federal resources flowed to local police with little categorical restraints onhow they should be spent.

(and what they teach by what they do) has implications beyond policing.

expanded, deepened, and routinizedpolice involvement in the daily lives of RCS communities

the growing infrastructure of immigrationpolicing

including those with police, jails, courts, bail offices, housingauthorities, and the gamut of other street-level bureaucracies that encircle the poor.

from too much government engagement—in the forms of supervision,interference, and predation.

has been diverted from serious political analysis of policing andrelated criminal justice operations by its steady focus on national contests over electoral and policyoutcomes

officials that exercise social control and encompassvarious modes of coercion, containment, repression, surveillance, regulation, predation, discipline,and violence.

second face

black and Latino residents as suspect populations andsubjecting them to similar projects of “extractive policing” and “government seizure”

extract revenues for the municipality

through coercion,containment, repression, surveillance, regulation, predation, discipline, andviolence

some

I G U R E

In a nationof 18,000 law enforcement agencies, discussions of“policing” writ large may often mask importantheterogeneity.

mean that effects areestimated with substantially more noise.

F I G U R E

Black and White offi-cers, and (3) Hispanic and White officers, as wellas conditional Democratic–Republican comparisonswithin

Democraticand Republican officers

all instances of the number “8”appear to have been manually deleted from datesand times in the use-of-force data, requiring imputa-tion to remedy; and (3) civilian ethnicity was excludedfrom stop data despite evidence that HPD tracks thisinformation for its annual reports

if officers from differ-ent social identities do not treat civilians differently,there is little reason to suspect AR is occurring.

We usethese covariates, along with 2015–2019 five-year Amer-ican Community Survey data, to evaluate PR

However,even in the district with the highest share of Repub-lican residents, civilians are roughly 9% Republican.

In the division with the lowest share ofRepublican residents, only 2% of civilians are Repub-lican, compared to 37% of officers

T A B L E

Results show police officers diverge from their juris-dictions on every attribute we measure.

if each current officer was insteadreplaced with a random draw from their respectivejurisdiction while holding agency sizes fixed

agency level. For officer race and gender, we rely onagency responses to federal surveys, avoiding the esti-mated voter file proxies. In our behavioral analysis ofChicago and Houston, we use voter file measures ofparty identification but rely on individual-level racialdata obtained through open-records requests.

officers can provide AR fortheir partisan group without identifying or knowinglyinteracting with individual copartisans.

so stronglytied to matters of race, officers may actively repre-sent copartisans indirectly through their treatmentof various civilian racial groups

tooccur when the salience of a relevant identity increases

staffing agencies withworkers who share values with the population at largewill promote desirable outputs

bureaucratic oversight is often incapable of ensuringthat bureaucrats will exercise discretion in desirableways

diverging from partisanpreferences on policing policy in the general popula-tion

Each test compares officers deployed tocomparable places, times, and tasks, ensuring officerbehavior is always evaluated against behavior by peersfacing common circumstances

e.g., selec-tively analyzing only the subset of situations whereofficers chose to make stops or issue citations

every sin-gle district in Chicago and nearly every division inHouston is policed by officers who skew more Repub-lican than local residents

police officers are notonly more likely to affiliate with the Republican Party,they also have higher household income, vote moreoften, and are more likely to be White.

And how do officers with differ-ing partisan affiliations behave when interacting withthose civilians?

olice officers experience nosuch constraint

Our results suggest that despite Republicans’ preference formore punitive law enforcement policy and their overrepresentation in polic-ing, partisan divisions often do not translate into detectable differences inon-the-ground enforcement.

Police skew more Republican than their jurisdictions,with notable exceptions.

We need toobserve the intervening political processes—and toaccount for a wide range of alternative mechanisms

nce local demographicsstabilize, and with them residents’ expectations,diverse localities face no special barriers to raisingtaxes.

ethnically charged battles in Massachu-setts communities in this period.

Above all, the centrality ofelites in the decision to hold a tax vote emerges fromthe newspaper articles.

can transform political agendas, with aspecial impact on easily ignored, long-term issues

But the evidence shows the opposite:that increases in diversity shape the holding of votesmore than their eventual success

Preference divergenceis one such mechanism

have markedeffects on those towns’ willingness to make long-termpublic investments

But whatmany researchers have missed is the role of changes indiversity.

