group of Europeans, the assignment of the treatment, i.e., a nation’s qualification, should
depend on the skills of players who participated in the qualification matches, so for some
European players selection into the treatment is not completely random at this stage;
however, even if some players’ outputs are indeed positively correlated with treatment
status, it seems reasonable to assume these are no longer correlated once we control for
constant skill/output differences between players by means of player fixed effects.
Our differences-in-differences analysis is based on the underlying assumption that de
facto qualification dates are relevant for determining the beginning of the treatment for
Euro 2008 - Europeans. One may argue though that even earlier on players from coun-
tries likely to qualify already exerted additional effort to increase their individual nom-
ination chances. We believe that our analysis is pertinent nonetheless. On the de facto
qualification date, there is a discrete upward jump in qualification probability (to 1) for
qualified nations, associated with a permanent increase in treatment intensity.27 The the-
ory predicts a positive reaction to such a permanent discrete change even if players were
already concerned with the nomination contest previously. In principle, we could esti-
mate pre-qualification treatment intensities combining the outcomes of completed qual-
ifications matches with national team ratings that determine winning probabilities for
upcoming qualification matches. A difficulty when using such intensities would be that
once we consider a longer time horizon, discounting, an issue we neglect in the present
analysis,28 should be quite important. Finally, it is unlikely that knowing more about
pre-qualification treatment intensities would allow us to yield additional insights concern-
ing what will be our main interest, the comparison of effects on players with different
characteristics.
As mentioned in the introduction, there are several possible explanations for why
an upcoming international Cup may have affect the outputs of players from qualified
nations. In order to discriminate between the nomination contest theory and alternative
explanations, we construct, as a proxy for player i’s anticipated nomination chance, the
following time-varying variable in [0, 1] measuring i’s national team exposure since World
27 Similarly, for all non-qualified nations there is a downward jump to zero at some point in time, in
many cases already long before the official qualification date. The group of players from these nations in
our sample is somewhat small (n = 23) to identify significant effects. Most Europeans from non-qualified
nations are from nations highly unlikely to participate in a Euro Cup; in particular, our sample contains
no players from England, a nation that was expected to qualify but in the end failed.
28 Testing for trends in the reaction to the Euro Cup treatment revealed no clear picture.
22