Power in the European Union: an evolutionary computing approach

Even the best existing model of legislative decisionmaking in the European Union, the compromise model, makes huge prediction errors when it is assumed that each actor’s power is determined by their formal voting weight. A few studies have attempted to improve the model’s predictive accuracy by examining alternative distributions of power, but extending their brute force approach poses daunting computational challenges. In this paper I illustrate how techniques from evolutionary computing can be employed to overcome these challenges. I then demonstrate the new possibilities that this approach opens up by identifying the relative power of each actor that best predicts policy outcomes from the EU-15 period. Some actors appear to punch significantly above or below their formal weight, with power varying dramatically across legislative procedures. My analysis highlights important unanswered questions about power in EU decisionmaking, and potentially indicates fundamental problems with the compromise model or the underlying data.