You are on page 1of 20

Informal Institutions and the Rule of Law: The Judicial Response to State Killings in Buenos

Aires and São Paulo in the 1990s


Author(s): Daniel M. Brinks
Source: Comparative Politics, Vol. 36, No. 1 (Oct., 2003), pp. 1-19
Published by: Comparative Politics, Ph.D. Programs in Political Science, City University of New
York
Stable URL: http://www.jstor.org/stable/4150157
Accessed: 02-12-2015 18:03 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Comparative Politics, Ph.D. Programs in Political Science, City University of New York is collaborating with JSTOR to
digitize, preserve and extend access to Comparative Politics.

http://www.jstor.org

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Informal Institutions and the Rule of Law

The JudicialResponseto StateKillingsin BuenosAiresandSao


Pauloin the 1990s

Daniel M. Brinks

While its very persistence in the face of multifaceted crises is no small achievement,
LatinAmericandemocracy in the last two decades has been criticized as incomplete,
unconsolidated, and lacking in the rule of law.1 The region's legal systems have
played a large part in this disillusionment for condoning corruption,favoritism, and
the continuation of repression. O'Donnell captures this failure of universalistic
democratic citizenship with his metaphor of "brown areas."2One aspect of these
brown areas is the inability,to differentdegrees, of the justice systems in Brazil and
Argentinato protectcertainbasic rights, as evidenced by the failure of their courts to
punish the large numberof police homicides that continue to be committed.
The problem is serious. In 1992 alone the Sao Paulo police killed more people
than the last military dictatorshipdid in the entire country in all its years of rule.3Yet
a policeman who kills someone in the course of routine policing in Sao Paulo has a
94 percent chance of escapingjudicial sanction. Moreover,the complete answermay
not be found in improvingthe organizationalcapacities of judicial institutions.Tens
of millions of dollars have been spent on judicial reform programs to address the
perceived weakness of judiciaries all over Latin America. Yet the consensus is that
the monies spent on judicial reform have not led to a significant improvement in
judicial services, and rights violations continue despite the proliferation of legal
norms that purport to outlaw them.4 This paradox is made evident in the judicial
response to state killings in So Paulo and Buenos Aires: while neither of the two
legal systems is very effective, the apparently stronger institutions in the former
coexist with a higher degree of impunityfor rights violators.
One of the argumentsthat have been proposed to explain democraticdeficiencies
in Latin America is that informal institutions contradict formal democratic ones.5
But despite valuable insights, insufficient attentionhas been paid both to the defini-
tion of informal institutions and to empirical analyses of their effect. This article
offers a precise definition of informalinstitutionsand evaluates one informal institu-
tion in an area vital to democracy,the enforcement of civil rights, and its effect on
the judiciaries in Sao Paulo and Buenos Aires. The data are drawnfrom a sample of
cases taken from the courts of So Paulo and Buenos Aires involving police homi-

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

cides, supplementedwith interviews and archivalresearch.The police in both coun-


tries observe an informalrule of behaviorthat goes far beyond what the formalrules
contemplateas the properuse of deadly force; this rule broadly condones killing, as
long as it occurs in the course of routinepolicing. The rule does not extendto the use
of violence by police officers in personal disputes. One of the essential featuresof
an institutionis its enforcementmechanism, and this informalrule can not count on
the judiciary for enforcementexcept in a subset of cases involving victims who are
perceived as having engaged in violence against society.
The interactionof these formal and informal rules generates a three-waydistinc-
tion among the cases. First, in those cases in which a police officer kills in the
course of a privateconflict, both the police and judicial actors are unified in the rule
of behaviorto be applied,and thus the courts would be expected to be at their most
effective. Here the formal rules govern the enforcement structure,and the formal
institutionis clearly the operative one. The results in both countries will be similar
depending only on the strength of the formal institution. Second, in those cases in
which the police and the courts are unified in supportingthe informalrule (the vio-
lent victim cases), the results would be expected to reflect the informalrule. In these
cases there is an informal institutionat work that includes the formaljudicial struc-
ture as its enforcement mechanism, and again the results will be similar in both
countries.Finally, in those cases in which the police and the courts are at odds with
each other on the rule of behaviorto be applied,the results would be expected to be
mixed and largely dependenton the circumstancesof the victim or the publicitygen-
eratedby the case. The police are the investigative arm of the legal system and thus
have a great deal of control over the informationthat enters the system; they will be
devoted to protecting those who behave according to its rules of behavior.
Prosecutionswill be successful only in those cases in which the victims' social net-
work is able to overcome this impedimentby supplyingthe informationdirectlyor in
which extraordinarymeasuresare taken by judges or prosecutorsto gatherthe requi-
site information in response to public pressures. Here, results will be worse in Sao
Paulo than Buenos Aires, because of the greaterdegree of social exclusion and fewer
resourcesof the victims. These expectationsare summarizedin Table 1.

Informal Institutions, Politics, and the Rule of Law

Since most Latin American countrieshave adopted laws that embody liberaldemoc-
ratic guarantees of individual rights, many of the democratic deficiencies that
observerspoint out can be restatedas the failure of actual practices to live up to the
formal laws. The problem of "brown areas," "incomplete citizenship," "illiberal"
democracies, or "unconsolidated"democracies is that the laws do not, for whatever
reason, structuresocial and political reality to a sufficient degree.6It is at least partly

