Here we develop a method for performing nonparametric Bayesian inference on quantiles.
Relying on geometric measure theory and employing a Hausdor base measure, we are able to specify meaningful priors for the quantile while treating the distribution of the data otherwise nonparametrically. We further extend the method to a hierarchical model for quantiles of subpopulations, linking subgroups together solely through their quantiles. Our approach is computationally straightforward, allowing for censored and noisy data. We demonstrate the proposed methodology on simulated data and an applied problem from sports statistics, where it is observed to stabilize and improve inference and prediction.

Much research and policy attention has been on socio economic gaps in participation at university, but little attention has been paid to gaps in earnings. This paper addresses this shortfall using tax and student loan administrative data to investigate the earnings of English graduates up to their mid thirties by socio economic background. We find that  graduates from higher income families (from the top fth of the income distribution of those enrolled in university) have average earnings which are 20% higher than those from lower income families. Once we condition on institution and subject choices, this premium roughly halves, to around 10%. The premium grows with age and is larger for men, in particular for men at the most selective universities. We follow Chetty et al. (2017) and estimate English mobility scorecards by university and subject, highlighting the good performance of medicine, economics, law, business, engineering, technology, math, computer science and architecture courses as well as the prominent London-based universities.

We define causal estimands for experiments on single time series, extending the potential outcome framework to dealing with temporal data.  Our approach allows the estimation of a broad class of these estimands and exact randomization based p-values for testing causal effects, without imposing stringent assumptions.  We further derive a general central limit theorem that can be used to conduct conservative tests and build confidence intervals for causal effects.  Finally, we provide three methods for generalizing our approach to multiple units that are receiving the same class of treatment, over time.  We test our methodology on simulated "potential autoregressions,"which have a causal interpretation.  Our methodology is partially inspired by data from a large number of experiments carried out by a financial company who compared the impact of two different ways of trading equity futures contracts.  We use our methodology to make causal statements about their trading methods.