Crime and Cities Edward L. Glaeser Taubman Center for State and Local Government, Harvard University.

Crime and Cities

Edward L. GlaeserTaubman Center for State and Local

Government, Harvard University

Thank You For Coming

• The past week’s events in Rio’s Favela do Alemão reminds us that– despite much progress– crime in Latin America remains a major problem.

• The justification for the cause– the 19 alleged criminals killed in a violent police operation against drug dealers– reminds us that there is still much room for improvement in policing.

• This is why I am so excited about having so many terrific minds here to help us learn.

Overview of Presentation

• The Costs of Crime– Cross-Country Comparisons– Cross-City Comparisons– The Rebirth of New York

• Fighting Crime: What We Don’t Know– Social Causes of Crime– Optimal Policing and Corruption– Punishment, Incapacitation and Deterrence

The Costs of Crime: Direct Costs vs. Indirect Costs

• Direct Costs # 1: Destruction of Life and Property– Value of lives lost usually trumps other losses

• Direct Cost # 2: Public and private costs of self-protection– Self-protection and destruction of property

• Indirect Cost # 1: A Disincentive for Investment• Indirect Cost # 2: A Disamenity for a city or a

country

The Rise of the Consumer City

• The world is increasingly mobile, both within and across countries, and few places have innate production advatages– This has made consumer amenities more and