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

Table 1 Expected Effect of Formaland InformalRules

Roleappliedby judge.sandprmosectrs
Fowerma i
dre infrsal Rule
High of conviction;
likelihood possible,and
Theoretically
For ..fo l rulesgovern (Private wouldpredict of
preeminence
raie V4iole~aca~e) theinformalrule,butnot
presentin my cases
Medium rateawith
conviction
levelsofimpunity
depending
largly ontherources
brouht Verylow likelihoodof
tobearbyinterested (.,
parties onviction; rulesdonot
formal
ride inBuenos
higher Airesthanin govern
(ViolentVictiMn
cas)
Seo Paulo)
(Reutine
Policing
cases) ----------

in this sense that many observershave pointed to the failure of the rule of law as one
of the key problems in LatinAmerica.7
This conceptualizationmay conjureup images of anomie, of lawlessness. In fact,
what is observed in many cases is not the absence of rules, but the presence of alter-
native rules, which are often labeled informal institutions or informal rules. These
alternative rules, where they exist, supersede, condition, or otherwise affect the
validity of formal rules. O'Donnell, for example, argues that many new democracies
are experiencing,not a negative condition, the lack of consolidation (or institutional-
ization), but rathera positive condition, the informal institutionalizationof alterna-
tive practices.8Helmke argues that the critical institutionsin understandingthe inde-
pendence of the supreme court in Argentinaare not only the formal ones that grant
secure tenure,but also the informal ones that allow new governmentsto oust justices
associated with a prior administrationor regime.9Da Matta argues that universal cit-
izenship rules in Brazil are conditioned by the preference for personalism in the
applicationof the law.10
There are many benefits in focusing on informal rules. However, because infor-
mal institutions are generally not written down, there is a risk that the concept can
become tautological and can become a residual category to explain empirical regu-
larities not otherwise accounted for, much as a vaguely defined "culture"has been
used to explain perplexing behavior.Without a precise definition, informal institu-
tions will remain fundamentallyunsatisfyingas explanatoryvariables.
Common to nearly all definitions of institutions is the idea that rules are an insti-
tution'sbasic building blocks. B. Guy Peters reviews the various strandsof institu-
tionalism and places the definitions in three camps: institutions as normative struc-
tures, institutions as rules, and institutions as regularized patterns of interaction.'1
But a sound definition of rules requiresboth a normative component and a regular-
ized pattern of behavior, encompassing all three conceptions of institutions. Some

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

definitions already include all these elements, at least implicitly. O'Donnell, for
example, puts patternsof behavior,rules, and norms into his definition when he says
that "institutionsare regularizedpatternsof interactionthat are known,practicedand
regularly accepted...by social actors who expect to continue interactingunder the
rules and norms formally or informallyembodied in those patterns."12 Juristsfurther
point out that there are both primary rules-rules of behavior-and secondaryrules
that determineamong otherthings how those primaryrules must be createdand who
is entitled to enforce them.13Institutionsare made up, not of a single rule, but of a
complement of primaryand secondaryrules that specify not only the regulatedcon-
duct but these other matters as well. One of their crucial features is the manner of
enforcement.14
Formal rules, and thus formal institutions, are usually easy to identify (though
Hartpoints out that they are quite a bit harderto define). They are generallywritten
standards for conduct produced according to specified procedures by authorities
legally invested with the power to do so. Laws, for example, are enactedby congress
and signed by the executive in accordance with constitutional procedures. Hart
would describe such enactmentas being in accordancewith the rules of recognition
that govern a particularsystem. Hart'swork offers the first-negative-clue to the
nature of an informal rule: an informal rule is one that was not enacted in accor-
dance with those rules of recognition, the formalities. But since informalrules are
not flagged by duly observed formalities at the time of their creation, they will be
easy to confuse with other empiricalregularities.
The challenge is in theorizingthe distinctionbetween mere behavioralregularities
(which might not be rules) and informalrules (which make up informalinstitutions).
For example, a distinction is intuitively recognized between the statements"people
take off their coat when they enter a restaurant"and "people take off their hat when
they enter a church."Both describe behavioralregularities,but the formerbehavior
respondsto a merely predictiverule, while the latterobeys a rule that is in some way
prescriptive.To aid in the distinction,Hart calls on an "internalcomponent"of rules,
according to which "some at least must look upon the behaviour in question as a
general standardto be followed by the group as a whole."15The distinguishingchar-
acteristic is, in theory, observable:"true"rules (that is, prescriptiveones) will sup-
port secondary (enforcement) conduct to be applied to rule breakers. Moreover,
those who are charged with enforcementmust generally assert the rule itself as the
basis for their enforcement action.16But a behavioralcomponent is also important.
As Ellickson argues, "a guideline for human conduct is a rule only if the existence of
the guideline actually influences the behavior either of those to whom it is addressed
or of those who detect others breachingthe guideline."17Therefore,a rule must have
both normativity(Hart'sinternal component) and facticity (Ellickson's requirement
that either its existence or its violation produce some behavioralconsequences).18

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

In thinking about informal institutions,the distinction between behavioral regu-


larities and rules is crucial. Otherwiseindistinguishablebehavioralpatternsmight be
either the product of a rule that is likely to generate its own dynamic in terms of
social disapproval and susceptibility to sanction or simply the regular result of
antecedent institutions, structures,or incentives. The danger is that any regularity
will simply be called an informal institution when the explanation should be taken
one step up, to the set of structures, institutions, or incentives that produces this
result time after time even in the absence of a prescriptiverule. In summary,a pat-
tern of behaviorresponds to an informal institution if it meets both a positive and a
negative condition. It must occur in response to certain primary rules that are
enforced by the relevant agents of social control (that is, the rules must show both
normativityand facticity), and these rules must be enforced even though they were
not createdusing the proceduresprescribedin the secondaryrules.
One last theoreticalpiece needs to be made explicit: the relationshipamong infor-
mal institutions,behavioralregularities,and the rule of law. Niklas Luhmannargues
that the legal system is normativelyclosed but cognitively open.19On the one hand,
any new principle (or ruling or decision) ought to be normativelyvalidatedby refer-
ence to the law itself. The intrusionof informal rules into legal decision making vio-
lates normative closure by incorporating rules of decision that were not created
accordingto legal procedures.On the otherhand, the legal system ought to be cogni-
tively open, as it depends on external sources for information.20When the system
consistently does not receive the informationit requires to decide certain cases cor-
rectly,the problemis a failure of cognitive openness.
The legal system fails to produce outcomes in accord with formal rules when it
either judges the wrong facts using the right standard(cognitive closure) or applies
the wrong rule of decision to the right facts (normativeopenness). Either failure can
produce a behavioral regularity,the former because there may be institutional or
structural reasons for a persistent blindness to the right information, the latter
because an institution in which all actors are enforcing a certain rule of behavior
(formal or informal) is likely to produce decisions in accordancewith that rule. But
if it is accepted that informal institutions must be made up of rules, only the latter
describes the situation when an informal institution has incorporatedthe legal sys-
tem as part of its mechanism of enforcement. Moreover, this theoretical difference
has real consequences: the necessary corrective measures in the case of cognitive
closure would not be the trainingof legal actors in the correct normativeframework,
but ratherthe tailoring of the institutionsto gather the requisite informationso that
legal actors can make informed decisions about the cases. Did the failure to punish
police killings in Buenos Aires and Sao Paulo respond to an informal rule of deci-
sion, or is it more appropriatelyclassified as a systematic and regular failure to
process the cases correctly,while attemptingto apply the correctrules of decision?

