Transcript Deadly Contests: An Economic Note on al Qaeda's Reward System

Al Qaeda and Jihadist Terrorism in the
light of contest theory
A theoretical note and empirical evidence
for the period 2004-2008
Raul Caruso
Università Cattolica del Sacro Cuore di Milano
[email protected]
Friedrich Schneider
Johannes Kepler University of Linz
[email protected]
1. Introduction
2
Economic Explanations of Terrorism
Krueger and Maleckova (JEP, 2003),Li ( JCR,
2005), Blomberg, Hess and Weerapana (EJPE,
2004), Abadie (AER, 2006), Berrebi (PEPS,
2007), Benmelech and Berrebi (JEP, 2007),
Freytag, Kruger and Schneider (wp,2008),
Sayre (PEPS, 2009).
18/01/10
1.1 Classical Economic interpretations
of terrorism
1) Opportunity cost argument. That is, the larger is
the set of economic opportunities for an individual
the lower is the likelihood or the willingness for
him to be involved in a terrorist activity.
Consequently, low-income and poorest countries
would be the natural incubators of terrorism.
2) Productivity argument which stresses the positive
relationship between education and terrorist
activity. That is, better educated individuals would
also become more productive and bloodier
terrorists.
4
1.2 Novelty of present work
In this work we present two main contributions to the economic
interpretation of terrorism:
(1) This paper interprets terrorist incidents in the light of the
contest theory.
(2) The paper studies only the Al Qaeda-style terrorism. This
makes the analysis peculiar and focused on radical jihadist
groups. It cannot be compared with foregoing studies.
(3) In the empirical application, the dependent variable is the
number of casualties and not the incidence of terrorism in
itself.
5
1.3 To start: Few insights
Al Qaeda does not retain a clear hierarchical line of command. In
particular, this flexibility allows for a novel recruitment
system. In fact, the recruiting process may now resemble a
kind of voluntary application to join the organization
In this view, new groups are involved in the organization as the
result of a selection process amongst different volunteers. The
rise of the so-called ‘self starters’ is taken as evidence of this,
i.e. groups with little or no affiliation with the network
perpetrating terrorist attacks on their own initiative
In particular, attacks in Instanbul (November 2003), Madrid
(March 2004), London (twice in July 2005) seemed to confirm
the emergence of such phenomenon. Cozzens (2005).
6
2. Contest Theory
7
Contests: theoretical underpinnings
A contest is commonly defined as a game in
which players compete for a prize by making
irreversible outlays. In other words, contests
are situations in which rational agents spend
resources in order to win a prize. The
characteristic feature of this interaction is that
resources are spent irreversibly. They are also
labelled as all-pay auctions.
18/01/10
Contests
Examples:
Rent-Seeking and Lobbying
Sport Tournaments
Patent Races
Lawsuits
Career Incentives
Academic Competition
Main References: Tullock (1980), O’Keeffe, Viscusi,
Zeckhauser (1984); Rosen (1986); Dixit (1987); Hillman and
Riley (1989), Nti (1999/2004); Baik (1998), Szymanski
(2003), Amegashie (2006). Moldovanu, Sela (2001),
Moldovanu, Sela, Shi (2007). Konrad(2009)
18/01/10
Contests: general
(1)The probability of winning of each agent in a contest
depends upon the ratio between its own effort and the
total effort.
(2)The effort of each contestant is increasing in the
value of the prize.
(3)The higher is the number of contestants the lower is
the effort of each contestant
(4)When the prize is equally evaluated, contestants exert
the same level of effort
(5)In the presence of asymmetry in the evaluation of the
prize, low evaluation players may give up
18/01/10
3. Deadly Contests
11
2.1 Al Qaeda as a contest organizer
Al Qaeda may be portrayed as a contest
organizer providing a prize to the best
terrorist group.
The prize could be assumed to be some sort of
ideological blessing (being accepted as a full
and honourable member of the organization) as
well as economic reward.
12
2.2 Deadly Contests
Candidate terrorists must then compete with each other
to prove their commitment and ability. Each
candidate group observes the results of some previous
attacks. Candidate terrorist groups compete by
maximizing their efforts to win the prize, i.e.
maximizing the number of casualties.
In particular, to maximize their own probability of
winning the prize, each group tries to make attacks at
least equally destructive as the previous attacks.
13
4. Peculiarities of our current
approach
14
3.1 Uncertainty about the number of
contestants
(1) Presumably, a crucial information which is
not publicly available is the number of
contestants. Namely, participants do not know
(at least not exactly) the number of contestants.
As noted in Munster (2006, PC) this increases
the total level of efforts exerted.
15
3.2 Information costs
(2) monitoring and information costs are close
to zero. When a terrorist group bombs an
embassy or a trade centre with dozens of
casualties somewhere in the world, the
event is extensively broadcast by
international mass media. (see Rohner and
Frey, PC, 2007 for an empirical evidence of
such association).
