Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
1
Religion, Terrorism and Public Goods
2
Prime Minister’s Office, Algeria – 4/11/071 Al Qaeda suicide bomber, 12 dead, 112 injured
3
The Global War on Terrorism
167* 10,000* Post 9/11
October 01 –September 06
1094,800Pre 9/11January 98 –August 01
Fatalities / monthFatalities
Fatalities due to Terrorist Attacks Worldwide1998-2006
Source: Memorial Institute for the Prevention of Terrorism, Oklahoma City, www.tkb.org•Excluding casualties in Iraq and Afghanistan. Including those fatalities the 10/01 – 9/06 figure is447 rather than 167.
• Note change in organizations threatening civilians and governments.
4
Outline
0. Introduction: The Violent Puzzles1. Background: Hamas, Hezbollah, Taliban, Insurgency
and Suicide Attacks2. Framework: Terrorist Clubs vs. Hard Targets 3. Testing: Clubs, Hard Targets and Suicide Attacks4. Policy Implications: Counterterrorism, Economic
Development and Nation Building5. Clubs vs. Rational Peasants, (or Gangs and Community
Policing (A&Y))6. Street Lights: Some results from Iraq
5
Hamas
6
Hamas 2006
7
Taliban
8
Hezbollah
9
Conclusions• Why so few terrorist organizations?
Defection constraint.• Why are religious radicals effective terrorists?
Solve defection constraint in benign activities.Not necessary theology.
• Why suicide attacks?Hard targets.
• What to do about it? Compete in providing benign services and competent governance, with muscular protection.
• Future WorkNeed research and evaluation
Distinguish club model from standard “rational peasant”approach to “winning hearts and minds”
10
Jewish Underground
11
Muqtada al-Sadr’s Mahdi Army
12
A Violent Puzzle Among Religious Sects
The Violent Puzzles:• Why are radical religious militias and terrorists so efficient at
violence?HamasHizbullahTalibanAl SadrAl Qaeda
They make the secular terrorist organizations of the 60s-90s look lame
• Why Suicide Attacks? • Our approach: Draw on
a) insurgency literature in IR, b) economics/sociology of religion, c) agency and collective action in organizations
13
What Motivates Terrorists? The Afterlife and Other Myths
• Is the advantage of radical religious terrorists due to the superior motivation that stems from theology and beliefs?- An ideology of hate? - Promises in the afterlife?
• Israeli psychiatrist Ariel Merari has spent years interviewing suicide attackers, their families and friends.
• He finds that: - Hamas and Jihad suicide attackers never mention religion or virgins in heaven as their primary motivation
• Consistent with experience in other countries- many suicide attackers worldwide are not religiousradicals, including the Tamil “Tigers” and the majority of attackers in Lebanon in the 1990s.
14
So what does motivate suicide attackers?
• Merari finds that there is no specific primary motivation- not economic depravity, - not depressed or suicidal or mentally ill
- consistent w/ research on Bader-Meinhoff, Red Brigade, ETA
- not ignorant - generally not seeking revenge
• Might be best thought of as well adjusted altruists, who truly believe that their courageous act will help their communities- combination of altruism and delusions of self-importance- close to profile of rational recruits to sects; triage needed
• Now that’s a frightening thought, because the world is full of self-motivated altruists who are willing to give their lives for some cause- and indeed there seems an ample supply of suicidal terrorists
15
Suicide Attacks as a Rebel Tactic
• Civil wars 1945-1999 (Fearon-Laitin)127 in 69 countriesdirectly account for 16m fatalities
• Rebel tactic is usually rural insurgency• Suicide attacks are very rare, but becoming more
common
16
Table 1: Suicide Attacks by Country of Perpetrator
17
Table 1: Suicide Attacks by Country of Perpetrator
18
Income, Insurgency and Suicide Terrorism
19
Religious Differences between Insurgents (Suicide Attackers) and Targeted Victims
(Table 3)
Insurgency
16.50%
83.50%
Suicide Attack
13%
87.4%
DifferentReligion
SameReligion
Unlike civil wars, this is the only consistent predictor.Why? Because coreligionists are soft targets.
20
2. Terrorist Clubs vs. Hard Targets
Militia activity - Coordinated rent captured involving violence.
• e.g. attacking occupying army, providing law and order, organizing and carrying out a clandestine activity (like terrorism).
• often involves personal risk. • Key aspect is sensitivity to defection.
21
Map of Afghanistan
22
A. Securing a Trade Route
31 2 4... ..N-1 N1
Convoy Destination
Checkposts
$B
23
Securing a Trade Route
C Ri C Rii
N
({ }) ( )==
∏1
Convoy will choose to set out only if all Ri = 1.
Payoffs: Club extracts surplus B and shares it equally among members, who buy goods at price P.
Benign local public goods provided by govt., G and club, A.Defector’s outside option is wi, but no access to C or A.
Incentive compatibility: Member loyal iff
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
24
Securing a Trade RouteC Ri C Ri
i
N
({ }) ( )==
∏1
If ICC fails this is an N player prisoner’s dilemma, resulting in an unsafe route, no convoy and no rents.
Adverse selection: Imagine two unobserved types (as above), such that
ICC holds if wi = wH (> wL) for all i. A club with a costly sacrifice as an initiation rite which successfully excludes all low wage types can secure the route and extract the rent.E.g., The Taliban
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
25
Figure 2: Selecting Low Wage Membership Allows Larger Projects
Project Value (B)
High wagedefector
Low wagedefector
Loyalmember
B*B* B**
Incentivecompatibilitylimit with lowwage members
26
B. Capturing a Hill
2
3
N
1
$B
27
3. Clandestine Violence
2
N
1
Conspirators
Target
Defection – The reason so few militias and terrorist organizationssurvive.. defection is common.
How do successful militias and terrorist organizations prevent defection?
They have an organizational advantage.
28
An analogous problem: Reducing free riding in
Religious Sects• Sect – a religious group that:
– imposes extreme prohibitions and requires distinctive sacrifices
– views secular society as corrupt, dangerous, and threatening – economic life: high levels of mutual aid, and local public goods
provided through volunteer work, e.g. education, health care, law and order, welfare services, orphanages, day care, soccer clubs
How can you trust members to apply full effort?
• Internal economies of sects rely on trust-based transactions- sacrifices are elicited early in life to signal commitment
e.g., education, missionary work, jail time- prohibitions distance members from market culture
29
30
Rational choice approach to religious sectsIannaccone (1992)
Formally..
(1) Ui = U (Si, Ri, C({Rj}) ), where S – consumption, R – religious activity,C – local public good .
(2) C({Rj}) = for j=1 to J .
C could be mutual insurance, health care, education.
(3) R = T – H . Budget constraint for time.(4) wH = S . Budget constraint for money.
Figure 1 illustrates optimal religious prohibitions.
∑=
=J
j
j
J
RR
1
31
Analogy: Seminar as a Club
• A seminar (like this one) is a club, where participants benefit from their own effort “R” and the average R of colleagues .
• A good citizen comes prepared, asks questions, provides good answers, all because she studies.
• Lacking a way to subsidize R, the club would like to tax outside activity of members.
• In principle, a research club should tax, or tithe, if it can.But it typically lacks tax authority.
32
Optimal Prohibitions for Seminar Participants
• Efficient proxy taxes on outside options might be:
• Prohibit alcohol with nonmembers
• Prohibit beach on Sabbath• Dress strangely• Limit eating with nonmembers
through dietary restrictions• Limit communication with
outsiders by speaking arcane language
• With enough prohibitions seminar participants would have nothing better to do with their time than study
• Enforcement could be through threat of expulsion or through peer pressure
If this example doesn’t work for you, think of a fraternity (or a team),where R is partying (training) and helping out other members.