more important.– Crime destroys those consumer amenities by

being unpleasant in and of itself and by making it hard to enjoy public spaces

Cross-Country Evidence: Change in Murder and Growth

dmurder8090

growth80s Fitted values

-1.7879 5.11137

-.5894

.742268

Angola Albania

United A

Argentin

Australi

AustriaBurundi

Belgium

Burkina BangladeBahamas

Brunei D

Botswana

Canada Switzerl

Chile

China

Côte d'I

Cameroon

CongoColombia

Cyprus

GermanyDenmark

Dominica

Ecuador

EgyptSpainFinland

Fiji

France

GabonGhana

Greece

Guyana

Honduras

Hungary

Indonesi

IndiaIreland

Iceland IsraelItaly

Jamaica

Jordan

Japan

Kenya

Korea, R

Sri Lank

Lesotho

Luxembou

Morocco

Mali

Mauritiu

Malawi

Malaysia

Niger

NetherlaNorway

Nepal

New Zeal

Pakistan

Panama

Peru

PhilippiPapua Ne

Portugal

Paraguay

Rwanda

Saudi Ar

SudanSenegal

Singapor

Sweden

Swazilan

Syrian A

Thailand

Trinidad

Tunisia

Turkey

Uruguay

United S

VenezuelZambia

Zimbabwe

Rule of Law and Trade/GDP

Rule of law -- 1996-2002 -- Kauf

trade Fitted values

-1.82 2.165

13.8

318.07

Angola

United A

Argentin

Australi

Austria

Burundi

Belgium

Benin

Burkina

Banglade

Bulgaria

Bahamas

Belize

BoliviaBrazil

Bhutan

Botswana

Central Canada

Switzerl

Chile

China

Côte d'I

Cameroon

Congo

Colombia

Comoros

Cape Ver

Costa Ri

Cyprus

Germany

Djibouti

DenmarkDominica

AlgeriaEcuador EgyptSpain

Ethiopia

Finland

Fiji

France

Gabon

United K

Ghana

Guinea

Gambia

Guinea-BGreece

Guatemal

Guyana

Honduras

Haiti

Hungary

Indonesi

India

Ireland

Iran (Is

IcelandIsrael

Italy

Jamaica

Jordan

Japan

Kenya

Korea, R

Kuwait

Lao Peop

Liberia

Sri Lank

Lesotho

Luxembou

Morocco

MadagascMexico

MaliMongolia

Mozambiq

Mauritan

Mauritiu

Malawi

Malaysia

Namibia

Niger

NigeriaNicaragu

Netherla

Norway

Nepal

New Zeal

Oman

Pakistan

Panama

PeruPhilippi

Papua Ne

Poland

Portugal

ParaguayRomania

Rwanda

Saudi Ar

Sudan

Senegal

Singapor

Solomon

Sierra L

El Salva

Suriname

Sweden

Swazilan

Syrian A

Chad

Togo

ThailandTrinidad

Tunisia

Turkey

United RUganda

Uruguay

United S

Venezuel

Vanuatu

YemenSouth AfCongo, D

Zambia

Zimbabwe

Murder and Trade/GDPtr

ade

lmurder-1.95782 6.97452

13.8

318.07

Angola

United A

Argentin

Australi

Austria

Burundi

Belgium

Benin

Burkina

Banglade

Bulgaria

Bahamas

Belize

BoliviaBrazil

Botswana

Central Canada

Switzerl

Chile

China

Côte d'I

Cameroon

Congo

Colombia

Costa Ri

Cyprus

Germany

Djibouti

DenmarkDominica

Algeria EcuadorEgyptSpain

Ethiopia

Finland

Fiji

France

Gabon

Ghana

Guinea

Gambia

Guinea-BGreece

Guatemal

Guyana

Honduras

Hungary

Indonesi

India

Ireland

Iran (Is

Iceland Israel

Italy

Jamaica

Jordan

Japan

Kenya

Korea, R

Kuwait

Lao Peop

Sri Lank

Lesotho

Luxembou

Morocco

MadagascMexico

MaliMongolia

Mozambiq

Mauritan

Mauritiu

Malawi

Malaysia

Namibia

Niger

NigeriaNicaragu

Netherla

Norway

Nepal

New Zeal

Oman

Pakistan

Panama

PeruPhilippi

Papua Ne

Poland

Puerto R

Portugal

ParaguayRomania

Rwanda

Saudi Ar

Sudan

Senegal

Singapor

Solomon

Sierra L

El Salva

Suriname

Sweden

Swazilan

Syrian A

Chad

Togo

ThailandTrinidad

Tunisia

Turkey

United R Uganda

Uruguay

United S

VenezuelYemen

South AfCongo, D

Zambia

Zimbabwe

Advantages of Cross-City Work

• Fewer critical omitted variables– Think about everything else that changes at the

country level • Better and higher frequency measurement,

especially of crime.• The ability to use crimes other than murder.• The big difference: mobility of population• Population and housing price growth are the

two major measures of urban success.

Crime in 1980 and Urban Decline

Crimes (known) per 100,000 Popul

dpop8000 Fitted values

0 17930

-.39141

1.12156

Abilene

Akron ci

Albuquer

Alexandr

Allentow

Amarillo

Anaheim Anchorag

Ann Arbo

Arlingto

Arvada c

Atlanta

Aurora c

Austin c

Bakersfi

Baton Ro

Beaumont

Bellevue

Berkeley

Birmingh

Boise Ci

Boston c

Bridgepo

Brownsvi

Buffalo

Burbank

Cambridg

Charlott

Chattano

Chula Vi

Cincinna

Clarksvi

Clearwat

Clevelan

Colorado

Columbus

Concord Corpus C

Costa Me Dallas cDaly Cit

Dayton c

Denver c

Detroit

Downey c

Durham c

El Monte

El Paso

Elizabet

Erie cit

Escondid

Eugene c

Evansvil

Fayettev

Flint ci

Fort Col

Fort Lau

Fort Way

Fort Wor

Fremont

Fresno c

Fullerto

Garden G

Garland

Gary cit

Glendale

Glendale

Grand Pr

Green Ba

Greensbo

Hampton

Hartford

Hayward Hialeah

Hollywoo

Honolulu

Houston

Huntingt Huntsvil

Independ

Inglewoo

Irving c

Jackson

Jersey C

Kansas CKnoxvill

Lafayett

Lansing

Laredo c

Las Vega

Lincoln

Little R

Livonia

Long BeaLos Ange

Louisvil

Lubbock Madison

Manchest

McAllen

Memphis

Mesa cit

Mesquite

Miami ci

Milwauke

MinneapoMobile c

Modesto

Montgome

New Have

New Orle

New York

Newark c

Newport Oakland

Oceansid

OklahomaOmaha ci

Ontario

Orange cOrlando

Oxnard c

Pasadena

Pasadena

Paterson

Philadel

Phoenix

Pittsbur

Pomona c

Portland

Portsmou

Providen

Provo ci

Pueblo c

Raleigh Reno cit

Richmond

Riversid

Rocheste

SacramenSalem ci

Salinas

Salt Lak

San AntoSan Dieg

San Fran

San Jose

Santa An

Santa Cl

Santa Ro

Savannah

Scottsda

Seattle

Shrevepo

Sioux Fa

South Be

Spokane

Springfi

Springfi

St. Loui

St. PaulSt. Pete

StamfordSterling

Stockton

Sunnyval

Syracuse

Tacoma c

Tallahas

Tampa ci

Tempe ci

Toledo c

Topeka cTorrance

Tucson c

Tulsa ci

Vallejo

Virginia

Washingt

Waterbur

West Cov

Wichita

Wichita

Winston-

Yonkers

Aurora c

Baltimor

Cedar Ra

Chesapea

Chicago

Columbia

Des MoinGrand Ra

Irvine c

Joliet c

Kansas C

Lakewood

Lowell c

Norfolk

Norwalk

Overland

Peoria c

Plano ci

Rancho C

Rockford

San Bern

San BuenSimi Val

Springfi

Thousand

Waco cit

Warren c

Westmins

Worceste

Better Work on Crime and Migration: Levitt and Berry-Cullen

• A key issue in all of this work is causality: does crime cause urban decline or the reverse.

