Professor Alan B Krueger Princeton University
Defining and Measuring Terrorism: macro-econometric evidence
Lionel Robbins Memorial Lectures
Christopher Johnson Chair
Outline of Lectures
1. Micro Evidence
2. Macro Evidence
3. Consequences
Lionel Robbins Memorial Lectures
International Terrorism:
Causes and Consequences
Alan B. Krueger
Outline
• U.S.’s International Terrorism Data
• Gov’t Bungling of Data
• Analysis of (Imperfect) Data
• Foreign Insurgents in Iraq
• Conclusions
References: “Education, Poverty and Terrorism: Is There a Causal Connection?” Journal of Economic Perspectives, with Jitka Malečková, Fall 2003.
“Kto-Kogo?: A Cross-Country Study of the Origins and Targets of Terrorism,” with David Laitin.
“Misunderestimating Terrorism: The State Department’s Big Mistake,” Foreign Affairs, with David Laitin, October 2004.
Papers on www.krueger.princeton.educ
The term ‘terrorism’ means premeditated, politically motivated
violence perpetrated against noncombatant targets by
subnational groups or clandestine agents, usually intended to
influence an audience.
The term ‘international terrorism’ means terrorism involving
citizens or the territory of more than one country.
… the term noncombatant is interpreted to include, in addition to civilians,
military personnel who at the time of the incident are unarmed and/or not
on duty. … We also consider as acts of terrorism attacks on military
installations or on armed military personnel when a state of military
hostilities does not exist at the site, such as bombings against U.S. bases in
the Persian Gulf, Europe, or elsewhere.
U.S. State Department Definition
Are we winning the war on
terrorism?
Although keeping score is difficult,
the State Department's annual
report on international terrorism,
released last month, provides the
best government data to answer this
question. The short answer is "No,"
but that's not the spin the
administration is putting on it.
"You will find in these pages clear
evidence that we are prevailing in
the fight," said Deputy Secretary of
State Richard Armitage. As
evidence, the "Patterns of Global
Terrorism" report says that
worldwide terrorism dropped by 45
percent between 2001 and 2003.
The report claims that the number
of terrorist acts committed last year
"represents the lowest annual total
of international terrorist attacks
since 1969."
Yet, a careful review of the report
and underlying data supports the
opposite conclusion: The number
of significant terrorist acts
increased from 124 in 2001 to 169
in 2003 -- 36 percent -- even using
the State Department's official
standards. The data that the report
highlights are ill-defined and
subject to manipulation -- and give
disproportionate weight to the least
important terrorist acts. The only
verifiable information in the annual
reports indicates that the number of
terrorist events has risen each year
since 2001, and in 2003 reached its
highest level in more than 20 years.
To be sure, counting terrorist acts is
not as straightforward as counting
the number of SARS victims.
Specialists have not agreed to any
test that would unambiguously
qualify an act as one of
international terrorism. But in the
words of the Congressional
Research Service, the State
Department's annual report is "the
most authoritative unclassified U.S.
government document that assesses
terrorist attacks."
So how did the report conclude that
international terrorism is declining?
It accomplishes this sleight of hand
by combining significant and
nonsignificant acts of terrorism.
Significant acts are clearly defined
and each event is listed in an
Faulty Terror Report Card May 17, 2004
Alan B. Krueger and David Laitin
__________________
● Consistent & clear
definition
● Verifiable
● Better vetting
● Career professionals first
release
● Create a statistical agency
Inspector General Report
Goodbye Appendix A
“Misunderestimating” Terrorism
Alan Krueger & David Laitin
Cross-Country Analysis
• State Department (Corrected) List
of significant terrorist events
• 1997-2003
• Negative Binomial Models
Log GDP per capita
Freedom Index
Religion
Log Population
• 149 x 149 Countries Rich data
Coded up the “Significant
Incidents” from 1997-2003
• 956 events in all
• Collected: place where
occurred; country of origin
of perpetrators; target of
attack; group responsible
or suspected; description
• Plus additional data just
on suicide attacks, incl.
domestic
Data Description
Some Issues
• Embassy (Place: country
where occurred; target:
country that owned
embassy)
• 280 events from India
(Kashmir); 104 from
Colombia; 50 in Israel
• Inconsistency: Chechnya;
multiple or single attacks
Correlations with Todd Sandler’s
Place-Based ITERATE Data
Primary
Origin Place Target
All Countries .57 .52 .23
Excl. India .89 .90 .41
No. Countries 138 139 136
17.5% more events in ITERATE data.