The results are also robust to the inclusion ofseveral economic and fiscal variables (e.g., unemploy-ment rate) and several measures of community stabil-ity (e.g., the percent of people in the same home in1985 and 1990). The results are not driven by housingprices or the change in housing prices, an alternativepathway through which diversity might shape prop-erty tax rates.

2.950

many of the most predictive variablesare economic measures such as the tax rate and themedian household income.

and included 15indicator variables denoting how long since the lastoccurrence of the dependent variable.15 Year indicatorvariables capture unobserved time effects.

This analysis shows that changing diversity—but notthe baseline level of diversity—matters in predictingdebt exclusion vote

Evidence: Diversity’s Impact

Avg. CommuteTime 90

it compares towns that pass increases to allothers, limiting the selection bias induced by thetown officials’ decision to hold the vote.

so wecannot use the election results as a straightforwardestimate of public preferences

leaders in diversecommunities will propose spending initiatives notsupported by their constituents—and so will seethem fail at the ballot box.

iverse communities’lower social capital might in turn dampen theircollective willingness to make public investments.

with opportunities toattract new voters, to gain recognition, and to ad-vance their agenda. These divisive questions can alsodivide local leaders and generate lasting cleavages,making it harder to bring together a coalition in sup-port of increased taxes.

So long as it shapesresidents’ long-term expectations, diversity couldmatter without becoming a visible topic of localpolitics

increasesin ethnic and racial diversity will mostly affect resi-dents’ support for capital projects and other long-termplans

Sociologists have demonstrated that racial andethnic considerations are often paramount when in-dividuals consider where to move.

moving decisions

Other work has shown that localdiversity shapes attitudes towards public spendingonly when race and ethnicity are highly politicized

a diverse locality might lead whites tosee public spending as having diminishing benefitsfor them

perceived

When called to the ballot box, local voters are likelyto know when the proposal will come to fruition.While current spending provides tangible benefits inthe near-term, capital investments take years, and sorequire both significant trust in the taxing authorityand a broad construction of one’s self-interest.

The second is that each ethnic group’sutility level for a given public good is reduced if othergroups also use it’

One is that different ethnic groups have differentpreferences over which type of public goods toproduce

By shaping whether towns ever considernew tax proposals, rising diversity can have a stealthimpact even absent visible and contentious localpolitical battles

the effect is strong only on votesabout long-term capital spending, suggesting thatdemographic changes operate in part by narrowingpeople’s time horizons.

willingness to hold or passmeasures that raise taxes

11 percentage points weighting towns by population.Massachusetts remains more homogeneous than Texas,but the same probability in Texas school districtsdropped by just two percentage points duringthe 1990s.

subsequent

ofeconomic decline or of urban labor markets

diverse environments generate distinctive opinions,which are then translated into policy by local leaders

If the candidate is good enough to hold his own withother voters, then the fact that he has the right kind of name maybe just the added fillip needed for success.319

ClearlytherewasnogreatdesertionofItalianDemocraticvoterstovoteforaRepublicancompatriot.‘Arethere,then,no exc

ontoadmitthatattimes,asSamuelLubell and other

Very frequently the nomination goes to the man withan “O” at the beginning or end of his name rather than to an indi-vidual who might make a good officcholder on other and perhapsmore significant grounds.

nklysaytheywillunlessthe leaders)demandsaremet?

nceofethnicrivalries,InthediscussionofNewHampshire,Tpointedouthowdevastatin

pwardsocialandeconomicmobilityofethnic‘minoritype

Democratic party.

leadership of the partyfand a general af

why substitute any moredebatable and contestable issu

ionofthesecond la

lprinciplesofthegreatMazzini, andhasourinterestsatheartbecauseheismarriedtooneofourpeople."Somepoliticiansnow claimthatLodgeoverdidthespecialappea

MereappealtotheOldSod,thevirtuesofMazziniorPulaskiisnotenough.Althoughabitmore

hepoliticalarrivaloftheItalians,thePolish,andtheFrench-CanadiansthreatenedthehegemonyoftheTrishleadershipoftheDemocraticpartyintheurbanareas,

eesas“Viewedtahistoricalperspe

Gnicfactors perse.Oveetime ethnicdistinctions anditacscseatssr'werdevay)Wigl

involved-—thereligiousdivisionsintheNewEag-faeelesosegare vottestprovlenseThaa-insomeresponstho factthat4

information is forbidden by law.

philosophic and social differences be-tween Protestantism and Catholicism,

snotuncommon:in1834aconventinChar

based as much on bigotry as land or building

building laws, lax health regulations, inefficient inspection of food andmilk, uncleaned streets, overcrowded school buildings, and unsupervisedplaygrounds”

housing developments be-came terrifying places to live, trapping people in prisons filled with vio-lence and drugs.

were the elite develop-ers, who were so powerful that city planning was often driven wholly bytheir preferences.

middle-class migrantstotheWestalsobenefitedfromreformgovernments’responsivenesstoresidentialdevelopers.

