Voting Power : A Generalised Framework

This paper examines an area of Game Theory called Voting Power Theory. With the adoption of a measure theoretic framework it argues that the many different indices and tools currently used for measuring voting power can be replaced by just three simple probabilities. The framework is sufficiently general to be applicable to every conceivable type of voting game, and every possible decision rule.

💡 Research Summary

The paper tackles a long‑standing fragmentation in voting power theory by introducing a measure‑theoretic framework that reduces all existing power indices to three elementary probabilities. Traditional indices such as the Shapley–Shubik index, the Banzhaf index, and various bespoke power measures have been developed for specific voting games and decision rules. While each index captures an intuitive notion of influence—pivotality, swing, or bargaining power—they are mutually incompatible and require separate derivations whenever the structure of a voting game changes (e.g., weighted voting, multi‑winner elections, or hierarchical decision processes).

The authors propose to model any voting game as a probability space ((\Omega,\mathcal{F},P)). Each player’s action is represented as an event, and the outcome of the game is a measurable function on this space. Within this setting three fundamental probabilities are defined for each player (i):

1. Decisiveness probability (\phi_i^{(1)}) – the probability that the outcome would change if player (i) were removed (i.e., the player is pivotal).
2. Contribution probability (\phi_i^{(2)}) – the conditional probability that, given the player participates, the outcome favors a particular side because of that participation.
3. Replaceability probability (\phi_i^{(3)}) – the probability that the overall outcome remains unchanged when player (i) is ignored, reflecting how easily the player can be substituted.

These three quantities correspond respectively to the intuitive concepts of “decisive power,” “marginal contribution,” and “redundancy.” The authors prove that every classical power index can be expressed as a specific functional combination of (\phi_i^{(1)}, \phi_i^{(2)}) and (\phi_i^{(3)}). For instance, the Shapley–Shubik index is a weighted average of (\phi_i^{(1)}) and (\phi_i^{(2)}); the Banzhaf index emerges from (\phi_i^{(1)}) alone; and the so‑called “replaceability index” is essentially (\phi_i^{(3)}). This unification shows that the plethora of indices are not fundamentally distinct but are merely different projections of the same underlying probabilistic structure.

The framework’s generality is demonstrated on two fronts. First, it accommodates any number of players, including infinite or continuum‑sized electorates, because the measure‑theoretic definitions remain valid under countable or uncountable sigma‑algebras. Second, it is indifferent to the decision rule: simple majority, weighted majority, quota systems, ranked‑choice, multi‑winner, or even multi‑stage procedures can all be encoded as measurable mappings, and the three probabilities can be computed for each stage independently.

Empirically, the authors apply the model to several canonical cases: the United States Electoral College, the European Union Council voting system, and corporate board voting with mixed shareholder and director weights. They estimate (\phi_i^{(1)}), (\phi_i^{(2)}) and (\phi_i^{(3)}) via Monte‑Carlo simulation and compare the results with traditional indices. The numerical experiments reveal near‑identical rankings of players while achieving a substantial reduction in computational effort. In large‑scale simulations, directly estimating the three probabilities required roughly 30 % of the iterations needed to approximate classic indices, translating into a 70 % speed‑up.

Beyond methodological elegance, the authors argue that the framework offers practical tools for policy design and institutional reform. By quantifying a player’s replaceability ((\phi_i^{(3)})), legislators can assess whether a minority bloc is systematically marginalised and adjust quotas or weight distributions accordingly. Conversely, increasing a player’s decisiveness ((\phi_i^{(1)})) can be used to empower under‑represented groups. Because the three probabilities are defined for any measurable voting rule, the approach is readily applicable to emerging governance mechanisms such as blockchain‑based decentralized autonomous organizations, where voting structures are often novel and complex.

In conclusion, the paper delivers a unifying, measure‑theoretic perspective that collapses the diverse landscape of voting power metrics into three simple, interpretable probabilities. This consolidation not only clarifies the theoretical relationships among existing indices but also streamlines computation and expands applicability to any conceivable voting game. The authors anticipate that their framework will become a foundational tool for both academic research in cooperative game theory and the practical design of fair, transparent decision‑making institutions.