6.5.1 Finding a link between the characteristics of the cases and the treatments

The ‘many-variables approach’ is aimed at finding a (causal) link between the characteristics of the case and the characteristics of the treatment. If there is a link between a certain characteristic of the case and the treatment, then this link should be present in all bank State aid cases where that particular characteristic is present. Otherwise, the principle of equal treatment will be violated.

This approach entails the following steps. The first step would consist of mapping all the relevant characteristics of the banks that received State aid.1 The next step would consist of mapping all the relevant characteristics of the State aid measures granted to those banks.2 Together, the characteristics of the bank and the characteristics of the State aid measures constitute the characteristics of the “case”. The third step would be to list all the characteristics of the “treatment” (i.e. all the modalities of the restructuring measures). The last step would be to link the characteristics of the case with the characteristics of the treatment. How can such a link be established?

Ideally, one should find two cases that are completely similar in all aspects except one. In other words: there is one characteristic on which the two cases differ; on all the other characteristics, the two cases score the same. So if the treatment of these two cases is different, the difference in treatment can only be caused by that one characteristic on which they differ. This way, the influence of that particular characteristic on the treatment can be established. In statistical terms, if a dependent variable (Y) is influenced by several independent variables (X1, X2, X3), then the influence of variable X1 can only be established, if one controls for the influence of the other independent variables (X2 and X3).

This approach can only work if there are cases which are completely similar in all aspects but one. However, in reality there are so many differences between cases, that one can never find two cases which differ only on one characteristic. So the isolated influence of a particular characteristic cannot be established. To conclude, there is no statistical way to establish a (causal) link between the characteristics of the case and the characteristics of the treatment.