  62

5.3.3  Consolidation  62

5.3.4  Diversification  63

5.3.5  Decline  63

5.4  Life Cycle Models of Terrorist Organisations  64

5.4.1  Formation  66

5.4.2  Growth  68

5.4.3  Adapting  69

5.4.4  Decline  72

5.5  Hawks and Doves: An Abridged Life Cycle Model  74

5.6  Organisational Size  76

5.6.1  Organisational Size and Growth  77

5.6.2  Separating the Effects of Organisational Age and Size  79

5.6.3  Size and Growth in Terrorist Organisations  80

5.7  Connecting Qualitative Life Cycles to Quantitative Life Spans  82

6.  MEASURING THE LIFE SPANS OF TERRORIST ORGANISATIONS  86

6.1  Methods for Survival Analysis  86

6.1.1  The Survival Function  86

6.1.2  The Hazard Function  87

6.2  Research Design  91

6.3  Non-parametric Analysis of Survival: The Kaplan-Meier Estimate  95

6.4  Semi-Parametric Analysis: Proportional Hazards Regression  100

6.5  Model Fitting Using Cox Regression  102

6.5.1  Cox-Snell Residual Analysis  108

6.5.2  Score Residuals  110

6.5.3  Deviance Residuals  112

6.5.4  Schoenfeld Residuals  113

6.6  Data Limitations  116

7.  FUTURE RESEARCH  120

8.  CONCLUSION  122

9.  REFERENCES  124

APPENDIX A:  DISTRIBUTION OF IDEOLOGY  135

APPENDIX B:  DISTRIBUTION OF FIRST ATTACKS  136

1. Introduction

While the life spans of well-known terrorist organisations such as the Provisional Irish Republican Army (PIRA), the Revolutionary Armed Forces of Colombia (FARC), Shining Path, Euskadi Ta Askatasuna (ETA), and Hezbollah can be measured in decades, the majority of terrorist organisations do not survive for more than a year. Al Qaeda’s current diffuse form is often cited as evidence in support of the view that modern terrorist organisations are complex adaptive systems. Yet the high failure rates of newly-formed terrorist organisations indicate that many of them are neither particularly complex nor adaptive. The vast majority of terrorist organisations are small, short-lived, and fail to achieve their stated strategic objectives.

Quantitative studies of the life spans of terrorist organisations are rare. This is a reflection of the difficulty in obtaining reliable longitudinal data on a large number of clandestine organisations, not a dismissal of the importance of studying the factors that influence mortality among terrorist organisations. One of the most widely-cited statistics concerning the life spans of terrorist organisations is David Rapoport’s (1992) suggestion that perhaps as many as 90% of terrorist organisations do not survive beyond their first year of operation. According to Rapoport (1992), of the few terrorist organisations that survived their first year of operation, fewer than 50% would survive until the age of 10. These figures, while approximate, convey the high failure rates experienced by terrorist organisations as they struggle to establish themselves. The high failure rates observed among terrorist organisations parallel the high failure rates seen among new business firms and voluntary organisations.

Many factors influence the life expectancy of a terrorist organisation. The availability of a large pool of potential recruits, external financial support, and safe

1

havens in nearby countries can greatly enhance the survival prospects of a terrorist organisation. The ideology through which a terrorist organisation views the world is also thought to have an important influence on its life span (Cronin, 2006). Gurr (1979), Wilkinson (1979), Crenshaw (1991), and Rapoport (1992) drew attention to the longer life expectancies of terrorist organisations that use separatist/nationalist appeals to mobilise support and justify their violence. The distinct ethnonationalist base available to a separatist/nationalist organisation confers considerable benefits in terms of its long-term viability.

The effect of ideology on a terrorist organisation’s life expectancy can be felt through the types of targets it attacks. The wrong choice of target can be fatal to a terrorist organisation (Bell, 1998). Egregious and indiscriminate attacks can result in the rapid withdrawal of crucial resources and support from the organisation’s repulsed constituents. Mistakes and disagreements over whom or what to target can cause irreconcilable internal schisms within terrorist organisations. Left-wing terrorist organisations are believed to exercise more care in the selection of targets in order to avoid harming members of the proletariat they claim to represent (Karber & Mengel, 1983; Fine, 2008). The disinclination to indiscriminate attacks among many left-wing terrorist organisations might enhance their survival prospects, as indiscriminate attacks often result in a loss of support from sympathisers and moderate members. While analysing the characteristics of a non-random sample of 77 terrorist organisations, Crenshaw (1991) noticed that leftwing and separatist/nationalist organisations appeared to be similarly wellrepresented among the 47 organisations that had survived for at least 10 years. Terrorist organisations mobilised around a religious ideology are believed to be much less discriminating in their targeting than secular organisations (Fine, 2008), a factor which, prima facie, has the potential to lead to rapid losses of support among their constituents and, as a result, higher rates of organisational failure.

