We consider the weighting scheme that results in a best-case scenario in the construction of the world governance indicators (WGIs), a proxy of institutional quality. To perform that, we use an approach that relies on consistent tests for stochastic dominance efficiency of a given index with respect to all possible indices constructed from a set of individual components. The test statistics and the estimators are computed using mixed integer programming methods. The results show that the equally weighted (fixed weights) composite WGI index is not the best-case scenario and that governance indicators at different years should be weighted differently. Furthermore, we revisit the institutions hypothesis in the empirical growth literature, where institutional quality is the main determinant of long-term development. We find that not only do institutions matter for economic development but also geography and macroeconomic policies do affect economic development directly.