16
5. Empirical application
17
4.1 Testable implication
In the presence of a contest, each candidate group
observes the results of some previous attacks. Hence
to maximize their own probability of winning the
prize, each group (maximizes its effort) tries to make
attacks at least equally destructive as the foregoing
attacks.
The testable implication is that: the number of
victims of terrorist attacks is significantly
associated with the number of victims of
past attacks.
18
4.2 Empirical application
(i) Another assumption is that presumably if a
tournament takes shape it does at a national
level. Perhaps, Al Qaeda tries to recruit
national champions.
(i) Secondly, it is possible that terrorist waves
may emerge. (namely a tournament takes
place at a certain time)
19
Jan-Febr
2004
Dec2004Feb2005
Terrorist waves
20
TABLE 1- VARIABLES, DESCRIPTIVE STATISTICS AND SOURCES
Description
Source
Obs.
Mean
Std. Dev.
Min
Max
NCTC
215
2.445
1.681
0
7.62
2.502
1.647
0
7.14
Opportunity
cost
215
7.526 argument
1.056
5.53
10.88
Productivity
argument
215
3.539
5.693
-10
10
0.4
0.98
Opportunity
cost
argument
215
0.698
0.099
0.4
0.95
UNDP
208
6.025
3.743
0.7
13.4
SIPRI
214
9.322
2.115
4.17
12.52
UNDP
215
50.288
18.691
16
95
UNDP
208
3.606
0,112
3.4
3.80
Victims of incidents
Victims
(logged)
Contest
Victims of previous
incident in the same
PastVict
country (logged)
GDP per capita
Gdppc
(logged),
NCTC
201
IMF, WEO
Polity IV project index,
bounded between -10
Polity
and 10.
Polity IV Project
Education
Education index
UNDP
Human Development
HDI
Index
215
0.715
0.165
UNDP
average annual change
in consumer price index
CPI
in 2004-2005 (logged)
Military expenditures,
Milex
(logged)
% people living in
Urban
urban areas
Gini
Gini index of income
Index
inequality (logged)
21
4.3 Empirical Evidence: Sample
The sample covers 22 countries for period Jan
2004 – March 2008.
(Algeria, Bangladesh, Egypt, Eritrea, Ethiopia,
France, India, Indonesia, Iran, Jordan,
Pakistan, Philippines, Qatar, Russia, Saudi
Arabia, Spain, Syria, Thailand, Turkey, UK,
Uzbekistan, Yemen)
22
4.4 Empirical Evidence: Data
Filters:
1) Only Islamist extremism and IED (as
reported in NCTC)
2) Potential Outliers excluded: Afghanistan,
West Bank and Gaza strip, Iraq, Israel;
See the
3) Attacks to facilities
(ex. Pipelines)
Manchester
Manual excluded;
4) Assasinations (oravailable
tentative)atof political
leaders excluded;
.
23
Table 2 - Dependent Variable: Log Victims by Event (Negative Binomial Regression)
1
2
3
4
5
6
7
8
9
10
0,06**
0,049*
0,047*
0,048*
0,054**
0,054**
0,051*
0,054**
0,055**
0,05*
-0,16***
-0,24***
-0,2***
Polity
-0,008
-0,009
Urban
0,004
Education
0,182
Pastvict
Gdppc
0,0004
-0,004
0,36
HDI
-1,21***
-1,22***
-1,08**
CPI
Milex
-0,253
-0,04*
-0,04*
(milex*polity)
Population
0,09***
Gini
(Gini
*
Polity)
(population*
polity)
Const
Obs
Log
Likelihood
LR - χ2
0,73***
1,91***
2,26***
2,02***
1,58***
1,58***
1,73***
1,10***
1,08***
-0,28
201
201
201
201
201
201
201
201
201
201
-374, 009
4,51
-369,023
16,25
-367,055
19,2
-368,407
18,96
-371,092
12,46
-371,091
12,5
-370,522
13,32
-372,66
8,01
-372,755
7,73
-371,023
13,46
Notes: robust standard errors in parenthesis, p-values in square brackets, ***significant at 1%, ** significant al 5%, *significant at 10%. For sake of readability statistically
significant coefficients are in bold.
24
6. Summary of findings and
Policy Implications
25
5.1 Summary of findings
1) The number of victims is significantly
and positively associated with the
number of victims of the previous
incident in the same country.
2) The number of victims is decreasing in
(i) the GDP per capita or alternatively
(ii) in the HDI.
26
5.2 Policy Implications
(1) A general improvement of standard of living has the
potential to reduce the likelihood (or even the
brutality) of terrorist attacks. In fact, results show a
negative association between number of victims and
HDI and GDP per capita alternatively.
(2) Since reward to would-be terrorist groups is
expected to be monetary, therefore, tracking
financial flows of terrorist organization becomes
a critical task. The argument for an international
cooperation on regulating financial flows is thus
strengthened.
27