Measuring Central Bank Independence : A Table of Subjectivity and of Its Consequences

The paper is organised around three objectives. First, it scrutinises closely each of the two indices of Central Bank Independence (CBI) most commonly used in the empirical literature and quantifies their "subjectivity" bias. It defines and discovers an impressive interpretation bias, a major criteria bias but a negligible weighting bias in those indices. Second, it examines the robustness, with respect to the particular index of CBI used, of the "common knowledge" on the benefits of CBI that previous empirical studies have gradually built. It finds that, when rankings produced by various CBI indices (rather than their subjectivity-prone values) are regressed with average inflation, average growth or their respective variance, as much as 87.5% of the regression coefficients are not statistically significant. Third, following recent theoretical developments which insist on the central role of the determinants of incentives, it suggests an alternative approach to the measurement of a Central Bank's operational status.