33
Fig 1: Optimal Taxation Through Prohibition
Work hours
Wages
Religious activity
34
Maimonides Rationale for Circumcision• Twelfth century philosopher Rav Moses Maimonides
explaining circumcision..• “It gives to all members of the same faith, i.e., to all believers
in the Unity of God, a common bodily sign, so that it is impossible for any one that is a stranger, to say that he belongs to them. For sometimes people say so for the purpose of obtaining some advantage ... ....It is also a fact that there is much mutual love and assistance among people that are united by the same sign when they consider it as [the symbol of] a covenant.
• [The Guide for the Perplexed, late 12th century, translated 1904. Chapter XLIX. Brackets are those of the translator. Italics are my own.]
• Signaling theory won a Nobel Prize in Economics, 9 centuries later
35
Rationalizing Sacrifices
Imagine heterogeneity in participants’ outside options, wj ,(call them “wages).
Members would prefer that other members have low wages wj , since that implies higher R and larger externalities .
Low R members are free-riders who it would be efficient to exclude, but wj is unobserved.
36
Rational Sacrifice (cont.)
• Voluntary sacrifices of time might exclude high wage individuals but include low wage for an efficient separating equilibrium.
e.g.s Insist on an arcane language that takes years to learn
Religious education with no market value
370
0
�
�
B1
A1
��A2B2
High wage, low C
Low wage, low C
High wage, high C,sacrifice
Low wage, high Csacrifice
Work Hours (H) º » R+Kκ∗H1
Figure 1: Rationalizing Sacrifices
38
Iannaconne (92): Sects and Churches in National Data
39
Subsidizing Sacrifice
Source: Berman (2000)
40
Evidence: Fertility and Schooling Results• Data : extensive search yielded household surveys in
Indonesia, India, Bangladesh, Cote D’Ivoire, Pakistan (B&S ‘04) and Israel (B ‘00)
• Women in families with Islamic and Ultra-Orthodox religious education have higher fertility in all 6 countries, by 2/3 to one more expected lifetime child.
• Islamic and Ultra-Orthodox education have significantly lower rates of return than secular education in 3 of 6 countries; insignificant results in other 3 countries
• Prevalence of radical religious schooling: 2-5% of Muslims in Rural Bangladesh, Pakistan, Cote
D’Ivoire, 5% of Israeli Jews14-25% in Indonesia and two Indian States
(Uttar Pradesh and Bihar)
41
Fig 1: Optimal Taxation Through Prohibition, and Fertility
Work hours
Wages
Religious activity& Fertility
42
Differential Fertility by Sect MembershipSix countries – Berman and Stepanyan (04)
AnyAnyOwnAnyOwnAnySect indicator
0.66(.39)
1.34(.46)
0.58(.27)
0.77(.43)
0.67(.26)
5.34(.30)
Diff. fertility
PakistanCote D’Ivoire
Bangla-desh
UP & Bihar
Indo-nesia
Israel
43
Differential Returns to Education by Sect Membership - Six countries (Berman and Stepanyan ’04)
-.048(.026)
.132(.006)
Pakistan
.175(.010)
.097(.007)
.122(.008)
.116(.005)
.094(.002)
Secular schooling
-.029(.070)
-.073(.034)
-.051(.229)
-.022(.013)
-.076(.006)
Religious schooling
Cote D’Ivoire
Bangla-desh
UP & Bihar
Indo-nesia
Israel
44
Fertility Differential for Ultra-Orthodox Jews (Berman ’03)
TABLE V TOTAL FERTILITY RATES OF ISRAELI SUBPOPULATIONSA. Source: Labour Force Survey
Period Full Population Jews Ultra-0rthodox Jewsc
All other Jews
1980-1982 2.99a 2.76 6.49 2.61
(0.04)b (0.04) (0.31) (0.04)
obs. 31347 27635 1040 26569
1995/96 2.66 2.53 7.61 2.27
(0.04) (0.05) (0.30) (0.05)
obs. 27866 22776 1021 21755
Change -0.33 -0.23 1.13 -0.34
(0.06) (0.06) (0.44) (0.06)
B. Source: Population Registry
Period Full Population Jews Christians Muslims
1980 3.14 2.76 2.66 5.98
1995/96 2.90 2.57 2.19 4.65
Change -0.24 -0.19 -0.47 -1.33
45
Analogy: Military unit as a Club• Like a sect, a unit is involved in cooperative
production, i.e., participants benefit from their own effort and the average effort of other members.
• A good soldier/member comes prepared, trains, works hard, covers his/her buddies, would never defect.. because they are devoted.
• Lacking a way to subsidize devotion, the unit would like to a) tax outside activity of members, b) select devoted members.
46
Why are sects effective at violence?
• Recall that militias and terrorist groups are organizations extremely sensitive to defection
• Sects have a strong advantage at coordinated violence because their benign service provision activities help thema) select operatives unlikely to defectb) influence operatives through their support of friends and family
• Testable implication: a sect will be more effective the stronger its’ social service provision
47
Figure 2: Selecting on Low Wage Membership Allows Larger Projects
Project Value (B)
High wagedefector
Low wagedefector
Loyalmember
B*B* B**
Incentivecompatibilitylimit with lowwage members
48
Figure 3: Benign Activity Increases a Militia’s Potential
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
Project Value (B)
Utility ofdefector
Utility of loyalmember
B**
Incentive CompatibilityLimit - augmented clubgood
B*
Utility of loyalmember - augmentedclub good
IncentiveCompatibility Limit
49
Resolving the Puzzle
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
• Taliban, Hamas, Hizbullah, Sadr’s Militia are all examples of remarkably effective violent radical Islamic organizations which started out as classic sects providing social services.
• Cooperative production of social services has the same “free-rider” problem, though less extreme e.g., mutual insurance is sensitive to defection
• An organization designed to limit defection in a benign context will have a huge advantage in the cooperative production of violence
50
TABLE 4: SOCIAL SERVICE PROVISION AND LETHALITY OF TERRORIST ATTACKS
ISRAEL AND LEBANON: 1968-2006
Group nameAttacks Injuries Fatalities
Injuries per attack
Fatalities per attack
(std. error)
Hamas 70 2202 413 30.2 5.9 0.87Hezbollah 90 387 449 4.3 5.0 2.82Palestinian Islamic Jihad 38 722 111 12.7 2.9 0.81Popular Front for the Liberation ofPalestine
38 376 107 9.9 2.8 1.03
Fatah/PLO 131 1465 279 11.20 2.1 0.48Democratic Front for the Liberationof Palestine 21 240 22 10.4 1.0 0.37
Unknown 427 1055 351 2.2 0.8 0.28
Social Service Providers: Hamas andHezbollah 160 2589 862 15.8 5.4 1.62
Others: DFLP, Fatah/PLO, PIJ, PFLP 228 2632 519 11.6 2.3 0.35Difference 3.1 1.67*
51
Suicide
SuicideSuicide Suicide
Suicide Suicide Suicide0
20
40
60
80
100
120
Att
acks
Hamas
Tanz
im
Al Aq
sa M
artyrs
PIJFa
tahPF
LP
Force
17
Terrorist Organization
Who Selects Suicide Attacks?
Suicide attacks are so damaging that only defection proof organizationscan succeed at them. Most do not try.