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

General Patterns

In Sao Paulo and Buenos Aires, at least, the institutionalstrength of the organiza-
tions that make up the legal system is inverselyrelated to their effectiveness in pros-
ecuting civil rights violations. Brazil's Ministirio Prblico has been praised for its
robustness and independence, while prosecutorsin Argentinaare underfundedand
politically subject.21The state-runlegal assistance office in Sao Paulo (a part of the
Ministcrio Priblico) is betterfunded than its counterpartin Buenos Aires. The courts
in Sdo Paulo, if not necessarily in all of Brazil, have more rationalizedcareers,big-
ger budgets, and better facilities than those in Buenos Aires and are more indepen-
dent. At the same time, a measure of the effectiveness of these two systems in deal-
ing with a serious problem like police homicides indicates that the courts in Buenos
Aires, though far from perfect, manage to produce better results than those in Sao
Paulo. To begin to resolve this paradox,it is necessary to go beyond the formal char-
acteristicsof the institutions.
The behavior at issue, then, is the regular and illegal use of lethal force by the
police. This behaviorwill be taken as given, with no attemptto explain it, in orderto
explore the reaction of the courts to the problem.22The substantivelaws on the use
of lethal force by the police are essentially the same in Argentinaand Brazil. Both
countries have signed conventions that protect human rights, among them due
process, access to the legal system, and freedom from summaryexecution and other
arbitrarypolice violence, and both of them have incorporatedthese laws into the
domestic legal framework.
Nevertheless, in the last decade the number of civilians killed in Sao Paulo and
Buenos Aires has remainedvery high and shows signs of worsening. On average,the
police in the Buenos Aires metropolitan area killed 160 persons every year from
1990 through 2000. The worst year was 1999, when they killed 277 persons, more
than five people a week. The trend has been generally rising. In Sdo Paulo the num-
bers are even higher, averaging680 victims annually from 1990 through2000. The
numberpeaked in 1992 at a stunning 1,428. Even if the immense populationof the
state of Sdo Paulo-around 36 million-is considered,many countriesdo not experi-
ence this number of casualties in the midst of a civil war. The difference in the size
of the two cities obscures the similarityin per capita figures. From 1990 to 1994 Sdo
Paulo's police homicide rate was twice as high as Buenos Aires's, 2.2 deaths per
100,000 to 1 per 100,000. But from 1995 through2000 the rates were nearly identi-
cal, 1.61 to 1.63.23
By way of contrast,other cities with large populationsand police forces and noto-
rious violent crime problems show indices that are ordersof magnitudelower in both
absolute and relative terms. In 1992 Sdo Paulo police killed thirty-one people for
every death in Los Angeles, the city whose police "kill more people in proportionto

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

the size of the force than any other major U.S. police department."Los Angeles
reports approximately.5 killings per 100,000 in 1991 and 1992. New York'sfigures
are lower; an average of about twenty-five persons were killed per year between
1991 and 1993 (.34 per 100,000).24The best available number for a single year for
Mexico City, an urban area larger even than Sdo Paulo, is still less than one per
100,000.
In summary,the members of the security forces in these two cities behave as if
they have the discretion to use lethal force at will against certain marginalized
groups. But can it accuratelybe said that there is an informal institutionat work that
gives the police discretion to kill members of this population, with an enforcement
structurethat includes the judiciary? Ellickson notes that the best evidence of the
existence of a rule is the presence of enforcementbehavior.25If violations are detect-
ed but not punished, then the law that prohibits the arbitraryuse of lethal force by
the police is not the operativerule.26The effective (informal)rule would at minimum
permit this behavior and at worst rewardit. On the other hand, even if breaches are
infrequentlydetected, if punishment generally follows upon detection, the law still
constitutes a valid rule, and the question becomes why violations are so infrequently
detected.
The low conviction rate in both countries shows that neither system punishes a
significant proportionof police killings, though the Argentine courts do considerably
better than their Brazilian counterparts.27Table 2 compares the results for the two
countries,based on the sample of cases collected during fieldwork in Argentinaand
Brazil.28In this sample, the conviction rate in Argentina is around24 percent but in
Brazil only 10 percent. Moreover,in the cases that go to trial, Brazil's conviction rate
is about 40 percent, while Argentina'sis nearly 90 percent. In short, in Argentina
fewer cases are thrownout early,and, if the case is tried, the chances of a conviction
are much greater.By way of comparison,a prosecutorin Sdo Paulo said their figures
show a conviction rate for ordinaryhomicide cases that go to trial as high as 80 per-
cent, while a study by Adorno of race and criminal prosecutions reports an overall
conviction rate around70 percent in homicide cases that go to trial.29Argentinaalso
has a large numberof pending cases, but they do not raise hopes for a higher convic-
tion rate. Interviews suggest that cases in which a conviction is likely to be legally
requiredbut unpopularwith the police, political powers, or public are often ware-
housed, either indefinitely or until they can be dismissed for proceduralreasons. The
data supportthis assertion.The mean delay for pending cases in Buenos Aires (mea-
sured to the sampling cut-off date) is fully 1.5 years longer than for all other cases,
suggesting that at least some cases in this category have been pending for an unusu-
ally long time.

