Ethnic Profiling in the European Union and the United States

Beyond the Directive, the general framework is the prohibition of discrimination in the European Convention on Human Rights read with the right to respect for private life, and the equivalent guarantees in the Charter of Fundamental Rights of the European Union. Both are the subject of a substantial case law of the European Court of Human Rights, which a student should look up rather than take from a summary.

Why it is contested in the United States

The American position differs in structure, and the difference is worth an answer.

The constitutional handles are two, and they operate differently. The Fourth Amendment governs stops and searches and asks whether they were reasonable. The Equal Protection Clause of the Fourteenth Amendment governs discrimination and asks whether the person was treated unequally on a prohibited ground.

The practical difficulty in American law is that an equal protection claim generally requires proof of discriminatory purpose and not merely of discriminatory effect. That is the reverse of the European Directive's approach to indirect discrimination, and it is why disparities in stop rates have been so much harder to litigate in the United States than the raw numbers would suggest.

Where American law has moved is through litigation about particular programmes, through consent decrees with individual police departments, through data collection requirements imposed on forces, and through State legislation. An Indian student should notice the pattern: the remedy has come from measurement and supervision of individual forces rather than from a single national rule.

Why profiling does not work

The efficiency argument is the one that persuades people who are not moved by the rights argument, and a good answer gives both.

It lowers the hit rate. Where a group is stopped at a much higher rate than others, the proportion of stops that find anything is generally lower for that group, not higher. That is the crucial measurement: it means officers are applying a lower threshold of suspicion to members of the profiled group, and a lower threshold produces worse decisions.

It is easily defeated. Any organised offender can select participants who do not fit the profile, so profiling displaces offending onto people the profile does not describe, while the police continue to look at people it does.

It wastes capacity. Stops that find nothing consume officer time that could have been used on behavioural indicators, which perform better.

It destroys cooperation. Policing depends on information from the public: complaints made, witnesses who come forward, communities that talk to officers. A group that experiences policing as harassment supplies less of all three, and the loss is invisible in any statistic of stops.

It is self-confirming. Concentrating attention on a group produces more detected offences among that group, which is then cited as the justification for the concentration. The statistic measures where the police looked, not where offending is.

The link to the previous two chapters

It is bias made operational. Chapter 520's implicit associations become a formal or informal practice.

It is structural inequality with an ethnic marker. Chapter 530's argument was that neutral criteria fall unequally by class; this is the same argument with ethnicity as the axis.

And it is what Bentham's third and fourth classes forbid. Punishment and enforcement that produce less benefit than harm, and that could be achieved by better means, are unprofitable and needless in his sense.

A worked example

A city's police introduce a stop and search operation in a commercial district after a rise in thefts. Officers are told to look for people who appear not to belong in the area.

Is this ethnic profiling? The instruction does not mention ethnicity, which is what makes it a good example. But "appears not to belong" is a criterion that officers will apply using dress, language, and appearance, and in operation it will fall on migrants and on the poor.

Apply the Directive's test. Is there an apparently neutral provision, criterion or practice? Yes, the instruction. Does it put persons of a racial or ethnic origin at a particular disadvantage? That is a question of measurement, answered by recording the ethnicity of those stopped and comparing it with the composition of people present in the district, which is the correct comparator and not the composition of the city. Is it objectively justified by a legitimate aim? Reducing theft is legitimate. Are the means appropriate and necessary? This is where the operation fails, if a criterion based on observed behaviour, such as trying door handles or watching a shop entrance, would achieve as much without the disparate effect.

And the efficiency test. Record the hit rate: what proportion of stops of each group produced any finding. If the rate is lower for the group stopped most, the operation is applying a lower threshold to that group and is producing worse decisions as well as unequal ones.

What a lawful operation would look like. Criteria stated in terms of behaviour; reasons recorded for each stop; ethnicity and outcome recorded and published; supervision of officers whose hit rates are outliers; and review of the operation against its own figures.

Distinctions that carry marks

What ethnic profiling is NOT

It is not all use of appearance. Acting on a witness's description of a particular offender identifies an individual and is not profiling.

