confint.PLR {LorenzRegression}

Confidence intervals for the Penalized Lorenz Regression

Description

confint.PLR provides confidence intervals for the explained Gini coefficient and Lorenz-R2 for an parm of class PLR.

Usage

## S3 method for class 'PLR'
confint(
  object,
  parm = c("Gini", "LR2"),
  level = 0.95,
  boot.method = c("Param", "Basic", "Perc"),
  which.pars = NULL,
  ...
)

Arguments

object

Output of a call to Lorenz.Reg, where penalty!="none" and Boot.inference=TRUE.

parm

Determines whether the confidence interval is computed for the explained Gini coefficient or for the Lorenz-R2. Possible values are "Gini" (default, for the explained Gini) and "LR2" (for the Lorenz-R2).

level

level of the confidence interval

boot.method

What bootstrap method is used to construct the confidence interval. Default value is "Param", which exploits the asymptotic normality and only bootstraps the variance. Other possible values are "Perc" (percentile bootstrap) and "Basic" (basic bootstrap). Percentile bootstrap directly plugs the quantiles of the bootstrap distribution. Basic bootstrap is based on bootstrapping the whole distribution of the estimator.

which.pars

Which values of the bandwidth h and the penalty parameter lambda should be used. Default is NULL, in which case the optimal values are used.

...

Additional arguments.

Details

Use this function only if Boot.inference was set to TRUE in the call to Lorenz.Reg. Otherwise, bootstrap was not computed and the confidence intervals cannot be determined.

Value

A matrix gathering the desired confidence intervals. Each row corresponds to a different selection method for the pair (h,lambda).

See Also

Lorenz.Reg

Examples

data(Data.Incomes)
set.seed(123)
Data <- Data.Incomes[sample(1:nrow(Data.Incomes),50),]
PLR <- Lorenz.Reg(Income ~ ., data = Data, h.grid = nrow(Data)^(-1/5.5),
                  penalty = "SCAD", eps = 0.02, seed.boot = 123, B = 40, Boot.inference = TRUE)
confint(PLR)