Country Events Events/Mil
Azerbaijan 1 0.13
Belgium 1 0.10
Eritrea 1 0.26
Germany 1 0.01
Guinea 1 0.14
Kenya 1 0.03
Nicaragua 1 0.21
Senegal 1 0.11
Thailand 1 0.02
Tunisia 1 0.11
United Arab E. 1 0.37
Zambia 1 0.10
Bahrain 2 3.11
Bangladesh 2 0.02
Chad 2 0.27
Cuba 2 0.18
El Salvador 2 0.33
Liberia 2 0.68
Macedonia 2 0.99
Morocco 2 0.07
Myanmar 2 0.04
Tanzania 2 0.06
Uzbekistan 2 0.08
Chile 3 0.20
Drc 3 0.06
Egypt 3 0.05
France 3 0.05
Kuwait 3 1.61
Peru 3 0.12
Venezuela 3 0.13
Bosnia 4 1.06
Cambodia 4 0.35
Israel 4 0.67
South Africa 4 0.10
Sudan 4 0.13
U.K. 4 0.07
Ecuador 5 0.41
Iran 5 0.08
Jordan 6 1.31
Lebanon 6 1.43
Ethiopia 7 0.11
Italy 7 0.12
Rwanda 7 0.86
Spain 7 0.18
Sri Lanka 7 0.37
Somalia 8 0.98
Yugoslavia 8 0.75
Georgia 9 1.65
Burundi 10 1.53
Uganda 10 0.48
Tajikistan 11 1.80
Indonesia 13 0.06
Algeria 14 0.47
Russia 15 0.10
Greece 16 1.52
Saudi Arabia 17 0.86
Afghanistan 18 0.72
Philippines 20 0.27
Sierra Leone 21 4.35
Iraq 26 1.17
Nigeria 26 0.21
Pakistan 26 0.20
Turkey 32 0.50
Angola 41 3.31
West Bank & Gaza 46 16.84
Yemen 49 2.95
Colombia 104 2.55
India 280 0.29
Country of Origin, No. of Events and Events per Million People
World Total: 0.16 per million people
Place, Target and Origin are Not the
Same
• P(Place=Origin) = 0.88
• P(Target = Origin) = 0.47
• P(Place = Target) = 0.53
• P(Place = Target = Origin) = 0.45
Note: Target is country most affected.
Some Initial Findings: Int’l Terrorism
• 67% of terrorist events had organizations involved
• 91% of events were carried out by multiple perpetrators.
• Religious difference in 62% of events; not very different than random people meeting
• Religious difference between perpetrator and target more prevalent for suicide attacks
• 3.5% of attacks
were on Embassies
• 12% of attacks
targeted the U.S.
• 7% of attacks were
on international
organizations
• 4.6% of attacks
were suicide
bombings
E(yij | x) = exp(x'ijβ1 +x'iβ2 + x'jβ3)
yij= Number of terrorist incidents from
country i to country j
x = Explanatory variables (e.g., income)
i = country of origin
j = target country
Estimation: Negative Binomial Model
Sample
• Drop Israel/West Bank & Gaza
• Assign Pakistan as perpetrator of Kashmir attacks
• Drop diagonal (attacks from country i on country i)
• Country must have at least 1 million people
• 149 Countries
[(149x149)-149]/2 = 11,026 pairs of countries
Main Findings for 11,026 Pairs of
Countries: Handout
1. Income of Target Matters (+) but not of Origin
2. Origin’s Civil Liberties (-); Target’s (+)
3. Lagged GDP growth (’90-’96) has no effect
4. Population of Target & Origin (+)
5. Volume of trade (-)
6. Geographic distance (-)
7. Literacy rate has no effect
8. Religion of Origin Country (Pct. Muslim and
Christian not different)
9. Occupier (+); Occupied (+)
0
1
2
3
4
5
6
Poorest 2 3 Richest
Income Quartile of Origin
Q1
Q2
Q3
Q4
Adjusted Number of Terrorist Incidents by
Income Quartile of Origin and Target
Notes: Negative Binomial regression adjusted for population, civil liberties and distance
between countries. Normalized relative to lowest income quartile.
