Reserves from an a priori expected loss ratio: the latest claims plus the share of the expected ultimate still to emerge, by the chain ladder's development pattern.
Arguments
- triangle
A triangle of cumulative claims, as for
chain_ladder, or the result ofchain_ladder(), whose development pattern is then used.The premium (or other exposure) of each origin.
- loss_ratio
The a priori expected loss ratio: one for all origins or one for each, as a share of the premium (0.7 for 70%).
- iterations
1 (the default) for Bornhuetter-Ferguson, 2 for Benktander.
- ...
Arguments passed to
chain_ladder()for the development pattern, such asaverageandtail, whentriangleis a triangle.
Value
An object of class netsimr_reserve: a list with by_origin,
a data frame with the latest claims, premium, a priori loss ratio, share
developed, a priori ultimate, ultimate, IBNR and ultimate loss ratio of each
origin, total, the same for all origins together, and
chain_ladder, the development pattern.
Details
With \(p_i = 1 / CDF_i\) the share developed of origin \(i\) (the inverse of its chain ladder cumulative development factor, with the tail) and \(U_0 = P_i \times LR_i\) the a priori ultimate, each iteration gives \(U_{k} = C_i + (1 - p_i) U_{k-1}\). One iteration is the Bornhuetter-Ferguson method, two the Benktander (iterated Bornhuetter-Ferguson) method, and as the iterations increase the ultimate tends to the chain ladder's.
References
Bornhuetter, R. L. and Ferguson, R. E. (1972). The actuary and IBNR. Proceedings of the Casualty Actuarial Society 59, 181-195.
Mack, T. (2000). Credible claims reserves: the Benktander method. ASTIN Bulletin 30(2), 333-347.
See also
cape_cod, which estimates the loss ratio from the
triangle, and chain_ladder.
Examples
premium <- c(50000, 52000, 54000, 56000, 58000, 60000, 62000, 64000, 66000, 68000)
bornhuetter_ferguson(raa_triangle, premium, loss_ratio = 0.35)
#> Bornhuetter-Ferguson
#>
#> origin latest premium loss_ratio developed prior_ultimate ultimate ibnr
#> 1981 18,834 50,000 0.350 1.000 17,500 18,834 0
#> 1982 16,704 52,000 0.350 0.991 18,200 16,870 166
#> 1983 23,466 54,000 0.350 0.974 18,900 23,950 484
#> 1984 27,067 56,000 0.350 0.943 19,600 28,184 1,117
#> 1985 26,180 58,000 0.350 0.905 20,300 28,108 1,928
#> 1986 15,852 60,000 0.350 0.813 21,000 19,782 3,930
#> 1987 12,314 62,000 0.350 0.694 21,700 18,959 6,645
#> 1988 13,112 64,000 0.350 0.546 22,400 23,284 10,172
#> 1989 5,395 66,000 0.350 0.336 23,100 20,728 15,333
#> 1990 2,063 68,000 0.350 0.112 23,800 23,195 21,132
#> Total 160,987 590,000 0.350 206,500 221,894 60,907
#> ultimate_loss_ratio
#> 0.377
#> 0.324
#> 0.444
#> 0.503
#> 0.485
#> 0.330
#> 0.306
#> 0.364
#> 0.314
#> 0.341
#> 0.376
# Benktander, with the chain ladder of simple averages
cl <- chain_ladder(raa_triangle, average = "simple")
bornhuetter_ferguson(cl, premium, loss_ratio = 0.35, iterations = 2)$total
#> latest premium loss_ratio prior_ultimate
#> 1.609870e+05 5.900000e+05 3.500000e-01 2.065000e+05
#> ultimate ibnr ultimate_loss_ratio
#> 2.271235e+05 6.613646e+04 3.849550e-01