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October 2003

Table 2 Effect of Formaland InformalRules on ConvictionRates

Rule applied by judgeaand.prosecutors:


.........
.r-U f
F afor te
.P Ar,Sni.Rate .R"
~iF fiLp~~ I
i ~?Cg:
:"

'k
:PO
rate

IirR Pcon
(i 0%a8~8
136: 0,2)~I~~~E
I
The Existence of a Competing Informal Institution

Some civil society actors accept the argumentthat the phenomenon at issue here is
the result of an official if informalpolicy to repressthe poor and preservethe liberal
economic orderthe state has put in place. But not all police killings go unpunished,
and treating these cases without differentiationobscures significant differences.An
effective informalrule precludes punishmentof police officers who kill personsper-
ceived to be violent criminals, encourages persistent systemic blindness in cases in
which the police kill in the course of routine policing activity, and allows the law to
be upheld relatively well in cases in which a police officer is accused of killing
someone in a private dispute. In terms of Luhmann'stheoretical framework,in the
case of a violent criminal who is executed by the police, impunity is producedby a
failure of normative closure-the intrusion of an informal rule of decision. In rou-
tine policing cases, the problem is more properly described as a failure of the sys-
tem's cognitive openness, which leads to mischaracterizingthe cases as legitimate
uses of force. This failure is systematic primarilybecause many of these cases arise
in a social context that is-because of police resistance and the judiciary'sdemand
for strong evidence, certainty, and other characteristics-especially opaque to the
eyes of the law. Neither of these failures afflicts the system in cases of privateuses
of violence by police officers.
To test the three-way distinction three kinds of cases from the datasetare exam-
ined. First are the cases of private violence (in which a police officer kills someone
in the course of a personal quarrelor domestic dispute, for example). As expected,
these homicides are generally punished in both countries. In Buenos Aires, among
decided cases of this kind, about 95 percent ended in conviction, while in Sao Paulo
the conviction rate is 100 percent. Inclusion of pending cases lowers the percentages
to 46 and 56 percent, respectively,but these figures are still far higher thanthe rates
in any other category. Interestingly, here the data support the argumentthat Sio

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

Paulo has the strongerinstitution in formal terms. When all the actors are united in
their efforts to prosecute offenders, the systems in both places produce a high con-
viction rate, and the courts in Sio Paulo, working more quickly and efficiently than
those in Buenos Aires, get betterresults.
The second category involves the execution of so-called dangerous criminals.
This categorization of the cases presents ideal types. In practice, victims may be
more or less taintedby a connection with violence, and the evidence of their involve-
ment may be more or less firm. Thus, the rule works more as a sliding scale; the
level of prosecutorialand judicial resources devoted to a case goes down (producing
a lower conviction rate) as the cases approachthe extreme. Forpurposes of this tabu-
lation, however, only clear-cut cases are included; only cases in which there was
solid thirdparty evidence that the victim had a close connection to a violent crime-
for example, unquestioned evidence of a shoot-out or hostage situation-are coded
as violent victim. As expected, none of the cases coded in this category led to a con-
viction, in eitherjurisdiction. On close examination,it is clear that, if the victim was
involved in a violent crime that presenteda serious risk to society, then the victim's
execution by the police was not punished even if it took place after the victim no
longer posed a threat.
In the last class of cases, routinepolicing, the police apply one rule, while judges
and prosecutors apply another. These cases involve the excessive use of force in
apprehendingpetty criminals, persons who run away from identity checks, and the
like. Again, the results are as expected. About 6 percent of the cases that fit this pat-
tern ended in a conviction in Sao Paulo, while the conviction rate is 21 percent in
Buenos Aires. In both jurisdictions, though to quite different degrees, prosecutors
will push cases, judges will bring them to trial, and some convictions will result if
the abuse of authorityis patent and the evidence suggests that the victim was not a
"dangerous"criminal. Some convictions result even when the victim was involved in
some petty criminal activity, though a conviction is more likely when the victim is
shot while fleeing out of a general fear of the police than when the victim was
actively breakingthe law. At the same time, the refusal of the police to cooperate in
the investigationmakes it difficult for the system to act effectively. Table 3 summa-
rizes the results, which match the expectationslaid out in Table 1. This simple tabu-
lation of the outcomes is clearly consistent with the hypothesis that in the violent
victim cases the courts apply a rule that precludes punishment- the informal vio-
lent victim exception to judicial protection.In the routine policing cases, in contrast,
judges and prosecutorsare occasionally willing to prosecute and can obtain convic-
tions, though far less often than in those cases in which the police is also supportive
of the basic rule that was allegedly violated.
A closer, more qualitativelook at the process offers not only supportingevidence
but also some insight into the mechanism by which it is accomplished. In at least
some of the violent victim cases, the facts that would support a conviction are

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

Figure 3 JudicialOutcomes by Jurisdiction

AimSao
Buenas Paulo
117 17
Pending 43% 10%
80 105
Dismissedwithouttrial 29% 64%

Acquitted 4% 16%
-67
17
Convicted 24% 10%
.Total274 165
Tol.. . .. . .. . . .. 100%.
. .. . . . .
100%

known, yet the violations are not punished. Also, judges and prosecutorsprosecute
routinepolicing cases if they consider that the evidence at trial will supporta charge
that the police officer exceeded the legitimate use of force, but prosecutions are
hamperedby the difficulty in obtaining good informationabout cases in which the
victims are predominantly marginalized, lower class males, especially since the
police actively oppose the prosecutions.
The zero conviction rate in violent victim cases may be attributablecoincidentally
to the impossibility of gathering sufficient evidence on which to base a conviction.
However,the cases in the sample suggest that, whetheror not all the informationthat
might be needed to produce a conviction is present, the killing of a person believed
to have taken up arms against honest citizens goes unpunished.Regiane Dos Santos
was a twenty-year-oldwoman who, with her husband,attemptedto rob a home. The
police arrivedin the middle of the holdup, and she and her husbandtook the family
hostage. An exchange of gunfire ensued during which the police killed her husband
and one of the hostages, a child. Regiane then indicatedher intent to surrenderand
turned over her gun to the owner of the house. The police entered the house,
removed her to the bathroom, and executed her. The policemen were tried and
acquitted by a panel of judges (before the transfer of jurisdiction to civilian jury
courts). According to the attorney who representedRegiane's children and mother,
the judges blamed Regiane for the death of the hostage and ignored the testimonyof
the owner of the house, who said she had surrenderedand relinquished her gun
before they took her away to be killed. The court of appeals affirmed the acquittal.
In Argentinathere are several similar cases. The best known is probablythe Villa
Ramallo case. In 1999 the police surroundedand incapacitateda car carrying four
people away from a bank robbery.They pulled the occupants out of the car and exe-
cuted all but one of them. One of those executed was a robber, but the other two
were employees of the bank who had been taken hostage. The same evidence sup-
ported the conclusion that all three men were killed while unarmedand defenseless

