Regression Quantiles

Abstract

REGRESSION QUANTILES' A simple minimization problem yielding the ordinary sample quantiles in the location model is shown to generalize naturally to the linear model generating a new class of statistics we term "regression quantiles." The estimator which minimizes the sum of absolute residuals is an important special case. Some equivariance properties and the joint asymptotic distribution of regression quantiles are established. These results permit a natural generalization to the linear model of certain well-known robust estimators of location. Estimators are suggested, which have comparable efficiency to least squares for Gaussian linear models while substantially out-performing the least-squares estimator over a wide class of non-Gaussian error distributions.

What can local authorities do to improve the social care-related quality of life of older adults living at home? Evidence from the Adult Social Care Survey

“…To account for non-normality of the error terms and heteroscedasticity in the models the Huber-White sandwich estimator was used to adjust standard errors (Huber, 1967;White, 1980). Finally, to examine the robustness of the results based on the OLS technique, we also employed a quantile regression model, which was introduced by Koenker and Basset (Koenker & Bassett, Jr., 1978).…”

The impact of health worker migration on development dynamics: evidence of wealth effects from Africa

“…Thus, based on this technique we are able to carefully assess how HHR migration affects development throughout the conditional distribution with particular emphasis on the least and most developed African countries. QR yields parameters estimated at multiple points in the conditional distribution of the dependent variable (Koenker & Bassett, 1978) and has gained attention in recent development literature (Billger & Goel, 2009;Okada & Samreth, 2012;Asongu, 2012abc).…”

Decomposing body mass index gaps between Mediterranean countries: A counterfactual quantile regression analysis

“…The quantile regression (QR) model (Koenker and Basset, 1978), specifies the conditional quantile as a linear function of observed covariates. Following Buchinsky (1998), let Q u (wjX) for u 2 (0, 1) denote the uth conditional quantile of the distribution of (log)BMI (w), given a vector, X, of k covariates.…”

“…We accomplish this first step by means of quantile regressions (Koenker and Basset, 1978), namely, by estimating models for the quantiles of the conditional BMI distribution (measured in logarithms 4 ). Unlike ordinary least-squares methods that focus on the impact of covariates upon the mean, we model the impact of covariates upon different quantiles of the conditional distribution.…”