The statistics of test Tn show that the best model appears to be the model 7. Thus, our empirical
evidence shows that, on this market of bottled water, manufacturers and retailers use two part
tariffs contracts without resale price maintenance (RPM) and this model also indicates that the
buyer power of retailers is affected endogenously by their outside opportunities.
Concerning this inference on the vertical relationship, other variants tested were also rejected.
For example, all the variants of two part tariffs contracts without RPM where uniform wholesale
pricing is imposed were also rejected, as well as models of collusion between manufacturers or
between retailers.
Let’s comment more on the preferred model. This is a model with two part tariff contract, and
no RPM, which contrasts with what was found in Bonnet and Dubois (2010) on the 1998-2000
period where RPM was found. It is interesting because in 2005, the Galland act was removed and
replaced by another law in order to redefine resale at loss by retailers and prevent the use of high
list wholesale prices to implement RPM. Actually, RPM is in principle forbidden in France but
the evidence found in Bonnet and Dubois (2010) was consistent with the worries of the compe-
tition authority that the Galland act (in force between 1996 and 2005) allowed manufacturers to
implement RPM equilibrium. Indeed, the definition of thresholds for resale at loss did not take
into account backward margins and only wholesale unitary list prices which could be set as high
as wanted to enforce minimum retail prices, while compensating retailers with backward margins.
After 2005, this became impossible because the definition of minimum retail prices to define resale
at loss did include part of the backward margins. It seems that the change in the law did succeed
in avoiding manufacturers to mimic RPM.
Also, for this preferred model, Table 4 shows that the average price-cost margins are of 35.9%
for mineral water and 63.3% for spring water. In absolute values, the price-cost margins are on
average 0.1∙3C for mineral water and 0.09C for spring water because mineral water is on average
more expensive. For this best model, the average total price-cost margins for national brands is
48.2% while it is of 26.4% for private labels. Remark that the high average margin for national
32