Target:
Countries of
Origin
Richest
Poorest
0
1
2
3
4
5
6
Poorest 2 3 Richest
Income Quartile of Origin
Q1
Q2
Q3
Q4
Adjusted Number of Terrorist Incidents by
Income Quartile of Origin and Target
Notes: Negative Binomial regression adjusted for population, civil liberties and distance
between countries. Normalized relative to lowest income quartile.
Target:
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Low Medium High
Civil Rights of Origin Country
Low
Medium
High
Adjusted Number of Terrorist Incidents
by Civil Liberties of Origin and Target
Target:
Notes: Negative Binomial regression adjusted for population, income, and distance
between countries. Normalized relative to lowest civil liberties countries.
Main Findings for 11,026 Pairs of
Countries
1. Income of Target Matters (+) but not of Origin
2. Origin’s Civil Liberties (-); Target’s (+)
3. Lagged GDP growth (’90-’96) has no effect
4. Population of Target & Origin (+)
5. Volume of trade (-)
6. Geographic distance (-)
7. Literacy rate has no effect
8. Religion of Origin Country (Pct. Muslim and
Christian not different)
9. Occupier (+); Occupied (+)
Others Find Similar Results
James Piazza (2004)
-- State Dept. Data on Incidents and Casualties
-- No effect of poverty, unemployment or economic growth
Alberto Abadie (2004)
-- Insurance Risk Rating Data of Domestic
and International Terrorism
-- “[T]errorist risk is not significantly higher for poorer countries, once the effects … of political freedom are taken into account.”
-- Nonlinear effect of political freedom (robust?)
Different from Civil War.
Diyala
Kirkuk
Tikrit
Bayji
IRAN Samarra
Hit
Balad Baqubah
Kuwait City
KUWAIT
Al Kut
Ar’ar
Umm Qasr
Diwaniyah
Al Hillah
Najaf/Kufah
Sulaymaniyah
An Nukhayb
Nasiriyah Qurnah
Basrah
Al Amarah
Samawah
SYRIA
Dahuk
Karbala
Tal Afar
Habur Gate
Ramadi
Ar Rutbah
Husaybah
Baghdad
Al Asad
Al Walid
SAUDI ARABIA
Fallujah
TURKEY
Irbil Mosul
COMBINED CF/ISF AND
INDEPENDENT ISF OPERATIONS 18 OCT 05
LEGEND
COMB CF & ISF OP
INDEPENDENT ISF OP
35 OPERATIONS
10 ISF Independent
25 ISF/CF Combined
8 in Baghdad
4 ISF Independent
4 ISF/CF Combined
Captured Foreign Nationals
55 56 57
73
26 26
18
0
10
20
30
40
50
60
70
80
APR MAY JUN JUL AUG SEP OCT
Since Apr ’05: 311
For the Year: 376
EGYPT: 06 JORDAN: 04 SYRIA: 03 SUDAN: 03
EGYPT: 78 SYRIA: 66
SUDAN: 41 SAUDI: 32
JORDAN: 17 IRAN: 13
PALESTINE: 12 TUNISIA: 10
ALGERIA: 08 LIBIYA: 07
TURKEY: 06 LEBANON: 03
BRITAIN: 02 QATAR: 02
UAE: 02 INDIA: 02
DENMARK: 01 FRANCE: 01
MACEDONIA: 01 MOROCCO: 01
SOMALIA: 01 YEMEN: 01 ISRAEL: 01
INDONESIA: 01 IRELAND: 01
KUWAIT:01 USA: 01
TOTAL: 311
TOTALs BY NATION
EGYPT: 78 SYRIA: 66
SUDAN: 41 SAUDI: 32
JORDAN: 17 IRAN: 13
PALESTINE: 12 TUNISIA: 10
ALGERIA: 08 LIBIYA: 07
TURKEY: 06 LEBANON: 03
BRITAIN: 02 QATAR: 02
UAE: 02 INDIA: 02
DENMARK: 01 FRANCE: 01
MACEDONIA: 01 MOROCCO: 01
SOMALIA: 01 YEMEN: 01 ISRAEL: 01
INDONESIA: 01 IRELAND: 01
KUWAIT:01 USA: 01
TOTAL: 311
TOTALS BY NATION Captured
SUDAN: 16 SAUDI: 11 EGYPT: 09 IRAN: 05
Primarily