• Levitt and Berry-Cullen use ACLU suits against prisons as instruments

• These suits cause a release of criminals and an increase in crime.

• Each crime leads to one extra out-migration• Effect is five times larger for high income than

low income people.

The Exodus of the Skilled: Murder 1975 and Later Change in BAs

Murder Rate per 100,000 Populati

dba Fitted values

0 47.4

-1.38928

11.7243

Abilene

Akron ci

Albuquer

Alexandr

Allentow

Amarillo

Anaheim Anchorag

Ann Arbo

ArlingtoArvada c

Atlanta

Aurora c

Austin c

Bakersfi

Baton Ro

Beaumont

Bellevue

Berkeley

Birmingh

Boise Ci

Boston c

Bridgepo

Brownsvi

Buffalo

Burbank

Cambridg

Charlott

Chattano

Chula Vi

Cincinna

Clearwat

Clevelan

Colorado

Columbus

Concord

Corona c

Corpus C

Costa Me

Dallas c

Daly Cit

Dayton c

Denver c

Detroit

Downey c

Durham c

El MonteEl Paso

Elizabet

Erie cit

Escondid

Eugene c

Evansvil

Fayettev

Flint ci

Fort Col

Fort Lau

Fort Way

Fort Wor

Fremont

Fresno cFullertoGarden G

Garland

Gary cit

Glendale

GlendaleGrand Pr

Green Ba

Greensbo

Hampton

Hartford

Hayward

Hialeah

Hollywoo

Honolulu

Houston

Huntingt

Huntsvil

Independ Inglewoo

Irving c

Jackson

Jersey C

Kansas C

Knoxvill

Lafayett

Lansing

Laredo c

Las Vega

Lincoln

Little RLivonia

Long Bea

Los Ange

LouisvilLubbock

Madison

Manchest

McAllen Memphis Mesa cit

Mesquite

Miami ci

Milwauke

Minneapo

Mobile c

Modesto

Montgome

New Have

New Orle

New York

Newark cNewport

Oakland

Oceansid

Oklahoma

Omaha ci

Ontario

Orange cOrlando

Oxnard c

Pasadena

Pasadena

Paterson

Philadel

Phoenix

Pittsbur

Pomona c

Portland

Portsmou

Providen

Provo ci

Pueblo c

Raleigh

Reno cit

Richmond

Riversid

RochesteSacramen

Salem ciSalinas

Salt Lak

San Anto

San Dieg

San Fran

San Jose

Santa An

Santa Cl

Santa Ro

Savannah

Scottsda

Seattle

Shrevepo

Sioux FaSouth Be

Spokane Springfi

Springfi

St. Loui

St. Paul

St. Pete

Stamford

Sterling

Stockton

Sunnyval

Syracuse

Tacoma c

TallahasTampa ci

Tempe ci

Toledo c

Topeka c

Torrance

Tucson c

Tulsa ci

Vallejo

Vancouve

Virginia

Washingt

Waterbur

West Cov

Wichita

Wichita

Winston-

Yonkers

Aurora c

Baltimor

Carrollt

Cedar RaChesapea

Chicago Columbia

Des Moin

Grand Ra

Irvine c

Joliet c

Kansas C

Lakewood

Lowell c

Napervil

Norfolk

North La

Norwalk

Overland

Peoria c

Plano ci

Rockford

San Bern

Simi Val

Springfi

Waco citWarren c

Worceste

City Growth and Crime in Brazil: Henderson and Others (2005)

• Henderson and co-authors look at Brazilian city growth between 1970 and 2000.

• They use a panel and find that a ten percent increase in homicides is associated with a 1.1 percent decrease in population growth over each five year period.

• The ability of crime to hurt cities is not limited to the U.S.

Amenities and City Growth

• Urban success is ultimately driven by the demand of individuals to live in that location.

• Productivity (Income) and Amenities are thought to be the two drivers.

• As people have gotten richer and as transport costs have declined, amenities have gotten more important.

• Amenity Measure: Prices that are higher than they should be controlling for income.

Amenity Index: Housing Prices Controlling for Income Levels

Highest Lowest

Honolulu, HI Stamford, CTSanta Cruz, CA Norwalk, CTSanta Barbara-Santa Maria-Lompoc, CA Anchorage, AKSalinas-Seaside-Monterey, CA Rochester, MNLos Angeles-Long Beach, CA Detroit, MISan Francisco, CA Midland, TXSan Jose, CA Trenton, NJSanta Rosa-Petaluma, CA Minneapolis-St. Paul, MNOxnard-Ventura, CA Nassau-Suffolk, NYSan Diego, CA Bloomington-Normal, IL

Source:

Table 4:United States MSA with Highest and Lowest Estimated Amenity Values

Note: Estimated Amenity Value measured as residual from an OLS regression of log median house value on log median income in 1990.

