A generating functions approach for computing the Public Good index efficiently

In the past years, a combinatorial method based on generating functions was introduced to compute Shapley–Shubik, Banzhaf and other indices for weighted majority games exactly and efficiently. In this paper, taking inspiration from what has already been done, in view of the efficiency of the generating functions method, we define a generating function for computing the Public Good index, maintaining the property of exactness of the resulting algorithm. The main difference with the existing algorithms derives from the fact that the Public Good index takes into account only minimal winning coalitions and counts how many swings of a player involve them. Moreover, we study the computational complexity of the algorithm and we evaluate the Public Good index for the vote share of the Russian Duma in 1995.