Skip to contents

The compensation for bearing the uncertainty of the claims, the risk adjustment of IFRS 17 or the risk margin of Solvency II, by one of two methods: a percentile of the distribution of the reserves, or the cost of the capital held until the claims are paid.

Usage

risk_adjustment(
  method = c("percentile", "cost of capital"),
  boot = NULL,
  level = 0.75,
  discounted = NULL,
  capital = NULL,
  coc_rate = 0.06
)

Arguments

method

"percentile" or "cost of capital".

boot

The bootstrap_chain_ladder of the reserves: needed for the percentile method, and for the confidence level of the cost of capital method.

level

The confidence level of the percentile method, such as 0.75.

discounted

The discount_reserves of the reserves: its payments give the run-off of the cost of capital method, its yield curve the discounting, and its ratio of discounted to undiscounted reserve scales the percentile method. Needed for the cost of capital method.

capital

The capital held now for the reserves, such as their Solvency II reserve risk (sii_premium_reserve_risk) or the bootstrap's 99.5% quantile less its mean. Needed for the cost of capital method.

coc_rate

The cost of capital rate a year: 6% under Solvency II before 2027.

Value

A list with the method, the risk adjustment, the confidence level (NA without the bootstrap for the cost of capital method), and for the cost of capital method the projected capital and its discounted cost by period.

Details

The percentile method takes the bootstrap's total reserves: the risk adjustment is their level quantile less their mean, and the confidence level is level.

The cost of capital method projects the capital capital held now over the run-off of the reserves, in proportion to the reserve still to be paid at the start of each period, and charges coc_rate of it for each period, discounted to now: \(RA = \code{coc\_rate} \sum_t C_t (1 + r_t)^{-t}\), as the Solvency II risk margin is found. With the bootstrap given too, the confidence level that IFRS 17 asks to disclose is the share of the simulated reserves at or below the mean plus the risk adjustment.

The discounting of the reserves applies to the risk adjustment too: with a discounted best estimate (discount_reserves), the percentile risk adjustment is scaled by the ratio of the discounted to the undiscounted reserve.

Examples

paid <- as_triangle(simulated_motor_claims, "accident_year", "development_year", "paid",
                    cumulative = FALSE)
boot <- bootstrap_chain_ladder(paid, n = 2000, seed = 1)
be <- discount_reserves(paid, yield_curve = 0.03)
risk_adjustment("percentile", boot = boot, level = 0.75, discounted = be)$risk_adjustment
#> [1] 610284.9
# the cost of 6% a year of the capital at 99.5%, over the run-off
capital <- boot$total[["p99.5"]] - boot$total[["ibnr"]]
ra <- risk_adjustment("cost of capital", boot = boot, discounted = be, capital = capital)
c(ra$risk_adjustment, ra$confidence_level)
#> [1] 360810.7193      0.6615