2

A terrorist organisation’s political goals and ideological imperatives are more likely to play a dominant role as the drivers of violence in the early stages of its life cycle. As the organisation ages, however, intra-organisational dynamics frequently displace ideology as the primary driver of violence (Oots, 1989). This implies that steps taken to ameliorate the grievance that provided the original stimulus for the formation of a terrorist organisation might actually accomplish very little if enough intra-organisational momentum has already been gained (Creshaw, 1985). This highlights the importance of understanding terrorist organisations from a longitudinal perspective. Strategies and tactics that show great promise when applied to newer terrorist organisations can fail dismally when applied to organisations in the middle or later stages of their life cycles. Sageman (2004) pointed out that intra-group dynamics can dominate the decision-making of contemporary jihadi cells from their very inception, especially if a terrorist cell’s formation was the result of a group of friends that gradually became radicalised. Wilson’s (1973) influential theory that survival was the overriding goal of any political organisation (and this includes terrorist organisations) can be contrasted against the actions of the 2005 London suicide bombers, who did not plan for their group to survive the bombings they carried out.

While there is no universally accepted life cycle model for terrorist organisations, descriptions of the stages through which many terrorist organisations pass as they age have been provided by Karber and Mengel (1983), Long (1990), Bell (1998), Sprinzak (1998), and Roberts (2003). Several models of terrorist organisational life cycles have quite similar stages to the four and five-stage life cycle models that emerged from studies of legal commercial organisations. Karber and Mengel’s (1983) and Roberts’ (2003) life cycle models emphasised the important behavioural thresholds terrorist organisations often cross when they realise that their tactical success has not translated into strategic success. The large variation in circumstances, goals, and structures among terrorist organisations mean that a

3

single life cycle model would be hard-pressed to cover them all. Sprinzak (1998) developed a life cycle model that described three distinct stages that many leftwing terrorist organisations pass through. The tendency for terrorist organisations to become less discriminating in their attacks as they age, to polarise along moderate and extreme lines, and to replace ideology with cohesion as the dominant driver of their violence emerge as leitmotifs from the various life cycle models discussed in this thesis.

The circumstances surrounding the formation of a terrorist group can vary widely. Terrorist organisations can form as a result of an increasingly radical faction hiving off from a larger, non-violent protest organisation. They can also form as a result of a merger between two or more organisations. Large terrorist organisations in their death throes can produce many smaller, distinct splinter groups, differentiated from each other on the basis of religion, geographic location, goals, and leadership rivalries. The sudden appearance of several radical splinter groups can complicate the environment for counter-terrorist planners and analysts. More importantly, it can lead to an increase in overall violence, as the smaller organisations ruthlessly compete with each other for recruits, publicity, and influence (de Mesquita, 2006). Terrorist organisations can end in as many ways as they begin. Some destroy themselves from within, some are eliminated by the authorities, while others decide to abandon terrorism as a strategy of political change. Some, like the African National Congress (ANC), succeed in achieving their goals and have no further use for terrorist tactics. Ross and Gurr (1989) pointed out that some terrorist organisations fail simply because of the members’ waning commitment to the organisation and its goal.

Agreement among scholars and governments on a definition of what constitutes terrorism has been, and is expected to remain, an elusive goal. While discussing Schmid and Yongman’s (1988) influential review of terrorism research, Silke (2004)

4

pointed out that a diligent researcher could easily uncover twice as many definitions as the 109 Schmid and Yongman (1988) found. Silke (2004) felt that perennial definitional debates, while important, often diverted researchers from much more productive lines of enquiry. The definition of terrorism used in this thesis is based on the definition provided by the Memorial Institute for the Prevention of Terrorism (MIPT). The MIPT defines terrorism as, “Violence, or the threat of violence, calculated to create an atmosphere of fear and alarm. These acts are designed to coerce others into actions they would not otherwise undertake, or refrain from actions they desired to take. All terrorist acts are crimes. Many would also be violation of the rules of war if a state of war existed. This violence or threat of violence is generally directed against civilian targets. The motives of all terrorists are political and terrorist actions are generally carried out in a way that will achieve maximum publicity. Unlike other criminal acts, terrorists often claim credit for their acts. Finally, terrorist acts are intended to produce effects beyond the immediate physical damage of the cause, having long-term psychological repercussions on a particular target audience. The fear created by terrorists may be intended to cause people to exaggerate the strengths of the terrorist and the importance of the cause, to provoke governmental overreaction, to discourage dissent, or simply to intimidate and thereby enforce compliance with their demands” (MIPT, 2007).