Originally appeared in Glaeser, Edward L., Jed Kolko, and Albert Saiz. "Consumer City." Journal of Economic Geography, Oxford University Press 1.1 (2001): 27-50.

Amenity Index and Population Growth

Figure 11: 1980-2000 Population Growth on Amenity IndexAmenity Index

1980-2000 Change in Population Fitted values

-.5 0 .5 1

-.5

0

.5

1

BeaumontSaginaw-

Detroit-Kalamazo

Johnstow

Grand RaIndianap

Fort WayCorpus C

YoungstoUtica-Ro

RochesteLansing-Syracuse

Duluth-S

Louisvil

Omaha, NMacon, G

McAllen-

San Anto

Rockford

Greenvil

Reading,

Jacksonv

St. Loui

Hickory-

Scranton

ChattanoEvansvil

Kansas CWichita,

Columbus

Buffalo-

SpringfiPhiladelAlbany-S

Augusta-

Allentow

Peoria-P

Appleton

Charlott

Pittsbur

Pensacol

Erie, PA

Biloxi-G

Canton-M

Davenpor

Harrisbu

Dayton-SToledo,

Greensbo

Shrevepo

OklahomaYork, PA

Binghamt

Richmond

Johnson

Memphis,Des Moin

Birmingh

Little R

Lakeland

Tulsa, O

Atlanta,

Montgome

Huntingt

CincinnaMobile,

Dallas-F

KnoxvillLancaste

Houston-

Columbia

Nashvill

ColumbusJackson,Charlest

Lafayett

El Paso,

Norfolk-

Charlest

Boston-W

Clevelan

Raleigh-

Orlando,

Providen

Melbourn

Tampa-St

Baton RoSpokane,

Milwauke

Minneapo

Austin-S

Lexingto

Chicago-Hartford

Boise Ci

Daytona

New Orle

Salt Lak

New York

Bakersfi

Madison,Albuquer

Phoenix-West Pal

Sarasota

Denver-BPortland

Washingt

StocktonColorado

New Have

Seattle-Miami-Fo

Tucson,

Las Vega

Modesto,

Eugene-S

SacramenFresno,

Los AngeSalinas,

San Dieg

Santa BaSan Fran

Honolulu

Crime as a Disamenity

• The hedonic literature has long documented the negative impact of crime on property values.

• Thaler (1975) is an early paper showing this connection.

• A vast number of studies have confirmed that higher crime rates are associated with lower property values (and hence amenities) within and across urban areas.

Housing Prices and Crime

1999 Crime Rate

median value of all owner-occup Fitted values

2944 13998

39577

422700

Albuquer

Anaheim

Anchorag

Arlingto

Atlanta Aurora c

Austin c

Boston c

Buffalo

Charlott

Cincinna

Clevelan

Colorado

Columbus

Corpus C

Dallas c

Denver c

Detroit El Paso Fort Wor

Fresno c

Honolulu

Houston Kansas C

Las Vega

Long Bea

Los Ange

LouisvilMemphis

Mesa cit Miami ci

Milwauke

Minneapo

New Orle

New York

Newark c

Oakland

OklahomaOmaha ci

Philadel

Phoenix

Pittsbur

PortlandRaleigh Riversid

Sacramen

San Anto

San Dieg

San Fran

San Jose

Santa An

Seattle

St. Loui

St. Paul

Tampa ciToledo c

Tucson cTulsa ci

Virginia

Washingt

Wichita

Chicago

Home Value Change and Crime Rate

Log Ch. Crime Rate 1980-2000

Log Ch. Home Values 1980-2000 Fitted values

-1.05487 .337192

-.280652

1.02699

Albuquer

Atlanta

Austin c

Birmingh

Boston c

Buffalo

Charlott

CincinnaClevelan

Columbus

Dallas c

Denver c

Detroit

El Paso

Fort Wor Honolulu

Houston

Kansas CLong Bea

Los Ange

Louisvil

Memphis

Miami ci

Milwauke

Minneapo

New Orle

New YorkNewark c

Oakland

Oklahoma

Omaha ciPhiladel

Phoenix Pittsbur

Portland

Sacramen

San AntoSan Dieg

San FranSan Jose Seattle

St. Loui

St. Paul

Tampa ci

Toledo c

Tucson cTulsa ci

Virginia

Washingt

Wichita

Chicago

Revival of NYC and Other US Cities• In the 1970s, almost all of the older American

cities looked like they were in trouble.• Economic decline, fiscal distress, and high crime

rates made them look like dinosaurs.• Since 1975, the high skilled cities have done quite

well, even in their downtowns.• One part of that is the rise of the older city as

consumer city (reverse commuting)• Crime can explain about 1/3 of NYC housing price

appreciation (Schwartz et al.)