52
Efficiency of Suicide Attacks in Israel and Lebanon
63
4437
7
31
2 17.2
17.3
4 2.9 2.8 0.5 00
10
20
30
40
50
60
70
Hamas
Hizbulla
h PIJPF
LP
Al Aq
sa M
artyrs Fa
tahSS
NP
Terrorist Organization
Suicide Attacks Ave. Fatal./Attack
53
Lethality of Suicide Attacks in Israel and Lebanon
63
4437
7
31
2 17.2
17.3
4 2.9 2.8 0.5 00
10
20
30
40
50
60
70
Hamas
Hizbulla
h PIJPF
LP
Al Aq
sa M
artyrs Fa
tahSS
NP
Terrorist Organization
Suicide Attacks Ave. Fatal./Attack
54
Recap: Sects and Violence
• Why so few terrorist organizations?Defection constraint.
• Why are religious radicals effective terrorists?Solve defection constraint in benign activities.Not theology.
• Why suicide attacks?Hard targets.
• What to do about it? Compete in providing benign services and competent governance, with muscular protection.
• How can we be sure it will work?Need research and evaluation, just like any weapon.
55
Sect CharacteristicsHamas Taliban
Localpublicgoods
schools,hospitalswelfare, militias
law and order,militias
Militia activity
assassination ofinformants,attacks onIsraeli civiliansand Israeli military
guardedsmuggling routes, law & order,conquered Afghanistan
Increasedstringency
dress codes, personal piety, worship,
personal piety
Sacrifice risking arrest,injury or death
madrassaattendance
These benign activities are the norm among religious sects for Christians (Iannaccone 92), Muslims and Jews (Berman 00, 03)
56
Constructive Efficiency : [Figure 7]
If ICC does not hold it may be efficient for a club to make it hold by
• 1) raising C(1), through some other investment that augments local public goods (e.g. welfare, hospitals, etc.),
• 2) making cash payments to members (e.g. families of martyrs),
57
Figure 8: Govt. Provision of Public Goods Reduces Militia’s Potential
(ICC) U( BN
,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
Project Value (B)
Defector - high govt services
Loyal member -high govt. services
B** B*
Defector - low govt. servicesLoyal member -low
govt. services
Incentivecompatibility limit -high govt services
E.g., Malaya, Phillipines, Egypt
58
Destructive Efficiency: [Figure 8]
• 3) reducing G , the public good available to members and nonmembers, (assassination of public officials),
• 4) limiting B (ban on heroin cultivation), • 5) raising P (general strikes, access to goods markets),• 6) lowering wi , the outside options of members,
(Madrassah, jail time, secluding women, harassing nonmembers, destroying or banning access to Israeli labor markets).
59
Implications
Violent radical religious groups thrive where.. • a) govt. provision of local public goods is weak
- Somalia, Nigeria, Pakistan, Algeria, Lebanon, Afghanistan, Palestine, Iraq
• b) local militias are popular..- Chechnya, Afghanistan during war, Kashmir, Palestine, Jordan (“Black September”), Palestine, Iraq
• c) wages are low.. - all of the above,
• d) where outside subsidies are available,- Kashmir, Lebanon, Palestine, Afghanistan, Iraq
60
Suicide Attacks, Terrorism and Insurgency
• Why suicide attacks?• Deadliest method of delivering explosives to a
target- precise- leaves no operative to interrogate
• Method of choice vs. “hard” targetsi.e., targets whose destruction implies a high probability of death or capture
61
Hard Targets• p(h) – probability of apprehension increases in
govt. investment in “hardening” target• Expected utility, loyal operative, suicide attack..
where D is damage, benefit B is proportional to D• Utility from defection:
• Utility from conventional attack:
• Choose suicide attack if (7a) > (7b), (7a) > (7c)
62
Figure 4: Strong Governments Harden Targets, Insurgents respond with Suicide Attacks
Utility of Defector -
Damage to Target (D)
DC
Utility of LoyalMember -Suicide Attack,
E
O
D
DD DE
Utility of LoyalMember -ConventionalAttack, low p
Utility of LoyalMember -ConventionalAttack, high p
C
Notes: a) Global decline in insurgency;b) Coreligionistsare usuallysoft targets; c) Foreign allies of govt. are hard targets – switch to suicide attacks puts them at risk.
63
TABLE 5: ATTACKS ON ISRAELI RESIDENTS BY LOCATION AND TACTIC
Sept 2000 through July 2003
64
Attacks on Israelis by Location and Tactic
Attacks on Israeli Residents
17405
730
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Palestine Israel
Attacks
Attacks on Israel Residents
341
511
8
401
0
100
200
300
400
500
600
Palestine Israel
Fatalities Suicide Attack Fatalities
Why? Because targets in Palestine are soft, whereas targets in Israel are hard.
65
Religious Differences between Insurgents (Suicide Attackers) and Targeted Victims
(Table 3)
Insurgency
16.50%
83.50%
Suicide Attack
13%
87.4%
DifferentReligion
SameReligion
What about this? Coreligionists are soft targets.
66
Coreligionists are soft targets
• Insurgents and terrorists often target coreligionists: political rivals, members of rival militias, collaborators, targets of extortion.
• They seldom use the suicide tactic to do so..• .. probably because it’s not necessary. A coreligionist
assailant can defeat profiling.• Exception are target well defended by means beyond
profiling: e.g., Sadat, Massoud, Rajiv Ghandi.• When members of other religions have similar appearance
suicide attacks are not used: N. Ireland.
67
Figure 5: Strong clubs choose more suicide attacks and do more damage
Utility of Defector -weak club
Damage to Target (D)DF
Utility of LoyalMember -Suicide Attack,
E
O
D
DD DE
Utility of LoyalMember -ConventionalAttack, high p
F
C
Utility of Defector -strong club
Note: Most benign policies vs. insurgents will affect the high damage margin, so they will reduce suicide attacks.
68
Evidence on Benign Activity
and a Militia’s Potential
69
70
Suicide
SuicideSuicide Suicide
Suicide Suicide Suicide0
20
40
60
80
100
120
Att
acks
Hamas
Tanz
im
Al Aq
sa M
artyrs
PIJFa
tahPF
LP
Force
17
Terrorist Organization
Who Selects Suicide Attacks?
• Suicide attacks are so damaging that only defection proof organizations can succeed at them. Most do not try.• Nonsystematic evidence from Iraq, Afghanistan, Sri Lanka and Chechnya is consistent with suicide attacks being reserved for hard targets.• Religious radicals specialize in s. attacks in Iraq, Afghanistan, Chechnya, but evidence on soc. service provision is weak.
71
Application: Jewish Underground
• Violent militia which drew members from Gush Emumin,a messianic settler movement but a weak sect, (weak prohibitions and sacrifices).
• Began settling illegally in West Bank in mid 1970s. • - vigilante activity had local public good aspect
• After Camp David (I) agreements frustrated settlers organized conspiracy to destroy Muslim holy sites on Temple Mount / Haram A-Sharif.
• - project aborted for lack of rabbinical authorization.
72
Testable Implications: Other countries
• Richer countries are less likely to have insurgencies and civil wars (Fearon-Laitin)
73
Why the Increase in Suicide Attacks?
• .. Because of the decrease in viable options for insurgents
• Insurgents attempt conventional tactics first, including against coreligionists. When these fail they turn to suicide attacks, generally reserving them for “hard” targets.
• As governments improve at counterinsurgency we will see more terrorism and suicide attacks- directed against both local targets and allies of govt.
74
4. Mosque and State –Implications for Counterinsurgency
Activity can be rationalized, so incentives matter, which implies that subtle instruments could work.