10

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

after being pulled from the car. But the prosecutionproceeded only to the killing of
the two bankers,not the killing of the bank robberand hostage taker,who, while an
unappealingindividual,was also entitledto legal protection.
While no prosecutor or judge interviewed admitted directly that they would not
convict underthese circumstances,in these cases the violation of the formal rule was
observed by the system, and still there was no sanction, suggesting the applicationof
an alternativerule of decision. Moreover,there is abundantindirect evidence of the
existence of this informal rule. Police officers who kill are often promoted. For
example, Gilson Lopes in Sao Paulo consistently climbed the ranks of the military
police in no small part due to his role in at least forty-two killings over the course of
a decade, before losing a homicide trial and being removed from the force. Carlos
Ruckauf, when he was governorof the province of Buenos Aires, offered a bonus of
up to five times their salary to police officers involved in "acts of bravery,"which he
loosely defined as the detention or killing of violent criminals.30This gesture echoes
the actions of the commanderof the Sdo Paulo military police, who also offered a
rewardand a five day leave to those who distinguishedthemselves by killing some-
one.31
The behavior of the police is itself evidence that they rely on this rule. In many
cases they carefully create the appearancethat the victim was a menace to society. In
more than half the cases in Argentinaand at least as many cases in Brazil some indi-
cation was given, either because judicial proceedings uncovered police deception or
because advocates for the victim made a concrete allegation, that the police had
either staged a confrontationafter the death of the victim, planted a gun, threatened
witnesses, produced false forensic reports, or tampered with the process in some
similar way. The methods are diverse and creative. One former policeman testified
that the Sdo Paulo police occasionally placed a gun in the dead victim's hand and
pulled the trigger so the skin would show gunpowderresidue; in a couple cases the
police shot their own car to simulate a confrontation;in others they were observed
transporting the victim's body to the place where the confrontation allegedly
occurred. Often the evidence of tampering is irrefutable,as when a second report
confirms that the shots entered the back of the head, rather than the front of the
body, or when a gun attributedto the victim is traced to the comisario who had
charge of the operation.Othertimes there is no more evidence than the (persuasive)
claim of an advocate for the victim.
Published studies of police violence in Buenos Aires and Sao Paulo support the
conclusion on this point.32In commentingon a recent case, the CELS writes:

[This case] occurredin the context of an institutionalpatternthat promotes complicity and cover-
ing up these facts, forging proofs and attempting to present these murders as confrontations. In
addition, it makes evident the refusal of those in charge of the Buenos Aires police to investigate
these events, to adopt any preventivemeasures as to those chargeduntil the courts requireit, or to
develop any policies whatsoeverto reversethe increase in police violence.33

11

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October 2003

Even the relatives of victims emphasize that their relative was not a marginal.It
is not uncommon to hear them justifying the killing of "lawless" criminals, even
while decrying the unjustified murderof their own relative. In the final instance,the
problemin these cases is that neitherthe prosecutornor the judge nor the jury is pre-
pared to convict a police officer when they consider that the victim has placed him-
self or herself outside the social contract.34It is easiest, therefore,to simply look the
otherway if there is any evidence thatmight suggest a conviction is legally required.
The situation is different in the routine policing cases that do not involve a so-
called violent victim. Most of these cases (more than half of the nonpendingcases in
Buenos Aires and nearly 65 percentof all cases in Sao Paulo) are dismissed early on,
ostensibly for lack of evidence, but occasionally one goes to trial. In Sdo Paulo, in
1984, the police removedtwo youths, ValdeciAntinio da Silva (seventeenyears old)
and RobertoThomaz de Oliveira (eighteen), from a taxicab run by an acquaintance.
Their bodies were found hours later, in an empty lot, with close range gunshot
wounds. The police officers who acknowledged killing them were broughtto trial
before a Sdo Paulo jury. According to the legal records, the prosecutordeemed the
case worthy of prosecution, and the judge agreed, but the jury acquittedthe police
officers on groundsthat they had exercised legitimate self-defense.
On the face of it, the result is inexplicable.The taxi drivercould have confirmed
that the boys were unarmed, were arrested without resistance, and were last seen
handcuffed in the back seat of a police vehicle. But at the trial the only witnesses
were the two police officers. Under Brazilian rules against self-incrimination,they
were able to tell their story without cross-examination.The prosecutorcalled no wit-
nesses because the only independentmaterialwitness, the taxi driver,also a resident
of the favela, could not be found after the long delay. Thus, the only version of
events presented to the jury with any evidentiary support was that the two young
men had met the police with armedresistance undercover of darknessand had been
killed when the police respondedby shooting into the overgrownlot wherethey were
hiding. Simply put, the jury judged the wrong set of facts.
On the other hand, in those cases in which witnesses are able and willing to testi-
fy against the police officers, there can be a conviction. In the case of JallioCesar
Antunes de Miranda, witnesses testified that the victim had tried to evade police
who wanted to check his identity,was cornered,and was pleading for his life when
he was killed. According to the detailed verdict, the jury was willing to approveof
the use of force in general but by a vote of four to three believed that the police offi-
cers had intentionally exceeded permissible limits on the use of force. Along the
same lines, the police officer that killed Eneas da Silva, twelve years old, after com-
ing upon him leafing through a pornographic magazine, was convicted and sen-
tenced to thirteen years in prison. In that case, the prosecution survived a falsified
forensics report, a planted gun, false testimony from the police officer, and the