Cities and Skills in the Colder Regionscoef = .01169126, se = .00161765, t = 7.23

e(

dp

op

| X

)

e( bagrad90 | X )-10.4384 24.0572

-.310423

.311965

Steubenv

Johnstow

Altoona,

Lima, OH

Mansfiel

Youngsto

Williams

LewistonJackson,

Joplin,

Janesvil

Wausau,

St. Jose

Kokomo,

Sharon, Scranton

Sheboyga

Canton--

York, PA

Jamestow

Elkhart-

Decatur,

Evansvil

Reading,

Saginaw-Terre Ha

Elmira,

Glens FaRockford

Utica--R

Sioux Ci

Fort Way

Erie, PA

Muncie,

Lancaste

Benton HDubuque,

Duluth--

Eau Clai

Peoria--

St. ClouAppleton

Waterloo

Allentow

Davenpor

Toledo,

Green Ba

Bangor,

Grand Ra

Harrisbu

Detroit-

Springfi

Pittsbur

Clevelan

Buffalo-

Dayton--

South Be

CincinnaLa Cross

Binghamt

Indianap

Kalamazo

Providen

St. Loui

Sioux Fa

Milwauke

Syracuse

Springfi

Pittsfie

Rapid Ci

Grand Fo

Wichita,

Cedar Ra

New LondSpringfi

Bismarck

PhiladelTopeka,

Omaha, N

Des Moin

Rocheste

Kansas C

Columbus

Chicago-

Albany--Lansing-

Fargo--M

New York

Lafayett

Hartford

Minneapo

Portland

Lincoln,

Boston--

Barnstab

BloomingBurlingt

New Have

Rocheste

State CoBlooming

Champaig

Madison,Columbia

Lawrence

Iowa Cit

Crime and NYC: The Long Haul

Crime in NYC and Chicago since 1980

Homicide Rate and Real Estate Prices in NYC 1970 - 2000

0

5

10

15

20

25

30

35

1970 1975 1980 1985 1990 1995

Ho

mic

ides

per

100

,000

Res

iden

ts

0.5

0.6

0.7

0.8

0.9

1

Rea

l Est

ate

Pri

ces

(In

dex

)

Economic and Social Roots of Crime

• What underlying societal factors explain the level of crime?

• Across countries, inequality seems important (Fajznlberger, Loayza and Lederman)– Overall wealth is less important

• Across cities, unemployment matters (Witte)• Urban size strongly predicts crime• Social multipliers seems to create crime waves

Inequality and Crime: Fajnzlberger, Lederman and Loayza (JLE)