Subtle Policies: Governments, Economic Rents and MarketsA. Improve Provision of Local Public Goods
by Secular Governments- the Kilcullen / Petraeus / SOC approach
B. Fiscally Separate Church and State if that Government is Radical Religious - so that it cannot lock itself into power
C. Reduce Rents Available to Militias and Smugglers- e.g., demand for Heroin, Cocaine and Oil
D. Improve Market provision of substitutes.
75
Once suicide attacks are being used, constructive intervention operates at that margin
76
The Kilcullen Approach
“23. Practise armed civil affairs. Counterinsurgency is armed social work; an attempt to redress basic social and political problems while being shot at. This makes civil affairs a central counterinsurgency activity, not an afterthought.”
“You need intimate cooperation with inter-agency partners here, national, international and local. You will not be able to control these partners . Many NGOs, for example, do not want to be too closely associated with you because they need to preserve their perceived neutrality.”
“Thus, there is no such thing as impartial humanitarian assistance or civil affairs in counterinsurgency. Every time you help someone, you hurt someone else . not least the insurgents. So civil and humanitarian assistance personnel will be targeted.”
Source: “Twenty-Eight Articles: Fundamentals of Company-level Counterinsurgency” (2006)Lt. Col. David Kilcullen, Ph.d. in Political Anthropology, Australian advising the Pentagon
77
DoD Policy Shifts Towards Social Science
• “Irregular warfare is about people, not platforms. IW depends not just on our military prowess, but also our understanding of such social dynamics as tribal politics, social networks, religious influences, and cultural mores. People, not platforms, and advanced technologies will be the key to IW success.”
Department of Defence, Irregular Warfare Joint Operating Concept, September 2007.
78
Nonstandard Development Economics
• Not designed to maximize growth or social welfare, but to undermine rebels
• Targeted at likely defectors and likely sources of intelligence
• Focus on programs that compete with services offered by clubs
• Benign programs will be targeted by rebels• Standard development programs can be captured by rebels
e.g., -“heros village”, LTTE controlled Sri Lanka- Sadr city garbage cleanup
Increased welfare, but reinforced rebels• Includes political development
79
5. Conclusion
• A “rational choice” economic model can explain the behavior of violent religious radicals- it succeeds on testable implications where the conventional wisdom about theological motivation fails
• That’s a relief: it provides benign options for dealing with violent religious radicals
• Those options are practical but are poorly understood and current implementation is awful
80
Conclusions• Why so few terrorist organizations?
Defection constraint.• Why are religious radicals effective terrorists?
Solve defection constraint in benign activities.Not necessary theology.
• Why suicide attacks?Hard targets.
• What to do about it? Compete in providing benign services and competent governance, with muscular protection.
• Future WorkNeed research and evaluation
Distinguish club model from standard “rational peasant”approach to “winning hearts and minds”
81
Is it Ideology?
• Well.. Ideological Shifts: • Taliban: from personal piety, local Islamic govt.
to international Jihad• Hamas: from personal piety, local Islamic govt. to
nationalist territorial struggle• Jewish Underground: from glorifying the state to
undermining it’s authority
82
Rational Choice Matters
• Can this fit in a rational choice model? - is that model helpful in predicting behavior?
• Policy implication: - can the shift to militia activity be reversed?- does behavior respond to incentives?- what could we recommend if it did not?
83
ReferencesBerman, Eli “Sect, Subsidy and Sacrifice: An Economist’s
View of Ultra-Orthodox Jews,” Quarterly Journal of Economics, 115(3) (August, 2000).
__________ “Hamas, Taliban and the Jewish Underground: An Economist’s View of Radical Religious Militias,”NBER WP 10004, (October 2003).
__________ and David Laitin, “Rational Martyrs: International Evidence on Suicide Attacks,” UC San Diego mimeo, (October 2003).
__________ and Ara Stepanyan, “How Many Radical Islamists? Evidence from Asia and Africa.” UCSD mimeo, 2003.
Iannaccone, Laurence R. “Sacrifice and Stigma: Reducing Free-riding in Cults, Communes, and Other Collectives,” Journal of Political Economy, C(1992), 271-291.
84
VI. Where Research can HelpLessons from Pacific Special Operations Command
1. Development and counterinsurgency in poorly governed spaces - now a Nat. Security concern- directly + indirectly through allies
2. USAID alone spending at ~$4B annually on this development effort, DOD spending more
3. USAID and DOD lack capability to do economic and political development in dangerous spaces. World Bank and NGOs not much better.Specifically, they lack
a) basic research on counterinsurgency & developmentb) a way to evaluate their development efforts
4 . Need: research and evaluation to guide development, governance and political violence- what DOD calls soc. sci research is mostly purchased validation of what they think is true already.
- like Great Society project, which Moynihan fixed- solution is intellectually independent research:
own core funding, university based
85
Nuclear Terrorism
• Defection constraints indicate that nuclear terrorism is very unlikely to come from most of the terrorists we see today
• Look for combination of state-backed expertise and terrorism: - Iran & Hezbollah, Pakistan (ISI) & Islamists
5. Streetlights and ViolenceEli Berman UCSD
(joint research with Jacob Shapiro PrincetonJoe Felter CTC West Point)
Results are Preliminary – Not for circulation
87
Outline
A. Two models that link public goods to violence• Clubs • Rational Peasants, Counterinsurgency &
Community Policing
B. Street Lights: Public goods and violence in Iraq
88
5.1 Clubs and Terrorism
2
N
1
Conspirators
Target
Defection – The reason so few rebel and terrorist organizationssurvive.. defection is common.
How do successful rebel and terrorist organizations prevent defection?
They have an organizational advantage.
89
Clubs and Terrorism
C Ri C Rii
N
({ }) ( )==
∏1
Operation initiated only if all Ri = 1.
Payoffs: Club extracts surplus B and shares it equally among members, who buy goods at price P.
Benign local public goods provided by govt., G and club, A.Defector’s outside option is wi, but no access to C or A.
Incentive compatibility: Member loyal iff
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
90
Clubs: Selecting Low Wage Membership Allows Larger Projects
Project Value (B)
High wagedefector
Low wagedefector
Loyalmember
B*B* B**
Incentivecompatibilitylimit with lowwage members
91
Clubs and TerrorismC Ri C Ri
i
N
({ }) ( )==
∏1
If ICC fails this is an N player prisoner’s dilemma, resulting in an unsafe route, no convoy and no rents.
Adverse selection: Imagine two unobserved types (as above), such that
ICC holds if wi = wH (> wL) for all i. A club with a costly sacrifice as an initiation rite which successfully excludes all low wage types can keep members loyal in larger projects. So religious radicals who operate mutual aid clubs have an advantage.
(ICC) U( B
N,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
92
Govt. Provision of Public Goods Reduces Militia’s Potential
(ICC) U( BN
,1, G%C(1)%A(R) ) $ U(B%wi,0 ,G)
Project Value (B)
Defector - high govt services
Loyal member -high govt. services
B** B*
Defector - low govt. servicesLoyal member -low
govt. services
Incentivecompatibility limit -high govt services
E.g., Malaya, Phillipines, Egypt
93
The Kilcullen Approach
“23. Practise armed civil affairs. Counterinsurgency is armed social work; an attempt to redress basic social and political problems while being shot at. This makes civil affairs a central counterinsurgency activity, not an afterthought.”
“You need intimate cooperation with inter-agency partners here, national, international and local. You will not be able to control these partners . Many NGOs, for example, do not want to be too closely associated with you because they need to preserve their perceived neutrality.”