Asmentioned in chapter 1, none of these leaders could afford to rest on biasalone. They worked for reelection as all politicians do, by cultivatingareliable constituency.

symbolic benefits that cost few resources.

Daley’scity“thatworked”metthedemandsofthiscon-stituency.He focusedon streetrepairs and beautification,garbagecol-lectionandsnow removal, lowpropertytaxes and preservationofprop-ertyvalues

Irishpatronagefaroutstripped populationshares.

spending on roads and decreased spending on healthand welfare.

but represent large sums of real money.

Public good

I analyze expenditures in the categories of health andpublic welfare.'

Reformers’ core constituents werewhite, middle-class voters, and, thus, city spending on sanitation ser-vices—a major concern of homeowners—is used to represent their de-mands

broader community demands.

Narrowly Distributed Benefits

Oncethe regime collapsed, these areas again turned out at higher rates.

eripheral Core marginturnout turnout of victory turnout

bynon-corecoalitionmembers.

theseregimescouldwinreelectionwithonlyasmall shareoftheeligibleelectoratevotingthembackintooffice

.38*

monopolies focus benefits towards governing coalition elites andcore coalition members, away from the broader community.

core supporters to turn out.

Nonetheless, the relative distinc-tion between monopolized and nonmonopolized cities remains a centraltheoretical contribution of this work.

consequencesdidmonopolyhaveforthepeoplelivinginthesecities?

hallengingthoseinpowerdifficultifnotimpossible

MexicanAmerican,lackedpavedroads,streetlights,sewerage systems, andrepresentationonthe"citycouncil. SanJose’sLatinosenduredpolicebrutalityanddiscrimina.torytreatmentinthejusticesystem,theschool system,andincityhirin

were arrested and incarcer-ated at disproportionate rates, discriminated against in city hiring, andprovided with lower quality city services.

he was prepared to defend his cityat any cost from the success of such revolutionary elements.

Daley

organization

electedpositionsintoappointedos w

y abolishing districts and choos-ing at-large elections, reform charters ensured that minority preferences,even those of substantial size, remained unrepresented in the city legis-lature

Daley’s machine relied on creative district linedrawing to ensure that neighborhoods with black and Latino majoritieswere dominated by white, machine-loyal representatives

andactivelydiscouragedothers,biasingoutcomesinfavorofreformcandidates.

Elections were held when agricul-tural workers were not living in the city

Chinese work-ers in particular were targeted for restriction from social and politicallife at the state level.

Los AngelesTimes

In Austin only 37 percent of adults over the age of twenty ue had theright to vote in 1933 because of suffrage restrictions ae the Polltax and literacy test

By constructing theirideal electorate through fraud and intimidation, machines biased thesystem in favor of their incumbent organizations.

Duringyearswhenmachinecontrolwas strongest, fraud wasmuchlesscommon.

and/orlacktheinstitutionalpowertochangetherulestofavortheircandidates

excessive corruption served to undermine a machine’s pow-er if it became too offensive to voters or attracted the attention of higherlevels of government.

knewthattheyneededthemachineon theirsidetopassinspec-tions,secureutilityextensions,ignoreclosing laws,sellliquorduringtheprohibition,runlotteries,

m‘tionoftheDemocre™”>achinefaction izationwouldtranslatetoalognto voters th any other orga ;ab electing an) _ As a result election outcomes be.of benefits including patronage jobs

oercive patronage serves as a barrierto entry for challengers and a barrier to exit for voters.

incumbent can use public money to back his promise.’

benefits are awarded to a particular segment

thereby using government fesourc-es to engender loyalty to the incumbent regime and pay political workers

hischangedafterreformleaderspurchasedthepaperin1924.

thereisnolegitimatebasisfordisagreement”

promised to use public funds to help thecities grow, and growth translated into a larger market for news con-sumption.

control over the local media.