Crime and Income: a weaker relationship

lngdppop80

lmurder Fitted values

4.99726 10.5927

-1.95782

6.97452

Angola

Albania

United A

Argentin

AustraliAustria

Burundi

BelgiumBenin

Burkina

Banglade

Bulgaria

BahamasBelizeBolivia

Brazil

Brunei D

Botswana

Central

Canada

Switzerl

Chile

China

Côte d'I

Cameroon

Congo

Colombia

Costa Ri Cyprus

GermanyDenmark

Dominica

Algeria

Ecuador

Egypt

Spain

Estonia

FinlandFiji France

Gabon

Georgia

Ghana

Gambia

Guinea-B

Greece

Guatemal

Guyana

Honduras

Hungary

Indonesi

India

Ireland

Iran (Is

Iceland

Israel

Italy

Jamaica

Jordan

Japan

Kenya

Korea, R

KuwaitSri Lank

Lesotho

Luxembou

Latvia

Morocco

Republic

Madagasc

Mexico

Mali

Mozambiq

Mauritan Mauritiu

MalawiMalaysia

Namibia

Niger

Nigeria

Nicaragu

Netherla

Norway

NepalNew ZealOman

Pakistan

Panama

Peru

Philippi

Papua Ne Puerto R

Portugal

Paraguay

Russian

Rwanda

Saudi Ar

Sudan

SenegalSingaporSolomon

Sierra L

El Salva

SurinameSweden

Swazilan

Syrian A

Chad

Togo

Thailand

Trinidad

Tunisia

Turkey

Uruguay

United SVenezuel

South Af

Congo, D

ZambiaZimbabwe

Cross Cities: Crime and Unemployment

Unemployment Rate, 2000

1999 Crime Rate Fitted values

1.7 12.9

2878

13998

Akron ci

Albuquer

Anaheim

Anchorag

Arlingto

Atlanta

Aurora c

Austin c

Bakersfi

Baton Ro

Birmingh

Boston c

Buffalo

Charlott

CincinnaClevelan

Colorado

Columbus

Corpus C

Dallas c

Denver c

Detroit

El Paso

Fort Way

Fort Wor

Fremont

Fresno c

Garland

Glendale

Greensbo

Hialeah

Honolulu

Houston

Jersey C

Kansas C

Las Vega

Lincoln

Long BeaLos Ange

Louisvil

Madison

Memphis

Mesa cit

Miami ci

Milwauke

Minneapo

Montgome New Orle

New York

Newark c

Oakland

Oklahoma

Omaha ciPhiladel

Phoenix

Pittsbur

Portland

Raleigh

Riversid

Rocheste

SacramenSan Anto

San Dieg

San Fran

San Jose

Santa An

Scottsda

Seattle

Shrevepo

St. Loui

St. Paul

St. Pete

Stockton

Tampa ci

Toledo c

Tucson c

Tulsa ci

Virginia

Washingt

Wichita

Chicago

Plano ci

Crime and Schooling

% persons 25+ with HS or higher,

SERIOUS CRIMES PER 100,000 POPU Fitted values

44.3 95.1

2829

18953

Abilene

Akron ci

Albuquer

Alexandr

Allentow

Amarillo

Anaheim Anchorag

Ann Arbo

Arlingto

Arvada c

Atlanta

Aurora c

Austin c

Bakersfi

Baton Ro

Beaumont

Bellevue

BerkeleyBirmingh

Boise Ci

Boston c

Bridgepo

Brownsvi Buffalo

Burbank

Cambridg

CharlottChattano

Chula Vi

Cincinna

Clarksvi

ClearwatClevelan

Colorado

Columbus

Concord

Corona c

Corpus C

Costa Me

Dallas c

Daly Cit

Dayton c

Denver c

Detroit

Downey c

Durham c

El Monte

El Paso

Elizabet

Erie cit

Escondid Eugene c

Evansvil

FayettevFlint ci

Fort Col

Fort Lau

Fort Way

Fort Wor

Fremont

Fresno c

FullertoGarden GGarland

Gary cit

Glendale

Glendale

Grand Pr

Green Ba

Greensbo

Hampton

Hartford

Hayward

Hialeah

Hollywoo

Honolulu

Houston

Huntingt

Huntsvil

Independ

Inglewoo Irving c

Jackson

Jersey C

Kansas C

Knoxvill Lafayett

Lansing

Laredo c

Las VegaLincoln

Little R

Livonia

Long BeaLos Ange

Louisvil Lubbock Madison Manchest

McAllen Memphis

Mesa citMesquite

Miami ci

Milwauke

Minneapo

Mobile c

Modesto

Montgome

New Have

New Orle

New York

Newark c

Newport

Oakland

Oceansid

Oklahoma

Omaha ci

Ontario

Orange cOrlando Oxnard cPasadena

Pasadena

Paterson

Philadel

Phoenix

PittsburPomona c

Portland

PortsmouProviden

Provo ci

Pueblo c Raleigh Reno cit

Richmond

Riversid

Rocheste

Sacramen

Salem ci

Salinas

Salt LakSan Anto

San Dieg

San Fran

San Jose

Santa An

Santa ClSanta Ro

Savannah

Scottsda

Seattle

Shrevepo

Sioux FaSouth Be

Spokane

Springfi

Springfi

St. Loui

St. Paul

St. Pete

Stamford

Sterling

Stockton

Sunnyval

Syracuse

Tacoma c

Tallahas

Tampa ci

Tempe ci

Toledo c

Topeka c

Torrance

Tucson c

Tulsa ci

Vallejo

Virginia

Washingt

Waterbur

West Cov

Wichita Wichita

Winston-

Yonkers

Baltimor

Cape Cor

CarrolltChandler

Chesapea

Columbia

Coral Sp

Fontana

Grand Ra

HendersoIrvine c

Kansas C

Lakewood

Lancaste

LexingtoMoreno V

Norfolk

Norwalk Overland

Palmdale

PembrokePeoria c

Plano ci

Port St.

