Unconditional and conditional quantile regressions are used to explore the determinants of labour earnings at different parts of the distribution and, hence, the determinants of overall labour earnings inequality. The analysis combines several household surveys to provide comparable estimates for 32 countries. The empirical work suggests that, in general, a rise in the share of workers with an upper-secondary or post-secondary non-tertiary degree, a rise in trade union membership, a rise in the share of public employment and a rise in the share of workers on permanent contracts are associated with a narrowing of the earnings distribution. By contrast, a shift in the sector composition of the economy is not found to have a large impact on overall earnings inequality. As for tertiary education, the impact remains ambiguous as there are several offsetting forces.