Like many aspects of research into terrorism, agreement over what constitutes a terrorist ‘organisation’ is not easy to reach and presents challenges on several fronts. There is a question of perspective. It would be rare to find an organisation that believed it deserved the label of ‘terrorists’. Those who use terrorism believe their violence is justified by their ideology and are acting on behalf of a larger, aggrieved or oppressed group of people. Many feel they are actually acting out of a sense of altruism towards society. The importance of perspective is neatly summarised by the aphorism, “One person’s terrorist is another’s freedom

5

fighter”. A terrorist organisation can use several aliases, leading to distorted counts and a confusing picture of the number of organisations active in a particular area or epoch. An organisation might change its name several times, while remaining relatively unchanged otherwise. Umbrella organisations often subsume, and are sometimes consumed by, a morass of competing or cooperating factions with fluctuating memberships, shared members, ideological niches, and shifting loyalties. A large number of organisations without a clear title or name are missed by analyses that require, first and foremost, a unique identifier in order to efficiently label, list, and catalogue them. In a discussion of system identity, Weinberg (1975) pointed out that existence implies an identity, and an identity implies the existence of an identifier. The challenges in cataloguing, counting, and measuring terrorist organisations with widely varying structures, resources, goals, environments, and opponents have been discussed by Mickolus (1977), Gurr (1979), Oots (1986), Crenshaw (1991), and Chasdi (2002). This thesis subscribes to the definition of a terrorist organisation given by the MIPT as, “A collection of individuals belonging to an autonomous non-state or sub-national revolutionary or anti-governmental movement who are dedicated to the use of violence to achieve their objectives. Such an entity is seen as having at least some structural and command and control apparatus that, no matter how loose or flexible, nonetheless provides an overall organisational framework and general strategic direction. This definition is meant to include contemporary religion-motivated and apocalyptic groups and other movements that seek theological justification or divine sanction for their acts of violence” (MIPT, 2007).

One of the most important tasks facing a researcher embarking on an empirical study of terrorist organisations consists of deciding which organisations to include in the study (Crenshaw, 1991). The organisations analysed in Chapter Six are drawn from open source materials such as the Encyclopedia of Terrorism (Kushner, 2003), the US Department of State’s Patterns of Global Terrorism (1976-

6

2003), MIPT, and peer-reviewed academic journals. The statistical analyses presented in Chapter Six are inferential, meaning that the results are intended to be generalised to the population from which the sample was drawn. Using archival data is a common way to study the members of a population of interest, although these studies can suffer from biases that skew the results and attenuate their generalisability. A realistic view of how well the sampling methodology covers the phenomenon under investigation needs to be communicated by the analyst. Gurr (1979), Oots (1986), Crenshaw (1991), and Merari (1991) outlined in clear and unambiguous terms the limitations of the data that quantitative researchers often used.

Identifying populations of organisations has been a significant methodological challenge for scholars of organisational dynamics (Hannan & Freeman, 1987). One of the most important issues to reflect upon when collecting and analysing archival data concerns the sample’s coverage. Thought needs to be given to the characteristics of a terrorist organisation which make it more, or less, likely to be included in the sample. Larger and older organisations are more likely to be found in this sample, as are organisations that, although small and new, carried out particularly egregious attacks. Gurr (1979) believed that larger and more publicityconscious organisations were over-represented in the data sources used by many analysts. Terrorist organisations that claimed responsibility for attacks on US and Western citizens or property are also widely-believed to be over-represented in archival records. A US and Western-centric bias cannot be taken for granted, however, as Weinberg and Eubank (2004) found a bias in the opposite direction with data drawn from similar sources. Developing countries with obdurate insurgencies frequently lack well-organised and transparent statistical reporting mechanisms. Data gathered under such circumstances often suffer from uneven coverage and varying quality. Stohl (1988) pointed out that most of the world’s terrorist attacks go unreported by the Western media. A terrorist attack within a

7

country whose government enforces strict media control is less likely to find its way into a database of terrorist attacks than an attack which happened in a country with a free press.