12

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

intimidationand beating of eyewitnesses. At the trial, a second, accuratereport was


introduced,the eyewitnesses testified, and the fact finder chose to believe that the
child who was killed was unarmed. In this case, the prosecutor ensured that there
would be an adequatefactual recordto supporta conviction (prompted,no doubt, by
extensive media and internationalattentionto the case).
At first glance, these results appearfortuitous:they depend on the luck of finding
an eyewitness or an aggressive prosecutor or on the vagaries of media attention. In
fact, the only useful generalization about these cases is that the prosecutions fail,
when they fail, for lack of sufficient reliable and accurateinformation.But the regu-
larity of the failures suggests that the results are not purely random. As to these
cases, qualitativeevidence suggests that the root of the problem is the cognitive clo-
sure of the system, primarilywhen the victims belong to the excluded social classes.
Cases involving police homicides overcome the two greatest informationaldiffi-
culties facing most other criminal investigations. Because they involve a homicide
and currently victims do not (usually) simply disappear, the legal system nearly
always learns that a potentially criminal act has taken place. Moreover, the person
who committed the act is nearly always identified. Nevertheless, police homicide
prosecutionsalways face special difficulties. The initial investigation is in the hands
of the police, and the first legal actors on the scene are the perpetratorsthemselves.
As a result, in both Sao Paulo and Buenos Aires there is a documented pattern of
crime scene corruption.The police typically remove the body under the pretense of
renderingassistance, though it is just as often true that victims arrive in the hospital
with more bullet holes than they had when they left the crime scene, or after long,
unexplained delays. They mangle the scene intentionally or through carelessness,
clean up real evidence, and plant false evidence.
Clearly,when the police do not performtheir investigative function, a lower con-
viction rate might be expected. Other actors-prosecutors and judges in these civil
law systems-can in theory compensate for this deficiency. Moreover,the social net-
work of the victims can also step in and supply information or demand a response
from the system. But when primaryresponsibility for generating informationis laid
on these other actors, the marginal social position of the typical victim creates spe-
cial informationaldifficulties. In the first place, those close to the victims are also
likely to be from marginal classes, and the police have much greater freedom to
threaten and intimidate complainants and potential witnesses from the margins of
society than a comparablemiddle class group of people. In the cases studied here the
police often knew both the victims and those who would be in a position to complain
about abuses. In some cases, the police create charges against those who are
involved in cases against them; in others they ostentatiously patrol the block on
which the complainants live; in yet others witnesses and complainants are them-
selves victims of violence. In one Buenos Aires case the police threatenedan entire

13

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

neighborhood (which had initiated demonstrationsprotesting the killing) with the


withdrawal of police protection. A less marginal group would have far greater
resources-access to lawyers, the press, and political leaders-with which to
respondto this pressure.
In addition, the repeated experience of being excluded from effective legal
redressjustifies this group's lack of confidence in the justice system and its belief
that the police will act with impunityboth in the initial violation and in carryingout
any threatsof retaliation.35As a result members of this group often conclude there is
little to be gained and much to be lost from becoming visible and vigorous advocates
on their own behalf; many times neighbors and friends actively discourage family
members from taking action. The media can remedy some of this inability to com-
plain, but the media, in these cities as much as anywhere,disproportionatelyreport
on violence affecting the middle class and virtually ignore events that take place in
the villas andfavelas unless they are truly spectacular.
Prosecutors, in turn, depend greatly on the initial police report to determine
whether to demand and wait for additional information, file the equivalent of an
indictment,or ask the court to dismiss the case without prejudice.Most prosecutors
will readily admit that they rely very heavily on the contents of the report;indeed,
the impression of outside attorneys and judges is that by and large they limit their
evaluation to the informationpresented to them. Prosecutorshave few resourcesto
conduct their own investigationsand rely on the police as their work force; they may
need the police's cooperation in hundreds of cases for every one that involves a
police officer. Clearly, it is in their interests to retain an amicable relationshipwith
the police. The more marginalthe victim is, the less likely it is that the prosecutor
will pay a political price for improperly dismissing a case. Finally, two different
prosecutorsin Sao Paulo told me that the single strongestpredictorof an acquittalby
the jury was the criminal record of the victim. This belief can become a self-fulfill-
ing prophecy, as prosecutorsthen rely on it to dismiss cases based on the unlikeli-
hood of a conviction.
Delays (never less than two years to trial in either city, and generally much
longer) also disproportionatelyaffect the underprivileged.A request to bring in a
witness might languish for a year, especially if there is no address because the wit-
ness lives in a villa or does not have a fixed place of employment.The typical com-
plainants and witnesses in these cases are transient,have no formal addressor fixed
employment, and are hard to track. Cumulatively, this interaction between social
class and the demand for information from the legal system has regular conse-
quences. At each stage people from the margins of society face special difficulties,
which build on each other until the end result is, predictably,the failure of the sys-
tem to respond effectively to their demands,not in all, but in the majorityof cases.
This explanationis consistent with the significantly differentresults in So Paulo
and Buenos Aires. Not only are overall levels of economic inequality greaterin the

14

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

former, but the sample shows that the underprivileged are more overrepresented
among victims of the police in Sao Paulo. Several of the lawyers interviewed in
Argentinasaid that some of their clients were so removed from the formal world of
government services and formal employment that they engaged the legal system
only with great difficulty. But there is a smaller proportionof victims drawn from
this class in Buenos Aires than in Sao Paulo. And social distances from the state,
includingthe legal system, are greaterfor the excluded sectors of Sao Paulo.
During the disturbancesin Argentinain December 2001, television crews filmed
an attackby rioters on a police officer. Three young men in a middle class neighbor-
hood were overheardby another police officer as they apparentlyexpressed some
supportfor the attackand were shot to death.An intervieweddemonstratoressential-
ly summarizedthe argumentmade here.

Peopleare indignantwiththe policebecausetheytriedto pass off the threekids as thieves.This


murderous policemanplanteda knifeon eachof themto justifythathe killedthembecausethey
triedto robthe servicestation.If the kids hadnot beenfromthisneighborhood,
you canbet that
thepolicewouldhaveclaimedtheywerethieves,andthejudgewouldhavebelievedthestory.36

In other words, the police nearly always try to present these cases as the repres-
sion of violent criminals, fabricating evidence if necessary. In those instances in
which the victims do not come from the middle class and do not have the requisite
resources and mobilization, the police are likely to succeed in that characterization,
either because the judge and prosecutorhave little incentive to dig deeper than the
police version or because there would be no credible witnesses available at an even-
tual trial to contradict the police version. The legal system demands credible wit-
nesses and solid proof to convict. The regular failure to convict police officers,
therefore,is not so much fortuitous as it is the probabilisticresult of the interaction
of the high informationaldemand of the legal system with the precariousliving situ-
ation of those most likely to be affected by the violations. The difference in convic-
tion rates between Buenos Aires and Sdo Paulo is a reflection of the greaterlevels of
inequalityand the greatermarginalizationof the targetpopulationin the latter.

