Confidence Interval

The QUANTREG procedure provides three methods to compute confidence intervals for the regression quantile parameter : sparsity, rank, and resampling. The sparsity method is the most direct and the fastest, but it involves estimation of the sparsity function, which is not robust for data that are not independently and identically distributed. To deal with this problem, the QUANTREG procedure computes a Huber sandwich estimate by using a local estimate of the sparsity function. The rank method, which computes confidence intervals by inverting the rank score test, does not suffer from this problem, but it uses the simplex algorithm and is computationally expensive with large data sets. The resampling method, which uses the bootstrap approach, addresses these problems, but at a computation cost.

Based on these properties, the QUANTREG uses a combination of the resampling and rank methods as the default. For data sets with more than either 5,000 observations or 20 variables, the QUANTREG procedure uses the MCMB resampling method; otherwise it uses the rank method. You can request a particular method by using the CI= option in the PROC statement.


Consider the linear model


and assume that , , are iid with a distribution F and a density , where in a neighborhood of . Under some mild conditions


where and Refer to Koenker and Bassett (1982).

This asymptotic distribution for the regression quantile can be used to construct confidence intervals. However, the reciprocal of the density function


which is called the sparsity function, must first be estimated.



can be estimated by the difference quotient of the empirical quantile function—that is,


where is an estimate of and is a bandwidth that tends to zero as .

The QUANTREG procedure provides two bandwidth methods. The Bofinger bandwidth


is an optimizer of mean squared error for standard density estimation, and the Hall-Sheather bandwidth


is based on Edgeworth expansions for studentized quantiles, where is the second derivative of and satisfies for the construction of confidence intervals. The quantity


is not sensitive to f and can be estimated by assuming f is Gaussian.

can be estimated by the empirical quantile function of the residuals from the quantile regression fit,


or the empirical quantile function of regression proposed by Bassett and Koenker (1982),


The QUANTREG procedure interpolates the first empirical quantile function and gets the piecewise linear version


is set to a constant if falls outside .

This estimator of the sparsity function is sensitive to the iid assumption. Alternately, Koenker and Machado (1999) considered the non-iid case. By assuming local linearity of the conditional quantile function in x, they proposed a local estimator of the density function by using the difference quotient. A Huber sandwich estimate of the covariance and standard error is computed and used to construct the confidence intervals. One difficulty with this method is the selection of the bandwidth when using the difference quotient. With a small sample size, either the Bofinger or the Hall-Sheather bandwidth tends to be too large to assure local linearity of the conditional quantile function. The QUANTREG procedure uses a heuristic bandwidth selection in these cases.

By default, the QUANTREG procedure computes non-iid confidence intervals. You can request iid confidence intervals with the IID option in the PROC statement.

Inversion of Rank Tests

The classical theory of rank tests can be extended to test the hypothesis : in the linear regression model . Here . See Gutenbrunner and Jureckova (1992) for more details. By inverting this test, confidence intervals can be computed for the regression quantiles that correspond to .

The rank score function can be obtained by solving the dual problem


For a fixed quantile , integrating with respect to the -quantile score function


yields the -quantile scores


Under the null hypothesis :


for large n, where .



Then from the constraint in the full model. In order to obtain confidence intervals for , a critical value can be specified for . The dual vector is a piecewise constant in , and can be altered without compromising the optimality of as long as the signs of the residuals in the primal quantile regression problem do not change. When gets to such a boundary, the solution does change, but can be restored by taking one simplex pivot. The process can continue in this way until exceeds the specified critical value. Since is piecewise constant, interpolation can be used to obtain the desired level of confidence interval; see Koenker and d’Orey (1994).


The bootstrap can be implemented to compute confidence intervals for regression quantile estimates. As in other regression applications, both the residual bootstrap and the -pair bootstrap can be used. The former assumes iid random errors and resamples from the residuals, while the later resamples pairs and accommodates some forms of heteroscedasticity. Koenker (1994) considered a more interesting resampling mechanism, resampling directly from the full regression quantile process, which he called the Heqf bootstrap.

In contrast with these bootstrap methods, Parzen, Wei, and Ying (1994) observed that


which is the estimating equation for the th regression quantile, is a pivotal quantity for the th quantile regression parameter . In other words, the distribution of can be generated exactly by a random vector , which is a weighted sum of independent, re-centered Bernoulli variables. They further showed that for large n, the distribution of can be approximated by the conditional distribution of , where solves an augmented quantile regression problem with observations with and sufficiently large for a given realization of u. By exploiting the asymptotically pivot role of the quantile regression "gradient condition," this approach also achieves some robustness to certain heteroscedasticity.

Although the bootstrap method by Parzen, Wei, and Ying (1994) is much simpler, it is too time-consuming for relatively large data sets, especially for high-dimensional data sets. The QUANTREG procedure implements a new, general resampling method developed by He and Hu (2002), which is referred to as the Markov chain marginal bootstrap (MCMB). For quantile regression, the MCMB method has the advantage that it solves p one-dimensional equations instead of p-dimensional equations, as do the previous bootstrap methods. This greatly improves the feasibility of the resampling method in computing confidence intervals for regression quantiles.