The true biases in the sample can only be evinced by testing the same hypotheses with alternative data sets. The sample analysed in this thesis would be most useful for those interested in larger or older organisations, although the fact that more than half of the 557 organisations in the sample had life spans of less than a year shows a serious attempt to include smaller and short-lived organisations in the analysis. Further data gathering efforts would be likely to net large numbers of obscure, short-lived, small organisations. Ignoring small organisations has its drawbacks, however. Although right-wing organisations are under-represented in databases, Crenshaw (1991) pointed out that the cumulative violence and impact of many small and fissiparous right-wing groups was as consequential as that of the more unified left.

Unfortunately, the simple lack of an appellation disqualifies an organisation from studies of organisational mortality. Studies of when, in a nascent or inchoate terrorist group’s life cycle, they assign themselves a name, and why, are few. A change of name can signify a change in intent or be one of many changes in a wider program of organisational renewal. Terrorist organisations can use several aliases. The operational cells of larger terrorist organisations sometimes adopt a new name with every attack to give the impression of a broader movement, confuse authorities, or simply to honour fallen comrades and martyrs (Mickolus, 1977). Clearly, identity does not reside entirely in an organisation’s name. Large changes within an organisation can take place while its name remains the same. Conversely, frequent name changes might take place within an otherwise stable organisation. Crenshaw (2007) pointed out that the Red Army Faction in Germany was not a monolithic entity - it was more like a succession of different groups

8

using the same name. Despite these limitations, there are few, if any, ways of identifying and cataloguing terrorist organisations that are as effective, efficient, natural, and uncontroversial as the names these organisations give themselves.

The majority of research on terrorism is qualitative. Case studies that focus on a single organisation can communicate a depth and richness of analysis that quantitative studies are unable to provide. The sheer variety of terrorism’s circumstances, organisational forms, and causes and effects, combined with imprecise and ambiguous data, has meant that attempts at sophisticated quantitative analysis have often fallen far short of the rigor and predictive power they presaged (Crenshaw, 1991; Merari, 1991). The heterogeneity of terrorism’s actors, methods, and environments can be a serious impediment when generalising any statistical patterns the analyst uncovers to a wider population. Crenshaw (1992) warned researchers that studies which combined a large number or wide range of terrorist organisations operating in different milieus into one single analysis risked conflicting and ambiguous results. A more selective approach, which involves partitioning data along particular types of activity, location, and epoch can ameliorate some of the difficulties presented by the heterogeneity among terrorist organisations. Moving from the general to the specific can rapidly whittle down a data set to a handful of observations, however. Ideally, the sampling method and data analysis methods chosen by the analyst will represent a judicial compromise between what Boulding (1956) considered the general that lacked useful content, and the specific that lacked broader applicability. Somewhere between the two poles the analyst must find, for their purpose, resources, data, and level of abstraction, the most effective level of analysis.

Statistical approaches to the study of terrorism assume that the behaviours of terrorist organisations might have some underlying patterns and that the discovery

9

of these patterns through even the simplest statistical procedures can be helpful (Mickolus, 2002). Any statistical analysis setting out to explore the possible association between ideological orientation and organisational longevity, for example, is based on the assumption that there are aspects of terrorism that are not completely random. Obviously, the usefulness of any particular approach depends greatly on the information being sought. If there are any consistent patterns or relationships within the data, statistical analysis can be a powerful tool for revealing them. Descriptive statistics can paint a clear picture of the variation along the dimensions that are amenable to measurement, such as life spans, target types, and attack frequencies. The information conveyed by descriptive statistics can also be a valuable complement to deeper and more context-sensitive qualitative studies. In his landmark study of the durations of terrorist campaigns in 87 countries, Gurr (1979) emphasised that although statistical estimates and large sample surveys were not a replacement for in-depth case studies, they were an essential complement to them. This thesis takes the view that there is no single best way to view the phenomenon of terrorism. Given the many subjective aspects of terrorism analysis, it is beneficial to use different methods to examine the same data or phenomenon.

The great diversity of sizes, structures, and ideologies observed among terrorist organisations may, prima facie, argue against including them in one sample. Yet such research designs are not without precedent. Hannan and Carroll (2000) pointed out that while there is often a great diversity of organisations, the assumption that each one is unique is too constraining for progress to be made in organisational demography. Terrorist organisations are more similar to each other than mining or banking organisations, for example, so knowing that an organisation is classed as ‘terrorist’ still conveys a great deal about what to expect from it. While Crenshaw (1992) advised against aggregating a wide variety of terrorist organisations into a single dataset, several justifications for aggregating

10

the 557 terrorist organisations that comprise this study into a single dataset can be made. Firstly, all of the organisations in the sample were involved in the intentional production of the same generic product - politically motivated violence. Secondly, the organisations in the sample used similar tactics, such as bombing, assassination, kidnapping, or arson. Thirdly, all of the organisations in the sample extracted a common set of material and social resources from their environment such as recruits, information, money, and weapons. The commonalities that exist among terrorist organisations become more salient when they are viewed from an open-systems perspective. These commonalities are explored in greater depth in Chapter Three.