Rancho C

San Buen

Santa ClSimi ValThousand

Waco cit

Warren c

West Val

Westmins

Crime and City Size

lpop80

Crimes (known) per 100,000 Popu Fitted values

10.1008 15.7716

0

17930

Abilene

Akron ci

AlbuquerAlexandr

AllentowAmarillo

Anaheim Anchorag

Ann Arbo

Arlingto

Arvada c

Atlanta

Aurora c

Austin c

BakersfiBaton Ro

Beaumont

Bellevue

Berkeley

Birmingh

Boise Ci

Boston c

Bridgepo

Brownsvi

Buffalo

Burbank

Cambridg Charlott

Chattano

Chula Vi

Cincinna

Clarksvi

Clearwat

Clevelan

Colorado

Columbus

Concord Corona c

Corpus CCosta Me

Dallas c

Daly Cit

Dayton c

Denver c Detroit

Downey c

Durham c

El Monte

El Paso

Elizabet

Erie cit

Escondid

Eugene c

Evansvil

Fayettev

Flint ci

Fort Col

Fort Lau

Fort Way

Fort Wor

Fremont

Fresno c

Fullerto

Garden G

Garland

Gary cit

GlendaleGlendale

Grand Pr

Green Ba

Greensbo

Hampton

Hartford

Hayward

Hialeah

Hollywoo

Honolulu Houston

Huntingt

Huntsvil

Independ

Inglewoo

Irving c

Jackson Jersey C

Kansas C

Knoxvill

LafayettLansing

Laredo c

Las Vega

Lincoln

Little R

Livonia

Long Bea

Los Ange

Louisvil

Lubbock Madison

Manchest

McAllen Memphis

Mesa citMesquite

Miami ci

Milwauke

MinneapoMobile c

Modesto

Montgome

New Have

New Orle

New York

Newark c

Newport

Oakland

Oceansid

Oklahoma

Omaha ci

Ontario

Orange c

Orlando

Oxnard c

Pasadena

Pasadena

Paterson

Philadel

Phoenix

Pittsbur

Pomona c

Portland

Portsmou

Providen

Provo ci

Pueblo cRaleigh

Reno cit

Richmond

Riversid

Rocheste

Sacramen

Salem ci

Salinas

Salt Lak

San AntoSan Dieg

San Fran

San Jose

Santa An

Santa ClSanta Ro

Savannah

Scottsda

Seattle

Shrevepo

Sioux Fa

South Be

Spokane Springfi

Springfi

St. Loui

St. PaulSt. Pete

Stamford

Sterling

Stockton

Sunnyval

Syracuse

Tacoma c

Tallahas

Tampa ci

Tempe ciToledo c

Topeka c

Torrance

Tucson c

Tulsa ci

Vallejo Vancouve

Virginia

Washingt

Waterbur

West CovWichita

Wichita Winston-

Yonkers

Aurora c

Baltimor

Cape Cor

Carrollt Cedar RaChandler

Chesapea

Chicago

Columbia

Coral Sp

Des MoinFontana

Grand Ra

Irvine c

Joliet c

Kansas C

Lakewood

Lancaste

Lowell c

Napervil

Norfolk

North La

Norwalk Overland

Pembroke

Peoria c

Plano ci

Rancho C

Rockford

San Bern

San Buen

Simi Val

Springfi

Thousand

Waco citWarren c

Westmins

Worceste

Why is there more crime in cities?• Cities have more available victims (about 1/5 of

the effect)– Distance between criminal and victim declines

• Enforcement is harder in cities (about ¼)– The poor of suspects is much larger

• Cities attract people who are crime-prone (about ½ of the relationship in the U.S.)– Cities have good amenities for the poor, like public

transportation– Cities might particularly have more weaker traditional

rules

Crime and Social Interactions• For many reasons crime appears to be

contagious. • Overcrowding of police (riots)• Transfer of crime-related human capital

(evidence from prisons by Bayer)• Legitimization of criminal activity• Standard peer effects work (Case and Katz)• High variance of crime rates (GSS)• If so, then there is a social multiplier in anti-crime

activities

Fighting Crime: Social Policy vs. Crime Policy

• The connection between social distress and crime does suggest that social policy has some role to play in fighting crime.

• Donahue and Siegelman review a series of interventions (like head start) and argue that some are cost effective relative to prisons (which cost around 40k/year).

• But there are many reasons to doubt the social policy will be effective on its own.

The Limitations of Social Policy

• The ability to create widespread social change is quite limited and surely takes decades.

• By contrast, many cities have been able to change its crime policy over a few years.

• Moreover, the elasticities of crime with respect to poverty are not all that strong.

• Without increasing the costs of crime, there is unlikely to be a major reduction in crime.

Crime and Punishment

• The economic approach to criminal policy has emphasized deterrence and incapacitation. – Little faith has been put in rehabilitation both

because of theory and high recidivism rates.

• The traditional theory (Becker, 1968) has particularly emphasized deterrence and has called for more fines than prisons.

• More recent work (Levitt) has emphasized incapacitation (prisons and abortion).

Crime and Incarceration

• Typical estimates of the elasticity of crime with respect to incarceration rates run from .15-.25.

• Levitt’s work using ACLU-forced prison releases suggests higher estimates (.3).

• The time series of incarceration certainly does seem to support the view that locking people up is effective, but at a terrible cost.

Two Graphs from Levitt

Incarceration and Homicides in Sao Paulo

Prisoners and Criminal Homicide in the State of São Paulo

0

20

40

60

80

100

120

140

160

180

1994 1996 1998 2000 2002 2004 2006 2008

Source: Secretaria de Segurança Pública

20

00

= 1

00

Prisoners

Criminal Homicides

Theory on Incapacitation and Deterrence

• The relative importance of incapacitation and deterrence depends mainly on the specialization of the criminal.

• If criminals are specialized, the punishment should work through incapacitation.

• When they are amateurs, then deterrence is more important.