“Thus, there is no such thing as impartial humanitarian assistance or civil affairs in counterinsurgency. Every time you help someone, you hurt someone else . not least the insurgents. So civil and humanitarian assistance personnel will be targeted.”
Source: “Twenty-Eight Articles: Fundamentals of Company-level Counterinsurgency” (2006)Lt. Col. David Kilcullen, Ph.d. in Political Anthropology, Australian advising the Pentagon
94
5.2 “Rational Peasant” ModelMotivation – “Hearts and Minds”
• Operating procedure of US and Allied Special Forces includes providing local public goods:
1. Control some territory2. Ask population what services they want3. Provide them4. Ask population for information about insurgents5. Use information to ambush or capture insurgents, allowing control of more
territory6. Repeat (1)-(5) until entire country is controlled
Note:a) This is less prosaic than “hearts and minds”b) Even the most disenfranchised population gets servicesc) “Rational Peasant” is due to Popkin
95
Akerloff-Yellen (94) Gang ModelMotivation
• Gangs are limited by community “norms” of behaviore.g., if the parents complain that a member is selling drugs in a primary school, the member is disciplined by the gang“…now I understand that if you ain’t got the community with you it’s just a matter of time before you got to close up shop.”- Duck (the gang member who learned his lesson)
96
A-Y model with public goods(and otherwise stripped down)
• Community: representative agent chooses whether to snitch to police: 0 = c = 1
• Gang: chooses level of crime: s = 0• Police: monitor criminals, m, and provide public
goods, g• Police move first, then gangs, then community
97
CommunityUtility from cooperating with police
Uc = (Bcsg - ac R) cwhere g = 0 is the level of govt. provided public goods, which complement public safety, and R is the expected value of gang retaliation, a constant.
It’s linear, so at the corner solution,c=0 iff ac R = Bcsg
“noncooperation constraint” ;
otherwise c=1
98
Gang
• Maximizes Ug = (Bg – Agmc) swhere Bg and Ag are positive constants,m is enforcement effort set by police, s is crime.
If Bg – Agmc = 0 when c=1, then gang will choose s so that the noncooperation constraint just binds, at s* = ac R / Bcg , so s > 0 and s decreases in g .
If Bg – Agmc > 0 when c=1, then gang chooses infinite s ..police will set m high enough to avoid that.
99
Police
• Minimize Ap s + Bp m + Cp g
.. which yields a corner solution for m = Bg / (Ag (avoiding infinite crime at c=1),
and an interior solution for g2 = (Ap ac R)/(BcCm).
Note: reducing cost of public goods Cm raises g & reduces s That could include reducing corruption in public good provision, or increasing the efficiency of taxation.
100
Back to Rational Peasant
Analogy to Rational Peasant model of insurgency, where even the disenfranchised noncombatants are favored with public goods by Special Forces.
Extensions• s and g need not be complements• Enrich g so it differs from a transfer?• If gangs can provide public goods can there be a race to the
top in public good provision?• If government can retaliate can there be a race to the
bottom in extortion?
101
Clubs or Rational Peasants?
• The two models share defection or snitching (cooperation with authorities)
• In the club model a combatant defects, while in the rational peasant model a noncombatant snitches
• Who cares which model is relevant? - predicting tactic choice by rebels:
- conventional tactics are cheaper, but share information with noncombatants (ambush, IED) while modern (club) tactics do not (suicide attack)- most organizations are not capable of club tactics because of defection constraints- targeting of police efforts- only club model relevant for domestic terrorism
102
Contents
1. Street Lights: Public services and violence2. What can we learn from CERP?3. Operational Initiative in the Philippines4. Other potential projects5. Research on Conflict: Building institutions6. Training: building in house analytical capability7. Theory and Doctrine development – upcoming
conferences8. Joint implementation plan
103
Why Statistics Can Help
6. SIGACTS (Significant Actions) in Iraq
104
6.1 Street Lights – a public good
Street lights are pretty durable, but some places improved and some got worse- These are measured on a scale of 1-bad to 3-good in a 2004 survey.
Al-Ka'im
Ana
Falluja
Haditha
Heet
Ramadi
Al-Mahawil
Al-Musayab
HashimiyaHillaAbu Ghraib
Adhamiya
Al Resafa
Al Sadr
Karkh
Khadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-Qurna
Al-Zubair
BasrahFao
Shatt Al-Arab
Amedi
Dahuk
Sumel
Zakho
Al-KhalisAl-Muqdadiya
Ba'qubaBaladroozKhanaqinKifriChoman
Erbil
KoisnjaqMakhmurMergasur
ShaqlawaSoran
Ain Al-Tamur
Al-Hindiya
Kerbala
Al-KahlaAl-Maimouna
Al-Mejar Al-Kabi
Ali Al-Gharbi
AmaraQal'at Saleh
Al-KhidhirAl-RumaithaAl-Salman
Al-Samawa
Al-ManatheraKufa
Najaf
Akre
Al-Hamdaniya
Al-Shikhan
Mosul
SinjarTelafar
Tilkaif
AfaqAl-Shamiya
Diwaniya
Hamza Al-Daur
Al-Shirqat
Baiji
BaladSamarra
Tikrit
Tooz
Chamchamal
Darbandihkan
DokanHalabja
KalarPenjwinPshdarRaniaSharbazher
SulaymaniyaAl-HawigaDaquq
Kirkuk
Al-Chibayish
Al-Rifa'i
Al-Shatra
NassriyaSuq Al-Shoyokh
Al-Hai
Al-Na'maniya
Al-SuwairaBadra
Kut
Al-Ka'im
Ana
Falluja
Haditha
Heet
Ramadi
Al-Mahawil
Al-Musayab
HashimiyaHillaAbu Ghraib
Adhamiya
Al Resafa
Al Sadr
Karkh
Khadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-Qurna
Al-Zubair
BasrahFao
Shatt Al-Arab
Amedi
Dahuk
Sumel
Zakho
Al-KhalisAl-Muqdadiya