This thesis proceeds by reviewing the previous literature on terrorist organisational life spans and life cycle models. Chapter Three discusses the use of a systems approach, specifically open systems theory, to describe the less obvious commonalities among seemingly diverse terrorist organisations. A description of the major structural and ideological characteristics of terrorist organisations, and their relationship with organisational survival, is presented in Chapter Four. A description of the various life cycle theories of terrorist organisations is presented in Chapter Five. The important relationship between organisational age and size is also discussed in Chapter Five. The description of the internal and external forces acting on an organisation, from its inception to dissolution, provides a contextual, and complementary, background to the quantitative analysis of organisational survival in the following chapter. A non-parametric and semi-parametric analysis of the life spans (or more precisely, the length of time, in years, between the date of an organisation’s first claimed attack until its last claimed attack) of 557 terrorist organisations is carried out in Chapter Six. The association between ideological orientation and the survival times of terrorist organisations is a major theme of the statistical analysis. The possible biases in the data, and limitations these may put on the generalisability of the statistical analysis, are also discussed in Chapter Six.

11

Several suggestions for further research are given in Chapter Seven. The thesis

concludes in Chapter Eight with a summary of the findings.

12

2. Literature Review

For counter-terrorism planners and analysts, an understanding of the major life cycle theories of terrorist organisations, and the assumptions they are based on, is vital. Counter-terrorism strategies need to be developed with an appreciation of the different forces that can influence the actions of a terrorist organisation at different stages of its life cycle. Counter-terrorist actions and strategies that are highly effective in the early stages of a terrorist organisation’s life cycle might be fruitless, or even counter-productive, if applied in the later stages of its life cycle. This chapter reviews the research on terrorist organisational life cycles and life spans, highlighting aspects of both that are relevant to the current study. This literature review proceeds along two general themes - terrorist organisational life cycles and terrorist organisational life spans.

2.1 The Organisational Life Cycle

Much of the research into the life cycles of terrorist organisations has drawn heavily on the idea of an organisation passing through various stages of birth, growth, maturity, and decline. These stage-based models have strong parallels to those found in the analysis of economic and voluntary organisations. The idea that organisations pass through a series of distinct and predictable developmental stages comparable to the birth, growth, maturity, and decline of biological organisms has its origins in 19 ^th Century Darwinism. Many modern life cycle models can trace their roots back to the work of economist Alfred Marshall. Marshall’s influential Principles of Economics (Marshall, 1890) compared the growth of business organisations to that of the life cycles of trees in the forest. The influential economist Kenneth Boulding (1950) did much to popularise the idea of a life cycle of organisations. Decades of cogent criticisms of the application of biological analogies of organisational development, spearheaded by Penrose

13

(1952), have done little to reduce the widespread appeal and continued use of this analogy, however flawed or controversial it remains.

One of the most influential stage-based models of organisational development was developed by Chandler (1962). Chandler (1962) described a four-stage life cycle model and outlined the circumstances that drove an organisation to move from one stage to another. Greiner (1972) outlined what became one of the most cited models of organisational evolution. Using a five-stage model, Greiner suggested that transition from one stage to the next was precipitated by crises that led to dramatic or revolutionary changes in an organisation’s processes and structures. The transition from one stage to the next was not seen as smooth or gradual. Greiner’s (1972) life cycle model is explored in greater depth in Chapter Five. In subsequent years, many life cycle models have been proposed (Kimberly & Miles, 1980; Churchill & Lewis, 1983; Mintzberg, 1984; Hanks, 1990; Dodge & Robbins, 1992). Lichtenstein, Levie, and Hay (2007) conducted a comprehensive review of the empirical evidence related to stage theory and concluded that despite its appeal, over 40 years of research in stage theory had not produced any consistent large-sample empirical confirmation of a particular stage-based life cycle model of economic organisations.

There is no universally applicable life cycle model of terrorist organisations. Given the great diversity of sizes, structures, ideologies, environments, and historical epochs, any such model would doubtless have many exceptions. Wilson’s (1973) observation that political organisations invariably shifted their focus from the achievement of a stated strategic goal to that of organisational survival has been a recurring theme throughout the literature. Wilson’s (1973) emphasis on organisational survival as the primary, albeit tacit, goal of political organisations (this includes terrorist organisations) was expounded on by Crenshaw (1985, 1991), Oots (1989), and Post (1990).