• Could the division of labor explain the 19th century rise in prisons (Foucault).

Application to Latin America

• The lack of national, systematic victimization surveys and crime report statistics compromise clearance rate estimates.

• Still the available data shows that homicides clearance rates in Rio de Janeiro are about 2.7% and about 12% in São Paulo (Piquet)

• Compared to 64% in the USA in 2002.

Can the difference in punishment explain the difference in crime?

• If we take an average clearance rate of 7.5% for Rio and Sao Paulo vs. 64% in the U.S.

• A typical large U.S. city might have about 15 murders per 100,000

• With an elasticity of .3, this predicts that the murder rate in the Brazilian cities should be about double the murder rate in the U.S.

• But the actual murder rate is about closer to four times as high

Reconciling the Figures

• Theory # 1: there is a lot more other than clearance rates that makes Brazil have a higher crime problem (inequality, etc.).

• There are many U.S. with a homicide rate that is far more than ½ of the rate in Brazil.

• Theory # 2: the crime elasticities are higher for more professional types of murders, and much of the difference between the regions reflects those more professional murders.

Effective Policing

• To economists, who know little about the business of policing, crime policy tends to come down to the probability of arrest and the length of sentence.

• To policing experts, much more matters and they are surely right.– Better Information Technology– Community policing– Corruption

Information Technology and Crime

• There is a long history of police using new information technology to fight crime– Dial 911 was a particularly dramatic innovation

• In the 1990s, information became used both to target responses and to evaluate precinct commanders (CompStat)

• Incentives tied to actual crime• NYPD Real Time Crime Center (opens in 2005)

The Rise of Community Policing

• Boston and NYC both had significant crime reductions in the 1990s, but followed different crime strategies (Chris Stone).

• NYC was police-centered; Boston focused on community partnership (Ten Point Coalition).

• Basic idea is to leverage police by using community resources.

• Reduces crime while building trust– but is it sustainable or transferable?

Corruption, Violence and Law Enforcement

• In 2004, 983 people were killed by the police at the State of Rio de Janeiro.

• This number is twice the number of justifiable homicides perpetrated by all US police departments together (about 450 a year according to NBJS).

Violence against Police

• Police officers are also at severe risk of being killed in the line of duty: in this same year (2004),

• 111 police officers were killed at Rio de Janeiro and 27 in São Paulo from a total of 50 thousand police officers in Rio and 120 thousand in São Paulo.

• At the US, 54 officers were killed in 2004 from approximately 796 thousand.

Corruption in Brazil

• From 1996 to 2006, 5567 police officers were placed under investigation and 4923 were dismissed from their positions.

• Officers dismissed from Sao Paulo over time:Year Number of Officers

Dismissed2000 3522001 4312002 5262003 8782004 4972005 427

Corruption and Murders

Corruption -- 1996-2002 -- Kaufm

lmurder Fitted values

-1.5625 2.39

-1.95782

6.97452

Angola

Albania

United A

ArgentinArmenia

AustraliAustria

Azerbaij

Burundi

BelgiumBenin

Burkina

Banglade

Bulgaria

Bahamas

Bosnia H

Belarus

BelizeBolivia

Brazil

Brunei D

Botswana

Central

Canada

Switzerl

Chile

China

Côte d'I

Cameroon

Congo

Colombia

Costa Ri Cyprus

Czech Re Germany

DjiboutiDenmark

Dominica

Algeria

Ecuador

EgyptEritrea

Spain

EstoniaEthiopia

FinlandFiji France

Gabon

Georgia

Ghana

GuineaGambia

Guinea-B

Greece

Guatemal

Guyana

Honduras

Croatia

Hungary

Indonesi

India

Ireland

Iran (Is

Iceland

Israel

Italy

Jamaica

Jordan

Japan

Kazaksta

Kenya

Kyrgyzst

Cambodia

Korea, R

Kuwait

Lao Peop

LebanonSri Lank

Lesotho

Lithuani

Luxembou

Latvia

Morocco

Republic

Madagasc

Mexico

The form

Mali

Mongolia

Mozambiq

Mauritan Mauritiu

MalawiMalaysia

Namibia

Niger

Nigeria

Nicaragu

Netherla

Norway

NepalNew ZealOman

Pakistan

Panama

Peru

Philippi

Papua Ne

Poland

Portugal

Paraguay

Romania

Russian

Rwanda

Saudi Ar

Sudan

SenegalSingaporSolomon

Sierra L

El Salva

Sao TomeSuriname

Slovakia

Slovenia Sweden

Swazilan

Syrian A

Chad

Togo

Thailand

Trinidad

Tunisia

Turkey

United R

Uganda

Ukraine

Uruguay

United S

Uzbekist

Venezuel

Viet Nam

Yemen

South Af

Congo, D

ZambiaZimbabwe

Improving Police Work