Ba'qubaBaladroozKhanaqinKifriChoman
Erbil
KoisnjaqMakhmurMergasur
ShaqlawaSoran
Ain Al-Tamur
Al-Hindiya
Kerbala
Al-KahlaAl-Maimouna
Al-Mejar Al-Kabi
Ali Al-Gharbi
AmaraQal'at Saleh
Al-KhidhirAl-RumaithaAl-Salman
Al-Samawa
Al-ManatheraKufa
Najaf
Akre
Al-Hamdaniya
Al-Shikhan
Mosul
SinjarTelafar
Tilkaif
AfaqAl-Shamiya
Diwaniya
Hamza Al-Daur
Al-Shirqat
Baiji
BaladSamarra
Tikrit
Tooz
Chamchamal
Darbandihkan
DokanHalabja
KalarPenjwinPshdarRaniaSharbazher
SulaymaniyaAl-HawigaDaquq
Kirkuk
Al-Chibayish
Al-Rifa'i
Al-Shatra
NassriyaSuq Al-Shoyokh
Al-Hai
Al-Na'maniya
Al-SuwairaBadra
Kut
Al-Ka'im
Ana
Falluja
Haditha
Heet
Ramadi
Al-Mahawil
Al-Musayab
HashimiyaHillaAbu Ghraib
Adhamiya
Al Resafa
Al Sadr
Karkh
Khadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-Qurna
Al-Zubair
BasrahFao
Shatt Al-Arab
Amedi
Dahuk
Sumel
Zakho
Al-KhalisAl-Muqdadiya
Ba'qubaBaladroozKhanaqinKifriChoman
Erbil
KoisnjaqMakhmurMergasur
ShaqlawaSoran
Ain Al-Tamur
Al-Hindiya
Kerbala
Al-KahlaAl-Maimouna
Al-Mejar Al-Kabi
Ali Al-Gharbi
AmaraQal'at Saleh
Al-KhidhirAl-RumaithaAl-Salman
Al-Samawa
Al-ManatheraKufa
Najaf
Akre
Al-Hamdaniya
Al-Shikhan
Mosul
SinjarTelafar
Tilkaif
AfaqAl-Shamiya
Diwaniya
Hamza Al-Daur
Al-Shirqat
Baiji
BaladSamarra
Tikrit
Tooz
Chamchamal
Darbandihkan
DokanHalabja
KalarPenjwinPshdarRaniaSharbazher
SulaymaniyaAl-HawigaDaquq
Kirkuk
Al-Chibayish
Al-Rifa'i
Al-Shatra
NassriyaSuq Al-Shoyokh
Al-Hai
Al-Na'maniya
Al-SuwairaBadra
Kut
Al-Ka'im
Ana
Falluja
Haditha
Heet
Ramadi
Al-Mahawil
Al-Musayab
HashimiyaHillaAbu Ghraib
Adhamiya
Al Resafa
Al Sadr
Karkh
Khadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-Qurna
Al-Zubair
BasrahFao
Shatt Al-Arab
Amedi
Dahuk
Sumel
Zakho
Al-KhalisAl-Muqdadiya
Ba'qubaBaladroozKhanaqinKifriChoman
Erbil
KoisnjaqMakhmurMergasur
ShaqlawaSoran
Ain Al-Tamur
Al-Hindiya
Kerbala
Al-KahlaAl-Maimouna
Al-Mejar Al-Kabi
Ali Al-Gharbi
AmaraQal'at Saleh
Al-KhidhirAl-RumaithaAl-Salman
Al-Samawa
Al-ManatheraKufa
Najaf
Akre
Al-Hamdaniya
Al-Shikhan
Mosul
SinjarTelafar
Tilkaif
AfaqAl-Shamiya
Diwaniya
Hamza Al-Daur
Al-Shirqat
Baiji
BaladSamarra
Tikrit
Tooz
Chamchamal
Darbandihkan
DokanHalabja
KalarPenjwinPshdarRaniaSharbazher
SulaymaniyaAl-HawigaDaquq
Kirkuk
Al-Chibayish
Al-Rifa'i
Al-Shatra
NassriyaSuq Al-Shoyokh
Al-Hai
Al-Na'maniya
Al-SuwairaBadra
Kut
11.
52
2.5
Stre
et li
ghts
04
1 1.5 2 2.5 3Street lights 02
Street lights 04 Street lights 04
105
Who improved? st
reet
ligh
t 04
street light in Dec 20021 1.5 2
1
1.5
2
Abu Al-K
Abu Ghra
Adhamiya
Afaq Ain Al-T
Akre
Al Resaf
Al Sadr
Al-Daur
Al-Hai
Al-Hamda
Al-Hawig
Al-Hindi
Al-Ka'im
Al-Kahla
Al-KhaliAl-Khidh
Al-Mahaw
Al-Maimo
Al-Manat
Al-Mejar
Al-Midai
Al-Muqda
Al-Musay
Al-Na'ma
Al-Qurna
Al-Rifa'
Al-RumaiAl-Salma
Al-Samaw
Al-Shami
Al-Shatr
Al-ShikhAl-Shirq
Al-SuwaiAl-ZubaiAli Al-G
Amara
Amedi
Ba'quba
BadraBaiji
BaladBaladroo
Basrah
ChamchamChoman
Dahuk
Daquq
DarbandiDiwaniya
Dokan
Erbil
Falluja
Fao
Haditha
Halabja
Hamza
HashimiyHeet
Hilla Kalar
Karkh
KerbalaKhadamiy
KhanaqinKifri
Kirkuk
Koisnjaq
Kufa
Kut
Mada'in
Mahmoudi
MakhmurMergasur
Mosul
Najaf
Nassriya
Penjwin
Pshdar
Qal'at S
Ramadi
RaniaSamarra
Shaqlawa
Sharbazh
Shatt AlShekhan
Sinjar
SoranSulayman
Sumel
Suq Al-S
Tarmia
Telafar
Tikrit
TilkaifTooz
Zakho
Abu Al-K
Abu Ghra
Adhamiya
Afaq Ain Al-T
Akre
Al Resaf
Al Sadr
Al-Daur
Al-Hai
Al-Hamda
Al-Hawig
Al-Hindi
Al-Ka'im
Al-Kahla
Al-KhaliAl-Khidh
Al-Mahaw
Al-Maimo
Al-Manat
Al-Mejar
Al-Midai
Al-Muqda
Al-Musay
Al-Na'ma
Al-Qurna
Al-Rifa'
Al-RumaiAl-Salma
Al-Samaw
Al-Shami
Al-Shatr
Al-ShikhAl-Shirq
Al-SuwaiAl-ZubaiAli Al-G
Amara
Amedi
Ba'quba
BadraBaiji
BaladBaladroo
Basrah
ChamchamChoman
Dahuk
Daquq
DarbandiDiwaniya
Dokan
Erbil
Falluja
Fao
Haditha
Halabja
Hamza
HashimiyHeet
Hilla Kalar
Karkh
KerbalaKhadamiy
KhanaqinKifri
Kirkuk
Koisnjaq
Kufa
Kut
Mada'in
Mahmoudi
MakhmurMergasur
Mosul
Najaf
Nassriya
Penjwin
Pshdar
Qal'at S
Ramadi
RaniaSamarra
Shaqlawa
Sharbazh
Shatt AlShekhan
Sinjar
SoranSulayman
Sumel
Suq Al-S
Tarmia
Telafar
Tikrit
TilkaifTooz
Zakho
Abu Al-K
Abu Ghra
Adhamiya
Afaq Ain Al-T
Akre
Al Resaf
Al Sadr
Al-Daur
Al-Hai
Al-Hamda
Al-Hawig
Al-Hindi
Al-Ka'im
Al-Kahla
Al-KhaliAl-Khidh
Al-Mahaw
Al-Maimo
Al-Manat
Al-Mejar
Al-Midai
Al-Muqda
Al-Musay
Al-Na'ma
Al-Qurna
Al-Rifa'
Al-RumaiAl-Salma
Al-Samaw
Al-Shami
Al-Shatr
Al-ShikhAl-Shirq
Al-SuwaiAl-ZubaiAli Al-G
Amara
Amedi
Ba'quba
BadraBaiji
BaladBaladroo
Basrah
ChamchamChoman
Dahuk
Daquq
DarbandiDiwaniya
Dokan
Erbil
Falluja
Fao
Haditha
Halabja
Hamza
HashimiyHeet
Hilla Kalar
Karkh
KerbalaKhadamiy
KhanaqinKifri
Kirkuk
Koisnjaq
Kufa
Kut
Mada'in
Mahmoudi
MakhmurMergasur
Mosul
Najaf
Nassriya
Penjwin
Pshdar
Qal'at S
Ramadi
RaniaSamarra
Shaqlawa
Sharbazh
Shatt AlShekhan
Sinjar
SoranSulayman
Sumel
Suq Al-S
Tarmia
Telafar
Tikrit
TilkaifTooz
Zakho
Abu Al-K
Abu Ghra
Adhamiya
Afaq Ain Al-T
Akre
Al Resaf
Al Sadr
Al-Daur
Al-Hai
Al-Hamda
Al-Hawig
Al-Hindi
Al-Ka'im
Al-Kahla
Al-KhaliAl-Khidh
Al-Mahaw
Al-Maimo
Al-Manat
Al-Mejar
Al-Midai
Al-Muqda
Al-Musay
Al-Na'ma
Al-Qurna
Al-Rifa'
Al-RumaiAl-Salma
Al-Samaw
Al-Shami
Al-Shatr
Al-ShikhAl-Shirq
Al-SuwaiAl-ZubaiAli Al-G
Amara
Amedi
Ba'quba
BadraBaiji
BaladBaladroo
Basrah
ChamchamChoman
Dahuk
Daquq
DarbandiDiwaniya
Dokan
Erbil
Falluja
Fao
Haditha
Halabja
Hamza
HashimiyHeet
Hilla Kalar
Karkh
KerbalaKhadamiy
KhanaqinKifri
Kirkuk
Koisnjaq
Kufa
Kut
Mada'in
Mahmoudi
MakhmurMergasur
Mosul
Najaf
Nassriya
Penjwin
Pshdar
Qal'at S
Ramadi
RaniaSamarra
Shaqlawa
Sharbazh
Shatt AlShekhan
Sinjar
SoranSulayman
Sumel
Suq Al-S
Tarmia
Telafar
Tikrit
TilkaifTooz
Zakho
106
