The nature of numerous strategies of firms is often discrete or countable. This adds difficulty to measuring and testing for the existence of complementarities among several strategies. This paper introduces a generalized multivariate count data model that allow estimating correlations of any sign among the pricing decisions of competing firms in a manner that is robust to the existence of unobserved heterogeneity leading to either over- and underdispersion of the distribution of counts. Thus, it is possible to overcome a major challenge in testing whether two decisions are strategic complements or substitutes, namely, dealing with the effect of unobserved heterogeneity. I study how firms actually compete in nonlinear tariffs by analyzing the interrelation between the incumbent and entrant's decisions to offer a given number of tariff options. Results document the existence of complementarity among tariff options regardless of whether they are dominated or not. This result supports the view that the implementation of nonlinear tariffs by means of a menu of self-selecting two-part tariffs has some strategic competitive value.