ER_Jan25.csv

<pre>The objective of this paper is to propose a nonparametric bootstrap-based confidence sets for the quantile ratio inequality index. Complications for inference on quantiles stem from their studentization as it inherits the well-known kernel density estimation problems and complications arising from the use of the delta method. The proposed method, which is based on bootstrapping the ratio directly, circumvents these difficulties as well as complications arising from the irregularities of the underlying distribution. Simulation results show that the bootstrap method admits almost exact coverage even with extreme quantiles drawn from heavy-tailed distributions with samples as small as 50 observations. To illustrate the empirical relevance of the proposed method, an application to world inequality is considered. Salient results show that the standard Wald-type confidence sets could lead to misleading conclusions and, subsequently, potential misguided policy actions.</pre>