Conclusion

The impact of informal institutions on the operation of formal institutions is being


increasinglynoted, but there are few empirical studies of the interactionof the two.
Informal institutions are more restricted than simple behavioral regularities. The
explanation of the high degree of impunity for police homicides is more complex
than a simple informally institutionalizedexclusion of the poor from the universal
benefits of citizenship.

15

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

In the relatively fewer cases that involve a victim believed to have used violence
against society, there is indeed an informal institution at work. This institution
includes a rule of behavior that excludes these victims from the protection of the
courts and an enforcementapparatusthat extends up into the judiciary.In these cases
convictions are extremely unlikely even where the violation is detected. In the more
numerous routine policing cases, however, while there is evidence that the police
continue to apply an informal rule permittingthe use of deadly force, there is little
evidence that this rule currentlypenetratesthe regime'sformal enforcementstructure
(the judiciary). Its efficacy is therefore limited to those cases that are not detected.
The inability of nonviolent but marginalizedmembers of society to claim judicial
protection in these cases is due primarilyto evidentiary failures. These failures are
more likely in a context of social inequality,which makes it more difficult for victim
advocates to produce the informationthe institutiondemands and reducesthe incen-
tives for judicial officers to dig up the informationthemselves.
In the cases in which the actors apply the wrong standard, the prescription
requiresthat the relevantlegal actorsbe instructedin the relevantcodes and thatthey
be applied. However,in the routinepolicing cases it is more importantto ensurethat
the enforcementagents have the tools (and incentives) they need to gatherthe requi-
site information.Not only is it necessary to have adequatebudgets, personnel,and
equipment,but the system must also be adaptedso it is open to informationfrom its
social environment,with all its peculiarities. In police violence cases, specialized
entities and procedureswill be requiredto extract informationfrom two very diffi-
cult contexts: the police environment,with its code of silence, and society's margin,
with its transientand socially distantdenizens.

NOTES

I am grateful for the encouragementand suggestions of Rossana Castiglioni, Michael Coppedge, Tulia
Falleti, Andrew Gould, Scott Mainwaring,Guillermo O'Donnell, and two anonymous reviewers;for the
financial assistance of the Social Science ResearchCouncil and the Kellogg Institute,Universityof Notre
Dame; and for the invaluable help of the people at CORREPIin Buenos Aires and C6rdoba,CELS in
Buenos Aires, and the Ouvidoriada Policia in Sdo Paulo.
1. Compare Scott Mainwaring, "The Surprising Resilience of Elected Governments,"Journal of
Democracy, 10 (July 1999); Guillermo A. O'Donnell, "On the State, Democratization, and Some
Conceptual Problems:A Latin American View with Glances at Some PostcommunistCountries,"World
Development,21 (August 1993).
2. O'Donnell, "Onthe State."
3. Paul Chevigny, Edge of the Knife: Police Violence in the Americas (New York:The New Press,
1995), p. 145.
4. Maria Dakolias, "A Strategy for Judicial Reform: The Experience in Latin America," Virginia
Journal ofInternational Law, 36 (Fall 1995); Nibaldo H. Galleguillos, "Judicialand Legal Reforms in the
DemocraticTransition:An Assessment of the Changing Roles of the Judiciaryin Chile,"paperpresented

16

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

at the meeting of the Latin American Studies Association, Washington,D.C., 2001; Linn Hammergren,
"FifteenYearsof Justice and Justice Reform in LatinAmerica:WhereWe Are and Why We Haven't Made
More Progress,"http://darkwing.uoregon.edu/-caguirre/papers.htm (1999).
5. See Gretchen Helmke and Steven Levitsky, "Informal Institutions and Comparative Politics: A
PreliminaryResearchAgenda,"paperpresented at the Annual Meeting of the American Political Science
Association, Boston, August 28-31, 2002.
6. O'Donnell; James Holston and TeresaP. R. Caldeira,"Democracy,Law and Violence: Disjunctions
of Brazilian Citizenship,"in Felipe Agilero and Jeffrey Stark, eds., Fault Lines of Democracy in Post-
Transition Latin America (Coral Gables: North-South Center Press, 1998); Larry Diamond, ed.,
Democracy in Developing Countries: Latin America (Boulder: Lynne Rienner, 1999), pp. 42-51;
Wolfgang Merkel, "Defective Democracies,"InstitutoJuan March de Estudios e Investigaciones, Centro
de Estudios Avanzados en Ciencias Sociales, WorkingPaper 1999/132 (1999); E Zakaria, "The Rise of
IlliberalDemocracy,"ForeignAffairs, 76 (November-December1997); LarryDiamond, Consolidating the
Third WaveDemocracies (Baltimore: The Johns Hopkins University Press, 1997); Larry Diamond,
"Democracyin Latin America: Degrees, Illusions, and Directions for Consolidation,"in Tom Farer,ed.,
Beyond Sovereignty:Collectively Defending Democracy in the Americas (Baltimore:The Johns Hopkins
University Press, 1996).
7. Diamond, ed., Democracy in Developing Countries: Latin America; Juan E. Mendez, Guillermo
O'Donnell, and Paulo Sergio Pinheiro, eds., The (Un)Rule of Law and the Underprivileged in Latin
America (Notre Dame: University of Notre Dame Press, 1999); M. Shifter, "Tensionsand Trade-Offsin
LatinAmerica,"Journal ofDemocracy, 8 (April 1997), 116.
8. Guillermo A. O'Donnell, Counterpoints: Selected Essays on Authoritarianism and
Democratization(Notre Dame: University of Notre Dame Press, 1999), pp. 175-94.
9. Gretchen Helmke, "The Logic of Strategic Defection: Court-Executive Relations in Argentina
underDictatorshipand Democracy,"AmericanPolitical Science Review, 96 (2002).
10. Roberto da Matta, Carnivals, Rogues and Heroes: An Interpretation of the Brazilian Dilemma
(Notre Dame: University of Notre Dame, 1991); Roberto da Matta, "The Quest for Citizenship in a
Relational Universe,"in John Wirth, Edson de Oliveira Nunes, and Thomas E. Bogenschild, eds., State
and Society in Brazil: Continuityand Change (Boulder:Westview Press, 1987).
11. B. Guy Peters, Institutional Theory in Political Science: The "New Institutionalism" (London:
Pinter, 1999), p. 146.
12. GuillermoA. O'Donnell, "DelegativeDemocracy,"Journal ofDemocracy, 5 (1994); Lee J.Alston,
Eggertsson Thrdinn,and Douglass Cecil North, Empirical Studies in Institutional Change (Cambridge:
CambridgeUniversity Press, 1996).
13. Robert C. Ellickson, Order without Law: How Neighbors Settle Disputes (Cambridge, Mass.:
HarvardUniversity Press, 1991); H. L. A. Hart, The Concept of Law (Oxford:ClarendonPress, 1961).
14. See Sue E. S. Crawford and Elinor Ostrom, "A Grammarof Institutions,"American Political
Science Review, 89 (September 1995); Ellickson.
15. Hart,p. 55.
16. Ibid., pp. 113ff. This definition excludes purely self-enforcing norms, the first category listed by
RichardA. Posner and Eric B. Rasmusen, "Creatingand Enforcing Norms, with Special Reference to
Sanctions,"InternationalReview of Law and Economics, 19 (September 1999). Norms in this category
are not properly labeled rules unless they are supportedby some enforcement mechanism that steps in
when self-interestfails.
17. Ellickson, p. 128.
18. By normativity,I do not mean moral or ethical sanction, but simply that the rule is held to pre-
scribe a behavioral standard.This terminology parallels Jiirgen Habermas,Between Facts and Norms:
Contributions to a Discourse Theory of Law and Democracy (Cambridge, Mass.: MIT Press, 1996),
thoughthe usage is slightly differentthan his.

