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...