A Comparative Analysis of Anti-Discrimination and Data Protection Legislations - Discrimination and Privacy in the Information Society - page 77

entitlements under the Community pension scheme would be wholly
unacceptable, and I cannot see that the use of the criterion of sex rather
than ethnic origin can be more acceptable.”
Recently, the ECJ addressed this kind of discrimination in Test-Achats (ECJ, C-
236/09, 1 March 2011), wherein the Belgian consumer organisation contested the
validity of art. 5(2) of the Gender Goods and Services Directive . Whereas art. 5(1)
prohibits “the use of sex as a factor in the calculation of […] individuals' premium
and benefits”, article 5(2) permitted member states to create legal provisions
derogating this prohibition when sex is a “determining factor” and when the risk
assessment is “based on relevant and accurate actuarial and statistical data.” The
ECJ declared the derogation of article 5(2) incompatible with gender equality and
invalid with effect from 21 December 2012. The decision caused an enormous stir
in the insurance sector. Possibly, the decision will lead to the use of proxy factors
(such as profession, education, lifestyle, etc.) in assessing risk, which in turn
might raise questions of indirect discrimination.
Though interesting, investigating these issues in more detail is beyond the
scope of this chapter. However, what is relevant for us to note here is that in the
Test-Achats case the proceedings were completely based on anti-discrimination
law, and do not relate to data protection at all. This can be explained by the facts
that the claimant was a consumer organisation and not an individual data subject,
and that the case did not concern an individual instance of differential treatment
but posed a direct challenge to a piece of AD legislation. Looking at the Test-
Achats case it is interesting to speculate whether the data processing related to the
gendered differentiation of insurance fees, had it been contested, would have been
considered legitimate from a data protection perspective. First, it is not even
crystal clear that DP can apply to this type of situation, since the question of
whether this type of data qualifies as “personal” in the meaning of Directive
95/46/EC is controverted. It will qualify as such if one refers to the
aforementioned opinion of the WP29 . However, this opinion is not uncontested,
and in any event not binding. Second, provided this insurance contract can be
considered as a legitimate aim to be pursued (art. 7), this would depend on
whether such processing is necessary to the performance of the insurance contract
(art. 6 DP ).
Would the applicability of DP be of any help? Often statistical discrimination
will not concern any of the protected grounds, rather, attributes such as income,
postal code, browsing behaviour, type of car, etc., or complex algorithmic
combinations of several attributes. AD could be eventually be resorted to if it
could be shown that any of these attributes, or algorithmic combinations of these
attributes, were used as proxies for any protected ground (indirect discrimination).
However, were this is not to be the case then, once more, the “art. 6 DP + general
equality”-route could prove to be a useful tool to supplement the limitative list of
protected grounds in AD .