It is not always deliberate. Much of it is produced by neutral instructions applied through implicit associations, which is why the indirect discrimination test matters more than any inquiry into intention.

It is not defended by the disparity in offending. Even if a group offended at a higher rate, that would not license stopping its members without individual suspicion, and the disparity itself is partly produced by where the police look.

It is not a purely foreign topic. The concept applies wherever policing uses group markers as proxies for suspicion, and every one of the analytical tools in this chapter, the hit rate, the correct comparator, the necessity test and the recording of reasons, can be applied in India.

Quick revision

1. Ethnic profiling is reliance on race, ethnicity, religion, national origin or language, without objective and reasonable justification, as a ground for stopping, searching, detaining or investigating.
2. The test: would the reason for the stop survive removing ethnicity from the description? If not, it is profiling.
3. Directive 2000/43/EC, article 2: direct discrimination is less favourable treatment on grounds of racial or ethnic origin; indirect discrimination is where an apparently neutral provision, criterion or practice puts persons of a racial or ethnic origin at a particular disadvantage, unless objectively justified by a legitimate aim and the means are appropriate and necessary.
4. Indirect discrimination is the operative concept, because forces rarely have a written instruction and always have a practice.
5. In the United States the Fourth Amendment governs the stop and the Equal Protection Clause the discrimination, and the requirement of discriminatory purpose makes disparate effect much harder to litigate. Remedies have come through data requirements and supervision of individual forces.
6. Why it does not work: it lowers the hit rate, which shows a lower threshold of suspicion is being applied; it is easily defeated; it wastes capacity; it destroys cooperation; and it is self-confirming.
7. A lawful operation: behavioural criteria, recorded reasons, recorded ethnicity and outcomes, published figures, supervision of outliers, and review against the numbers.

Test yourself

1. Define ethnic profiling and give the test that identifies it. It is reliance by law enforcement on race, ethnicity, religion, national origin or language, without objective and reasonable justification, as a basis for deciding to stop, question, search, detain or investigate. The test is whether the ground of suspicion is individual behaviour or category membership: if the reason for the stop would survive the removal of the person's ethnicity from the description, it is ordinary criminal profiling; if it would not, it is ethnic profiling.

2. How does European Union law reach a police practice that mentions nobody's ethnicity? Through the definition of indirect discrimination in article 2 of Council Directive 2000/43/EC, under which discrimination occurs where an apparently neutral provision, criterion or practice would put persons of a racial or ethnic origin at a particular disadvantage compared with others, unless that provision, criterion or practice is objectively justified by a legitimate aim and the means of achieving it are appropriate and necessary. Since a force will rarely have a written instruction naming a group but will have practices that fall unequally in operation, this definition rather than the definition of direct discrimination does the work, and the word "necessary" requires the authority to show that no less discriminatory means would achieve the aim.

3. Why has profiling been harder to challenge in the United States? Because the two constitutional handles operate differently and neither maps onto disparate effect. The Fourth Amendment asks whether a stop or search was reasonable, which is a question about the individual encounter. The Equal Protection Clause reaches discrimination but generally requires proof of discriminatory purpose rather than discriminatory effect, which is the reverse of the European approach to indirect discrimination. The result is that statistical disparity, however large, does not by itself establish a claim, and reform has come instead through litigation against particular programmes, consent decrees, data collection requirements and State legislation.

4. Explain the hit rate argument. The hit rate is the proportion of stops that produce any finding. If a group is stopped far more often than others and yet a smaller proportion of those stops finds anything, the difference cannot be explained by that group offending more; it shows that officers are applying a lower threshold of suspicion to its members. A lower threshold means worse decisions, so profiling reduces the accuracy of policing at the same time as it increases its unfairness, which is why the efficiency argument and the rights argument point the same way.

5. Why is profiling described as self-confirming? Because concentrating attention on a group produces more detected offences among that group, and the resulting statistics are then cited as the justification for the concentration. What such a statistic measures is where the police looked rather than where offending occurred. The way to break the circle is to record the ethnicity and outcome of every stop, to compare stops against the composition of the population actually present in the area rather than the whole city, and to publish the hit rates, since only then does the disparity become visible as a fact about policing rather than about offending.