17

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
ComparativePolitics October2003

19. Niklas Luhmann,A Sociological Theory of Law (London: Routledge & Kegan Paul, 1985), pp.
283ff. It is unclear whetherLuhmannintends this statementas a descriptionof the law or as a normative
standard.Clearly,it does not in any sense describe the law as it operates almost anywhere,and there are
some difficulties with its use as a normativestandard,as arguedby RichardLempert,"TheAutonomyof
Law: Two Visions Compared,"in GuntherTeubner,ed., Autopoietic Law: A New Approachto Law and
Society (Berlin: Walterde Gruyter, 1987). But its clear distinctionbetween the source of rules and infor-
mation is useful in this context.
20. William M. Evan, Social Structureand Law: Theoreticaland Empirical Perspectives (Newbury
Park:Sage, 1990), pp. 41-42; Niklas Luhmann,"Closureand Openness:On Reality in the Worldof Law,"
in Teubner,ed., p. 20.
21. Alvino Oliveira Sanchez Filho, "Incentivos E Constrangimentos Institucionais: O Ministerio
Piiblico Da Bahia E O ControleExterno Da Atividade Policial,"in Nelson de Oliveira,Lutz MulertSousa
Ribeiro, and Jose Carlos Zanetti, eds., A Outra Face Da Moeda (Salvador: GraificaEnvelope, 2000),
MariaTerezaSadek, ed., Justiga E CidadaniaNo Brasil (Sdo Paulo:EditoraSumare,2000).
22. For explorations of the roots of the problem, see Chevigny; Holston; Jerome H. Skolnick and
James J. Fyfe, Above the Law: Police and the Excessive Use of Force (New York:The FreePress, 1993).
23. The figure of 36 million inhabitantsfor the state of Sdo Paulo was reportedon the state govern-
ment's official website; the figure of 12 million for the Buenos Aires metropolitanarea came from the
Encyclopedia Britannica's list of largest metropolitanareas. The available figures for Sdo Paulo suggest
that the proportionof killings in the capital versus the rest of the state roughly correspondsto the capital's
relativepopulation.
24. Chevigny,p. 46.
25. Ellickson, p. 128.
26. Detection has a special significance here, sinc6 not all police killings are rights violations.What is
importantis whether the features of the act that characterizeit as a violation are known to the relevant
actors:prosecutors,judges, andjuries.
27. The point is not that the conviction rate should be 100 percent.The institutionshould also protect
the rights of those who are accused on the basis of flimsy or inaccurateevidence. Thus, a certainnumber
of acquittals is to be expected and is, indeed, a sign of health. But the rate of appropriateconvictions
should be higher where the investigative,prosecutorial,andjudicial functions are most effective.
28. In Argentina the point of departurefor the samples are lists of cases compiled by civil society
organizations(CELS and CORREPI)that conduct daily newspaperreviews and have networksof infor-
mants in marginal areas. They exclude cases in which they judge the victim was armed and resisting
police action. In Sdo Paulo the startingpoint was a list compiled by the Ouvidoriada Policia, a state-run
ombudsman that also reviews press accounts and receives complaints; here I excluded cases in which
available informationconfirmed that the victims were armed and forcibly resisting police action. In both
countries I followed up with interviews of participants,reviews of court documents, online newspaper
archive searches, or all three.
29. S6rgio Adorno, "Discriminaqdo Racial E Justiqa Criminal Em Sdo Paulo," Novos Estudos-
CEBRAP,43 (November 1995).
30. La Nacion, Oct. 4, 2000, p. 17.
31. Chevigny.
32. Centro de Estudios Legales y Sociales/Human Rights Watch (CELS/HRW), La Inseguridad
Policial. Violenciade las Fuerzas de Seguridaden la Argentina(Buenos Aires: EditorialUniversitariade
Buenos Aires, 1998); Chevigny.
33. Centro de Estudios Legales y Sociales (CELS), "La Continuidad de los Patrones de Violencia
Policial"(Buenos Aires: CELS, Archivos de Hoy, 2001).
34. S6rgio Adorno, in a personal conversation, suggested this way of describing the withdrawalof
legal protectionfrom those perceived to be marginais (outlaws).

18

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions
Daniel M. Brinks

35. See, for example, "ParaFavela, Policia Assusta Mais que Trafico,"Folha de Sdo Paulo, Feb. 4,
2001, p. C1; "Cuandono todos son iguales ante la ley,"La Naci6n, June 20, 2000; "La Justicia, vista por
la gente,"La Naci6n, June 24, 2000.
36. La Naci6n, Jan. 6, 2002, p.15.

19

This content downloaded from 130.64.11.153 on Wed, 02 Dec 2015 18:03:40 UTC
All use subject to JSTOR Terms and Conditions

You might also like