The lack of party cues to assist voters in the fo,mation of preferences resulted in systems biased in favor of candidatewith independent wealth or fame and incumbents

Nonpartisan elections also allowed reformers to build alliances actoparty lines more easily.

In converting electionsto nonpartisan contests, reformers sought to minimize divisions in theelectorate and among elites.

For some machines at least, this strategy of bias was not neces-sary to dominate elections.

bribing editors or reporters, contributing heavily in advertising funds, orby offering publishers or editors public jobs

city agencies to destroy evidence, provide extended leaves to potentialwitnesses, and otherwise prevent people from cooperating with prosecu-tors

somecoalitionsachievedcontrolprimarilythroughtheuseofgovernmentresourcesforpoliticalend

They fall alonga spectrum.

but also to advantage those who have already won.

Put simply, the decentralized neighbor-hood control of district elections may trade spatially concen-trated inequalities (new housing units) for a spatially diffuseburden (citywide housing costs)

First, we find that district elections constrain the ability ofcities to permit new housing.

substantiverepresentation for racial minorities

We find that moving to district elections signifi-cantly decreases the disparity in permitting between whiteand minority neighborhoods.

Figure 2

As before, we remove the middle tercile of data

The positive in-teraction term in both models suggests that the effect of districtelections is smaller and less predictable in cities with larger andless overrepresented majority populations.

suggesting that citieswith lower levels of segregation may experience less dramaticchange from district elections.

new equilibrium that was below their pretreat-ment levels and below their causal counterfactual.

selection into districtelections on the basis of past permitting behavior and pre-emptive changes to housing outcomes in anticipation of elec-toral reform.

we interact the treatment in-dicator with an indicator for being in the top or bottom tercileon segregation

districtit

ri 1 ht 1 zi t

housing units permitted.

that would ultimately switch to districts, that havemore than 50,000 residents, and where there is at least oneunderrepresented minority that comprises more than 20% ofthe population

there was a great deal of random chance in thetiming of treatment

Race will become less predictive of a neighbor-hood’s housing burden under district elections com-pared to at-large, all else equal.

District elections will decrease the permitting ofmultifamily housing in cities where the council ma-jority is significantly overrepresentative of that racialgroup’s population share

District elections will decrease the permitting ofmultifamily housing in cities with low majoritypopulations.

Next, existing research has found the effect of districtelections on descriptive representation to be greatest in citieswith large shares of minority residents, where majority-minoritydistricts can be more easily drawn

District elections will decrease the permitting ofmultifamily housing in residentially segregated cities.

First, district elections are more likely to improve descrip-tive representation when minorities are segregated enough toform majority-minority districts

District elections will primarily decrease thepermitting of multifamily rather than single-familyhousing.

Counterintuitively, renters may op-pose new market-rate housing not only because it harms theirquality of life but also because they believe it will attract demandto their neighborhoods, causing rents in their neighborhoods toincrease

As a result, the six-unitproject is largely insulated from political pressure that couldeither downsize or even block the proposal.

finds that a nationwide sample of cities that switchedto district elections between 1980 and 2018 experienced adecline in housing units permitted annually

they are incentivized toshift LULUs out of their districts

, generally meaningwealthier, more highly educated white voters

but pricesout those seeking to move to cities with high upward incomemobility, exacerbating long-run income inequality

unwanted housing is morelikely to be concentrated in minority neighborhoods, all elseequal

endsthe disproportionate channeling of new housing into mi-nority neighborhoods, causing cities to more equally dis-tribute new housing between their majority and minorityconstituencies.

decreases the permitting of multifamily housing

Or, they may be assignedto smaller, single-member districts, with each citizen voting foronly one candidate (district elections).

Because LULUs are perceived to threatenthe property values, safety, or general quality of life of nearbyresidents, they historically have been channeled into the po-litically weakest areas

Central to the70democratic experience is contact with difference—other races, other nationalities, other economic classes,other language groups. And, too often, the end of urbanism has undermined that experience by promotingsocial homogeneity within municipalities, leading to the evolution of regional hierarchies in which“purified communities”

It is kept primarily by an intricate, almostunconscious, network of voluntary controls and standards among the people themselves, and enforcedby the people themselves.

to be an active agent in society, and to do sowithin the relatively public setting of city life.