from SYRIA:17 EGYPT: 10 SUDAN: 05 SAUDI: 05
Primarily
from
SYRIA: 17 EGYPT: 12 JORDAN: 07 SAUDI: 07
Primarily
from
EGYPT: 36 SYRIA: 10 SUDAN: 04 SAUDI: 04
Primarily
from
SUDAN: 06 SYRIA: 05 EGYPT: 04 TURKEY: 02
Primarily
from Primarily
from
SYRIA: 09 SUDAN: 03 PALESTINE: 02 LEBANON: 02
Primarily
from
New Results: Iraq Foreign Fighters
• “They’re identified in terms of questions we asked them, interrogations we do, papers that they might have on them. But with authority, we can say they came from those countries.” Maj. Gen. Rick Lynch
• Source: MNF-I & MNC-I reports (N=311)
• April 2005 – October 2005
• Sample: 81 countries in Middle East, North Africa, Europe (26 countries had captured foreign nationals)
• Also tried limiting to 47 countries within 3,000 Km. of Baghdad
“We believe that a major piece -- not numerically but in terms of the effects of their attacks -- of the insurgency is terrorists and foreign fighters. We believe that they're coming -- a majority of them come from Syria through the Euphrates River valley into Iraq. Since April of this year, we have captured 311.” - Maj. General Rick Lynch
Explanatory Variable (1) (2)
Ln Population 0.80
(0.27)
0.78
(0.24)
Ln Distance -1.83
(0.52)
-1.29
(0.50)
Ln GDP per capita -0.15
(0.36)
1.14
(0.48)
Literacy Rate (%) --- -0.024
(0.028)
Civil Liberties --- -0.45
(0.21)
Coalition Member --- -0.40
(1.01)
Muslim (%) --- 0.023
(0.011)
N=76
Negative Binomial Estimates for Foreign Fighters
Summary Iraq Foreign Insurgent
Analysis
Origin Country Characteristics:
1. Population (+)
2. Distance to Baghdad (-)
3. GDP per capita (+)
4. Literacy is insignificant
5. Gini coeff. insig.
6. Infant Mortality (-) [knocks out income]
7. Civil Liberties (-)
8. Political rights span space as Civil Liberties
9. Coalition Member has insig. effect
10. Percent Muslim of Origin Country (+)
[A] lack of civil liberties seems to be a main cause of terrorism around the
world. Support for civil liberties should be part of the arsenal in the war
against terrorism, both at home and abroad. -- Alan Krueger,
The New York Times May 29, 2003
“The more people live under democracy, with human liberty intact, the less
inclined they or their states will be to indulge terrorism or to engage in it.
This may be open to debate – though personally I agree with it – but it
emphatically puts defeating the causes of terrorism alongside defeating the
terrorists.” -- Tony Blair, Jan. 27, 2005
Destroying Civil Liberties to Build Democracy?
• Guantanamo Bay
• Abu Ghraib
• Paul Bremer closes Al-Hawza newspaper March 28, 2004
• Lincoln Group – plants stories in Iraqi newspapers
• “Data-free” war on terrorism
• Caveat: Lack of correlation is not proof of lack of causality
• Civil liberties seem to matter
• Indirect Links to Economic Conditions: Civil war? Political freedoms?
• Terrorism should be viewed as a violent political act, not a response to economic conditions
• Many reasons to improve education and reduce poverty, but reducing terrorism probably not one of them.
Conclusion
Top Related