Skip to contents

The Paid-incurred Chain model (Merz, Wuthrich (2010)) combines claims payments and incurred losses information to get a unified ultimate loss prediction.

Usage

PaidIncurredChain(triangleP, triangleI)

Arguments

triangleP

Cumulative claims payments triangle

triangleI

Incurred losses triangle.

Value

The function returns:

  • Ult.Loss.Origin Ultimate losses for different origin years.

  • Ult.Loss Total ultimate loss.

  • Res.Origin Claims reserves for different origin years.

  • Res.Tot Total reserve.

  • s.e. Square root of mean square error of prediction for the total ultimate loss.

Details

The method uses some basic properties of multivariate Gaussian distributions to obtain a mathematically rigorous and consistent model for the combination of the two information channels.

We assume as usual that I=J. The model assumptions for the Log-Normal PIC Model are the following:

  • Conditionally, given \(\Theta = (\Phi_0,...,\Phi_I, \Psi_0,...,\Psi_{I-1},\sigma_0,...,\sigma_{I-1},\tau_0,...,\tau_{I-1})\) we have

    • the random vector \((\xi_{0,0},...,\xi_{I,I}, \zeta_{0,0},...,\zeta_{I,I-1})\) has multivariate Gaussian distribution with uncorrelated components given by $$\xi_{i,j} \sim N(\Phi_j,\sigma^2_j),$$ $$\zeta_{k,l} \sim N(\Psi_l,\tau^2_l);$$

    • cumulative payments are given by the recursion $$P_{i,j} = P_{i,j-1} \exp(\xi_{i,j}),$$ with initial value \(P_{i,0} = \exp (\xi_{i,0})\);

    • incurred losses \(I_{i,j}\) are given by the backwards recursion $$I_{i,j-1} = I_{i,j} \exp(-\zeta_{i,j-1}),$$ with initial value \(I_{i,I}=P_{i,I}\).

  • The components of \(\Theta\) are independent and \(\sigma_j,\tau_j > 0\) for all j.

Parameters \(\Theta\) in the model are in general not known and need to be estimated from observations. They are estimated in a Bayesian framework. In the Bayesian PIC model they assume that the previous assumptions hold true with deterministic \(\sigma_0,...,\sigma_J\) and \(\tau_0,...,\tau_{J-1}\) and $$\Phi_m \sim N(\phi_m,s^2_m),$$ $$\Psi_n \sim N(\psi_n,t^2_n).$$ This is not a full Bayesian approach but has the advantage to give analytical expressions for the posterior distributions and the prediction uncertainty.

Note

The model is implemented in the special case of non-informative priors.

References

Merz, M., Wuthrich, M. (2010). Paid-incurred chain claims reserving method. Insurance: Mathematics and Economics, 46(3), 568-579.

Author

Fabio Concina, fabio.concina@gmail.com

Examples

PaidIncurredChain(USAApaid, USAAincurred)
#> $Ult.Loss.Origin
#>            [,1]
#>  [1,]  983113.3
#>  [2,] 1078696.6
#>  [3,] 1145761.4
#>  [4,] 1245171.4
#>  [5,] 1371964.4
#>  [6,] 1433857.4
#>  [7,] 1415963.7
#>  [8,] 1410065.3
#>  [9,] 1320414.5
#> 
#> $Ult.Loss
#> [1] 11405008
#> 
#> $Res.Origin
#>              [,1]
#>  [1,]    965.2681
#>  [2,]   3159.6104
#>  [3,]   7386.4026
#>  [4,]  18521.4454
#>  [5,]  47232.3846
#>  [6,] 113727.4102
#>  [7,] 230663.7420
#>  [8,] 443903.3046
#>  [9,] 778393.5002
#> 
#> $Res.Tot
#> [1] 1643953
#> 
#> $s.e.
#> [1] 113940.2
#>