BootMackChainLadder: Boot Mack Chain Ladder model

In lrnv/mbmcl: Multi Boot Mack Chain Ladder boostrap the Braun model in claim reserving

Description

This function implement a simple bootstrap of the residuals from the mack model with a one-year reserving risk point of view.

Usage

BootMackChainLadder(Triangle, B = 100, distNy = "normal", seuil = NA,
  zonnage = FALSE, BF.premiums = NULL, BF.param = c(5, 5),
  stab = NA)

Arguments

Triangle	A simple triangle from che Chain-ladder package.
B	numeric. The number of bootstrap samples you want
distNy	character. Distribution of next-year incremental payments. Either "normal" (default) or "residuals"
seuil	numeric. A value of NA (default) will prevent exclusion of residuals, and a numerci value (e.g 2) will exclude all residuals that have an absolution value greater than 2.
zonnage	logical. Do you want to force residuals to be resampled inside zonnes ?
BF.premiums	If a Bornhuetter-fergusson is needed, input a vector of ultimates premiums here. Otherwise, the BF code will not be triggered.
BF.param	A vector of 2 interger that represent (respectively) the number of year of averaging Loss-ratios for the bornhuetter fergusson and then the number of year of applying the bornhuetter fergusson to.

Details

The bootstrap that is implemented here consist in a resampling of residuals obtained by the Mack model (or simulated standard normal residuals if you choose so), and on thoose samples we construct a one-year point of view of the mack model, allowing us to bootstrap one-year quantities like the CDR or next year IBNRS. Using thisfunction properly, you could check that the proposed bootstrap is convergent with the merz-wuthrich formula if you take standard normal r<c3><a9>siduals, but not otherwise.

Value

A BootMackChainLadder object with a lot of information about the bootstrapping. You can plot it, print it and str it to extract information.

Examples

data(ABC)
BootMackChainLader(Triangle = ABC, B = 100, distNy = "residuals", seuil = 2)

lrnv/mbmcl documentation built on May 24, 2019, 2:52 p.m.