14

Many life cycle models of terrorist organisations emphasise how the growing tension between moderate and extreme members often leads to organisational schisms, internecine violence, and splinter groups (Bell, 1998; Roberts, 2003). Karber and Mengel (1983) drew attention to the pivotal role the decision to use indiscriminate violence, as opposed to discriminate and symbolic attacks, played in the evolution of terrorist organisations. Sprinzak (1998) described a life cycle model of left-wing organisations in democracies, describing three stages through which a peaceful non-violent political protest group went through on their way towards engaging in strategic terrorism. The first stage involves a crisis of confidence, where an enraged ideological challenge group opposes a detested leader or specific policy through verbal, and usually peaceful, protest. There may be small scale unplanned violence at protests and strikes. The second stage involves a conflict of legitimacy, where the entire system is viewed as rotten and in need of overturning or transformation through violence. The third stage involves a crisis of legitimacy, where everyone remotely associated with the system is seen as a potential target. The world becomes bifurcated into good and evil. A marker of this stage is the intentional use of rhetoric that dehumanises those associated with the regime, thus making it morally easier to kill them. This stage is associated with the adoption of strategic terrorism. Sprinzak (1998) pointed out that very few angry protest groups reached the crisis of legitimacy stage. Gurr (1979) believed that the perpetrators of terrorist acts seemed to be more often motivated by hostility towards particular policies or political figures than revolutionary aspirations.

The distinction between external strategic focus and internal organisational focus was brought into sharp relief by Crenshaw (1985) who contrasted instrumental and organisational approaches to understanding the behaviour of terrorist groups. The instrumental approach considers terrorist behaviour to be driven primarily by its publicly stated goals of liberating its constituency or influencing government behaviour. The organisational approach considers terrorist group behaviour to be

15

driven primarily by the internal dynamics of the group, the most basic of which is group survival. Oots (1989) and Post (1990) emphasised the frequency with which instrumental considerations gave way to organisational considerations as a terrorist organisation aged, thereby accommodating the two different views. One of the implications of this change is that belated, but genuine, attempts by the government to ameliorate the original grievance may have little effect on a terrorist organisation whose motivations for violence have become thoroughly dominated by intra-organisational dynamics.

It is not a straightforward task to reconcile the idea that organisations become more homogeneous and extreme in their opinions over time (Post, Ruby, & Shaw, 2002) against life cycle models such as Karber and Mengel (1983), Bell (1998), and Roberts (2003) which view an organisation-rending schism between moderates and radicals as almost ineluctable. The growing size of a terrorist organisation over time also means an increase in the ideological diversity within the organisation, perhaps serving as a counter-weight to the increasing homogeneity of opinions Post, Ruby, and Shaw (2002) referred to. Sprinzak’s (1998) three-stage model of left-wing organisational development, where each successive stage attracts different types of recruits, and sheds more moderate members, is consistent with an increasingly smaller, extreme, and homogeneous terrorist organisation.

Using a graphical approach analogous to industrial control charting, Lockett (1994) examined the level of violence of four terrorist organisations over their lifetimes. This life cycle model drew attention to the way terrorist organisations often failed to keep their level of violence between a level that was too high, thereby alienating supporters, or too low, thus failing to influence anyone. A terrorist organisation that manages not to stray too far from the optimal level of violence for their environment has a better chance of survival. Overestimating the maximum level of violence supporters will accept is a common cause of failure among terrorist

16

organisations (DeGhetto, 1994; Bell, 1998; Gerringer, 2002; McAllister, 2004). Attacks that are tactically brilliant may turn out to be strategically unwise.

While not expressly studying the life cycles or life spans of terrorist organisations, Chasdi’s (2002) extensive statistical analysis of how age, size, and ideology influenced terrorist organisation target selection in the Middle East demonstrated that as Middle Eastern terrorist groups aged, they tended to become less discriminate in their targeting.