Street lights by themselves don’t predict much
Mean of Sig. Acts in ‘04 is 0.60 per 1000 per yearRegression slope is - 0.19 (not stat. significant)
Al-Ka'im
Ana
FallujaHadithaHeet
RamadiAl-Mahawil
Al-MusayabHashimiyaHilla
Abu Ghraib
Adhamiya Al ResafaAl Sadr
KarkhKhadamiyaMada'in
MahmoudiyaTarmia
Abu Al-KhaseebAl-MidainaAl-Qurna Al-Zubair BasrahFaoShatt Al-Arab Amedi DahukSumelZakhoAl-Khalis
Al-Muqdadiya
Ba'quba
BaladroozKhanaqinKifri
Choman ErbilKoisnjaqMakhmurMergasurShaqlawaSoranAin Al-TamurAl-Hindiya KerbalaAl-Kahla
Al-Maimouna
Al-Mejar Al-KabiAli Al-GharbiAmara
Qal'at SalehAl-KhidhirAl-RumaithaAl-Salman
Al-SamawaAl-ManatheraKufa NajafAkre
Al-Hamdaniya
Al-Shikhan
Mosul
SinjarTelafar
TilkaifAfaqAl-ShamiyaDiwaniyaHamza
Al-Daur
Al-Shirqat
Baiji
Balad
Samarra
Tikrit
ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya
Al-HawigaDaquq Kirkuk Al-ChibayishAl-Rifa'iAl-Shatra NassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniya
Al-SuwairaBadra Kut0
510
15
1 1.5 2 2.5Street lights 04
Incident per 1000 persons Fitted values
107
Multivariate regression: 2004
• All statistically significant, F=• - conditional on service provision after invasion, good services under Saddam predict violence vs.
coalition forces- conditional on service provision under Saddam, post-invasion service provision predicts less violence vs. coalition forces
05
10
15e(
p_i
nc |
X )
-.5 0 .5e( street_light | X )
coef = -2.2283984, (robust) se = .715622, t = -3.11
(conditioning on street lights 02)Incidents 04 vs Street lights 04
05
10
15e(
p_i
nc |
X )
-.5 0 .5e( street_light_02 | X )
coef = 1.95034, (robust) se = .6454984, t = 3.02
(conditioning on street lights 04)Incidents 04 vs Street lights 02
108
Street Lights and Violence: 2004, 05, 06, 07
Could it be reverse causality?
Violence could be bad for street lights.… so look at the timing:
Interpretation: Granger causality runs from lights in 02 and 04 to subsequent violence?
Tactical implication:Location of escalation was predictable based on 02 and 04 data.
010
2030
40e(
p_i
nc |
X )
-.6 -.4 -.2 0 .2 .4e( street_light | X )
coef = -2.2283984, (robust) se = .715622, t = -3.11
Incidents 2004
010
2030
40e(
p_i
nc |
X )
-.6 -.4 -.2 0 .2 .4e( street_light | X )
coef = -4.8966916, (robust) se = 1.1795619, t = -4.15
Incidents 2005
01
020
3040
e( p
_inc
| X
)
-.6 -.4 -.2 0 .2 .4e( street_light | X )
coef = -9.4834806, (robust) se = 2.3861742, t = -3.97
Incidents 2006
01
020
3040
e( p
_inc
| X
)
-.6 -.4 -.2 0 .2 .4e( street_light | X )
coef = -10.599431, (robust) se = 2.8844079, t = -3.67
Incidents 2007
(conditioning on street lights 02)Incidents vs Street Lights 04
109
Street Lights and Violence: 2004, 05, 06, 07
OLS results for Incidents per 1000 persons, 2004-2007Dependent variable:Incident per 1000 persons 2004 2005 2006 2007Street lights -2.23 -4.90 -9.48 -10.6
(0.72) (1.18) (2.39) (2.88)Street lights 02 1.95 3.56 7.56 8.70
(0.64) (1.08) (2.24) (2.81)Constant 0.98 2.89 4.89 5.07
(0.55) (1.12) (2.28) (2.14)
R squared 0.05 0.05 0.06 0.06Sample size 100 100 100 100Average incident per 1000 persons 0.63 1.13 2.38 2.60
110
Is it literally Street lights?
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Tot
al N
umbe
r of
Inci
dent
s
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Hour of the day
Mar 2006 - Dec 2007
Total Number of Incidents
111
Is it literally Street lights?OLS results for Day vs Night Incidents per 1000 persons, 2006-2007Dependent variable:
2006 2007 2006 2007Street lights -6.00 -7.71 -6.43 -6.50
(1.44) (2.07) (2.33) (2.68)Street lights X night 3.43 4.96 -0.09 0.23
(1.64) (2.24) (3.71) (4.13)Street lights 02 5.00 6.54 6.55 7.08
(1.42) (2.06) (1.81) (2.08)Street lights 02 X night -3.05 -4.49 -0.58 -1.14
(1.57) (2.20) (2.89) (3.17)Night (= 1) -1.24 -1.64 0.29 0.59
(1.52) (1.62) (1.88) (2.25)Constant 2.75 3.32 -1.43 -1.97
(1.36) (1.48) (1.30) (1.53)
R squared 0.09 0.10 0.19 0.19Sample size 200 200 170 176Average day-time incident per 1000 persons 1.46 1.81Average night-time incident per 1000 persons 0.68 0.76
Incident per 1000 persons Log of incident per 1000 persons
112
Is it literally Street lights?
Similar results if public garbage collection is used instead of 04 street lights
Similar results if road quality is used instead of 02 street lights
Implication: It looks likePast (current) public service provision predicts (reduced) violence
05
1015
e( p
_inc
| X
)
-.6 -.4 -.2 0 .2 .4e( street_light | X )
coef = -2.2283984, (robust) se = .715622, t = -3.11
(conditioning on street lights 02)Incidents 04 vs Street lights 04
05
1015
e( p
_inc
| X
)
-.5 0 .5 1e( pub_garbage | X )
coef = -.47651547, (robust) se = .30347421, t = -1.57
(conditioning on street lights 02)Incidents 04 vs Garbage collection 04
05
1015
e( p
_inc
| X
)
-.4 -.2 0 .2 .4 .6e( street_light_02 | X )
coef = 1.95034, (robust) se = .6454984, t = 3.02
(conditioning on street lights 04)Incidents 04 vs Street lights 02
05
1015
e( p
_inc
| X
)
-1 -.5 0 .5 1 1.5e( road_qual | X )
coef = .45150932, (robust) se = .12967818, t = 3.48
(conditioning on street lights 04)Incidents 04 vs Road quality 04
113
Recap: Data supports both club and rational peasant models
The next steps: • Models have an additional prediction about tactics and
geography: - In areas where noncombatants do not cooperate with government we expect conventional attacks, - while in areas where noncombatants cooperate, suicide attacks and other tactics which do not share information with noncombatants are required.