Roberts (2003) proposed a five-stage model that bore a resemblance to the traditional four and five-stage formation, growth, maturity, decline, and collapse models used in the organisational analysis of business firms. The traditional maturity stage, however, was re-interpreted by Roberts (2003) as an adapt-orcollapse stage, where the organisation invariably struggles with its failure to translate tactical successes into strategic success. Most life cycle models of legal organisations have a maturity or consolidation phase as their mid-life, as well as less emphasis on decline. The emphasis Roberts’ (2003) places on the emergence of factions and schisms in the middle of the life cycle is justified given the widespread factionalism and infighting observed in many terrorist organisations (Oots, 1986; Cragin & Daly, 2002; Zirakzadeh, 2002; Siqueira, 2005; de Mesquita, 2006). Terrorist organisations often splinter along moderate-radical fault lines in the middle and later stages of their life cycles, a phenomenon which is not immediately evident among legal organisations. The description of a generic life cycle for a terrorist organisation in Chapter Five of this thesis draws heavily on the life cycle models of Karber and Mengel (1983), Long (1990), Bell (1998), and Roberts (2003).

17

2.2 The Organisational Life Span

Research concerned with measuring the life spans of organisations has made use of statistical models drawn from mortality studies and demography, such as the Gompertz distribution and Makeham-Gompertz distribution. Stinchcombe (1965) was influential in promulgating the theory of a liability of newness, which referred to the very high rates of failure among new organisations of all kinds. The effects of organisational age and size dominate investigations into the failure rates and life expectancies of organisations, but the effects of environmental changes, founder personality, previous experience, and circumstances of organisational formation have also been investigated (Hannan & Freeman, 1987; Hannan & Carroll, 2000). Rapoport’s (1992) suggestion that perhaps as many as 90% of terrorist organisations fail in their first year comported with Stinchcombe’s (1965) liability of newness theory and highlighted the fact that the vast majority of terrorist organisations fail more quickly than their founders and members anticipate. Gurr’s (1979) study of the duration of terrorist campaigns found that campaigns of violence were typically short-lived and conducted by tiny groups. One of the goals of this thesis is to apply statistical methods from the field of failure and survival analysis in order to analyse the survival times of a large number of terrorist organisations.

The association between ideological orientation and longevity of terrorist organisations has been addressed in the literature on a mainly descriptive basis, but there has been little disagreement over the much longer duration of nationalist/separatist organisations. The advantages an ethnically, linguistically, or religiously distinct supporter base confer on a terrorist organisation mobilised around a revanchist, separatist, or nationalist cause are well-known (Gurr, 1979; Crenshaw, 1991; Rapoport, 1992). In fact, Rapoport (1992) claimed that without a distinct nationalist or ethnic base, terrorist organisations almost always fail. Less

18

well-known is the effect of left-wing, religious, and right-wing ideologies on organisational life expectancy. Crenshaw (1991) compiled a list of terrorist organisations (n = 77) and noted that separatist/nationalist and left-wing organisations appeared to be roughly equally represented among the 47 organisations in the list that had survived for more than 10 years. The data analysed in this thesis allow a similar, but not direct, comparison to be made by looking at the predominance of certain ideologies and certain combinations of ideologies ascribed to the organisations in the sample that have operated for 10 or more years (see Appendix A, Table 4). Crenshaw (1991) noted a number of unavoidable conceptual problems that arise when measuring the life spans of terrorist groups. The first problem was deciding which organisations to include. The second problem was specifying the time period during which the organisation was active - taken as the year when the strategy of terrorism appeared to be adopted. Although acknowledging that the nature of the data did not lend itself to definitive conclusions, Crenshaw (1991) drew attention to the durability of the separatist organisations in the list. The statistical analyses presented in Chapter Six of this thesis build upon Crenshaw’s (1991) study by using a much larger data set (n = 557) and Cox proportional hazards regression to investigate if ideological orientation can help to explain the variation in life spans observed among terrorist organisations.

Rapoport (1992), while tracing the rise of contemporary religious terrorism from the mid to late 1970s, drew attention to the longevity of religiously-inspired violent clandestine organisations such as the Hindu Thuggee cult (13 ^th to 19 ^th Century India) and the Ismaili Shiite Nizari Assassins (11 ^th to 13 ^th Century Iran) against the comparatively fleeting existence of modern terrorist organisations. While raising doubts over the relevance of traditional organisational analysis and life cycle models in an era of web-based recruitment and leaderless resistance, Cronin (2006) nonetheless encouraged more research on the life spans of terrorist organisations,

19

particularly those that used religion as their primary mobilising ideology. The wave of religious terrorism that emerged in the late 1970s and early 1980s is much more recent than the separatist/nationalist wave which began soon after 1920, and the USSR-sponsored social revolutionary/left-wing wave, which began in the 1960s. The relative recency of contemporary religiously-inspired terrorism, which derived its impetus and early succour from the 1979 Iranian Islamic revolution and the USSR’s 1979 invasion of Afghanistan, has meant that fewer life cycle models have been developed to reflect the distinct characteristics of terrorist groups mobilised around religious ideals.

Using open-source data, Blimes, Berohn, and Mewhinney (2005) used Cox proportional hazards regression to assess the effect of various county-level independent variables on the duration of terrorist organisations in 160 countries. Their study excluded terrorist organisations that had not carried out 10 or more attacks, thus biasing their sample towards more tactically successful and aggressive organisations. Truncated and biased datasets are very common in organisational studies, where the majority of organisations studied tend to be the larger and more successful ones (Hannan & Carroll, 2000). While acknowledging the limitations of their data, Blimes, Berohn, and Mewhinney (2005) found that terrorist organisations based in countries with a high proportion of Muslims, large urban populations, and liberal political institutions operated for a significantly longer duration. Their study illustrated the importance of incorporating environmental variables into the analysis of the terrorist organisational life spans.

The great diversity of organisations that use terrorist tactics has meant that life cycle models that reflect a segmented approach to their applicability are more realistic. Sprinzak’s (1998) life cycle model of left-wing terrorist groups in a democratic society demonstrates this more finely-tuned and contextually sensitive

20

approach. The diversity among terrorist organisations is a common obstacle to meaningful and generalisable statistical studies.

The next chapter uses an open systems framework approach to highlight the way diverse organisations can be described, understood, and compared in terms of their generic sub-systems and functions.

21

3. Terrorism through an Open Systems Framework

This chapter discusses the view of terrorist organisations as systems. The early 20 ^th Century theories that promoted the view that organisations were analogous to machines are contrasted to the more modern view of organisations as analogous to organisms. One of the main themes of this chapter is the contribution General Systems Theory made towards characterising organisations as open systems which imported, transformed, and outputted material to their environment.

3.1 The Organisation as a Machine

Early organisational theorists drew heavily on the analogy of the organisation as a machine. This view was prominent in Taylor’s (1911) pioneering research on scientific management. The mechanistic view of organisational behaviour is entwined with bureaucracy and hierarchy. It was rapidly adopted by an increasingly industrial workplace, one where workers were seen as little more than cogs in a giant machine. By working in unison, the organisation’s goals could be reached with maximum efficiency. Taylor’s focus was on task and structure. There was little emphasis on the environment with which the organisation interacted, nor was much thought given to the human ‘cogs’ that ran the organisation. The focus on task and structure reflects the typical closed system view of an organisation. A closed system is one that is totally insulated from its environment. Although it is unrealistic to claim that any organisation is an entirely closed system, Jackson (1991) pointed out that the most influential organisational theories of the first half of the 20 ^th Century disregarded the relationship between an organisation and its environment in favour of the pursuit of internal operational efficiency and control. One of the reasons for this oversight was the increasingly mechanised industrial environment within which Taylorism flourished. A production line built on the theory of scientific management can be treated as a closed system to the extent that it is sufficiently insulated from fluctuations in demand and supply (environmental

22

contingencies) through the stockpiling of raw materials and finished goods to keep it in a relatively static environment.

Taylor’s (1911) theories were heavily influenced by industrial engineering and formed the basis for Weber’s (1947) bureaucratic theory of organisations. Weber (1947) emphasised the importance of hierarchical power structures and formal rules in order to establish clear lines of authority and control, reduce diversity and ambiguity, insure stability and uniformity, facilitate the division of labour, and increase task specialisation. This epitomised the reductionist approach to organisational analysis, one that sought to understand the behaviour of an organisation by decomposing it into ever-smaller components. Weber (1947) believed that a deeper understanding of the cause and effect relationships among these components was a necessary step towards greater control over the organisation’s behaviour.

The bureaucratic and hierarchical model of organisations, along with its machine analogy, had two major flaws; it did not deal well with internal conflict, nor did it adequately address the issue of structural adaptation in response to a changing environment. Barnard (1938) believed that organisations were cooperative systems, analogous to mechanical systems in equilibrium. One of Barnard’s most important contributions to systems theory was the idea that effectiveness and efficiency could be achieved by taking into account the formal and informal interactions within the organisation. The formal aspects of the organisation are the consciously coordinated activities that define a common purpose, reward organisational members, and put individuals in communication with one another. Informal structures arise without a common or coordinated joint purpose. Barnard (1938) believed that the informal organisational structures that existed alongside the formal structures were often overlooked by management theorists. Despite the limitations arising from of his reliance on the organisation as a machine analogy,

23