114
Where did these data come from?
• Capable of pulling classified data from military sources, declassifying it, and making it available for civilian researchers.
• In this case, the SIGACTs and survey data come from the US Central Command- don’t ask for these two datasets until we’ve examined, cleaned and written a paper
www.ctc.usma.edu
115
What can we learn from CERP?(Commanders Emergency Response Program)
• CERP is spent on public service provision, among other things
• CERP per capita and sig. acts:• But CERP is also directed disproportionately at the most troublesome areas
Al-Ka'im Al-RutbaAnaFallujaHadithaHeetRamadiAl-MahawilAl-MusayabHashimiyaHilla
Abu Ghraib
AdhamiyaAl ResafaAl Sadr KarkhKhadamiyaMada'inMahmoudiyaTarmia
Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis
Al-Muqdadiya
Ba'qubaBaladroozKhanaqinKifriChomanErbilKoisnjaqMakhmurMergasurShaqlawaSoranAin Al-TamurAl-HindiyaKerbalaAl-Kahla
Al-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-RumaithaAl-SalmanAl-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj
Al-Hamdaniya
Al-ShikhanHatra
Mosul
ShekhanSinjarTelafarTilkaifAfaqAl-ShamiyaDiwaniyaHamza
Al-Daur
Al-ShirqatBaiji
Balad
SamarraTikrit
ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniyaAl-HawigaDaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-SuwairaBadraKut
Al-Ka'im
Al-Rutba
Ana FallujaHaditha
HeetRamadi
Al-MahawilAl-MusayabHashimiyaHilla
Abu Ghraib
AdhamiyaAl ResafaAl SadrKarkhKhadamiyaMada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis
Al-Muqdadiya
Ba'qubaBaladroozKhanaqin Kifri
ChomanErbilKoisnjaqMakhmurMergasurShaqlawaSoran Ain Al-TamurAl-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-Rumaitha Al-SalmanAl-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj
Al-Hamdaniya
Al-ShikhanHatra
Mosul
ShekhanSinjarTelafar
TilkaifAfaqAl-ShamiyaDiwaniyaHamza
Al-Daur
Al-Shirqat
Baiji
Balad
Samarra
Tikrit
ToozChamchamalDarbandihkan DokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya
Al-Hawiga
DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-SuwairaBadraKut
Al-Ka'im
Al-RutbaAnaFalluja
HadithaHeet
Ramadi
Al-MahawilAl-MusayabHashimiyaHilla
Abu Ghraib
AdhamiyaAl ResafaAl Sadr
KarkhKhadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-KhaseebAl-MidainaAl-QurnaAl-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakho
Al-Khalis
Al-Muqdadiya
Ba'quba
BaladroozKhanaqin
KifriChomanErbil KoisnjaqMakhmurMergasurShaqlawaSoranAin Al-Tamur
Al-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-Rumaitha Al-SalmanAl-SamawaAl-ManatheraKufaNajaf AkreAl-Ba'aj
Al-Hamdaniya
Al-ShikhanHatra
Mosul
ShekhanSinjarTelafar
TilkaifAfaqAl-ShamiyaDiwaniyaHamza
Al-Daur
Al-Shirqat
Baiji
Balad
Samarra
Tikrit
ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya
Al-Hawiga
DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-Suwaira
BadraKutAl-Ka'im
Al-RutbaAna FallujaHaditha
HeetRamadi
Al-Mahawil
Al-MusayabHashimiyaHilla
Abu Ghraib
AdhamiyaAl ResafaAl Sadr
Karkh
Khadamiya
Mada'in
Mahmoudiya
Tarmia
Abu Al-Khaseeb
Al-MidainaAl-Qurna Al-ZubairBasrahFaoShatt Al-ArabAmediDahukSumelZakhoAl-Khalis
Al-Muqdadiya
Ba'quba
BaladroozKhanaqinKifri
ChomanErbilKoisnjaq MakhmurMergasurShaqlawaSoranAin Al-Tamur Al-HindiyaKerbalaAl-KahlaAl-MaimounaAl-Mejar Al-KabiAli Al-GharbiAmaraQal'at SalehAl-KhidhirAl-RumaithaAl-Salman
Al-SamawaAl-ManatheraKufaNajafAkreAl-Ba'aj
Al-Hamdaniya
Al-ShikhanHatra
Mosul
ShekhanSinjarTelafarTilkaifAfaqAl-Shamiya
DiwaniyaHamza
Al-Daur
Al-Shirqat
Baiji
Balad
Samarra
Tikrit
ToozChamchamalDarbandihkanDokanHalabjaKalarPenjwinPshdarRaniaSharbazherSulaymaniya
Al-Hawiga
DaquqKirkukAl-ChibayishAl-Rifa'iAl-ShatraNassriyaSuq Al-ShoyokhAl-HaiAl-Na'maniyaAl-Suwaira
BadraKut010
2030
40
0 100 200 300 400 500CERP spending per capita
Incident per 1000 persons Fitted values
116
CERP per capita and sig. acts. 06(zoom on CERP pc <100)
CERP per capita
sig acts / 1000 Linear prediction
0 50 100
0
5
10
15
20
Abu Al-K
Adhamiya
Afaq
A i n A l - T
Akre
Al ResafAl SadrAl-Ba'aj
Al-Chiba
Al-Hai
Al-Hamda
Al-Hawig
Al-Hindi
Al-Ka'im
Al-Kahla
Al-Khali
Al-Khidh
Al-Mahaw
Al-Maimo
Al-ManatAl-Mejar
Al-Midai
Al-Muqda
Al-Musay
Al-Na'maAl-Qurna Al-Rifa'Al-Rumai
Al-Rutba
Al-SamawAl-ShamiAl-Shatr
Al-Shikh
Al-ShirqAl-Suwai
Al-ZubaiA l i A l - GAmara Amedi
A n a
Ba'quba Badra
Baiji
Baladroo
Basrah
ChamchamChoman
Dahuk
Daquq
Darbandi
DiwaniyaDokan
Erbil
Falluja
Fao
Haditha
Halabja HamzaHashimiy
Hatra
HillaKalar
Karkh
Kerbala
Khadamiy
Khanaqin
Kifri
Kirkuk
Kufa Kut
Mada'in
Mahmoudi
MakhmurM e r g a s u r
Mosul
NajafNassriyaP e n j w i n
PshdarQal'at S
Ramadi
Rania
Samarra
ShaqlawaS h a r b a z h
Shatt Al
Shekhan
SinjarSoranSulayman SumelSuq Al-S
Telafar
Tikrit
Tilkaif
Tooz
Zakho
117
Can Hearts & MindsBe Bought?
• Huge investment ~ $41B • Uncertain impact• Reconstruction spending
– IRRF– CERP/CHRRP
• Community characteristics – WFP– ILCS
118
How to use CERP efficiently?
• To answer that research question we need access to more data
• In particular, we need a way to filter out the reverse causality due to CERP spending being directed at locations expected to be dangerous.
• One solution: Troop rotation data (declassified in the same way these data have been)