actuar: An R Package for Actuarial Science

Vincent Goulet

École d’actuariat, Université Laval

 Abstract

The actuar project is a package of Actuarial Science functions for the R sta-
tistical system. The project was launched in 2005 and the package is avail-
able on CRAN (Comprehensive R Archive Network) since February 2006.
The current version of the package contains functions for use in the fields
of risk theory, loss distributions and credibility theory. This paper presents
in detail but in non technical terms the most recent version of the package.

Keywords: R, actuar, loss distributions, risk theory, credibility theory, soft-

ware, programming language.
1 Introduction
R is a programming language and an environment for statistical computing
and graphics (R Development Core Team, 2007). It is a free software version
of the award winning S system whose commercial counterpart is S-Plus by
Insightful Corporation. Despite some important fundamental differences
between S-Plus and R, the two systems are largely compatible and most
code written for one runs unaltered in the other.
R can also be considered as a dialect of the S programming language, a
language for “programming with data” developed at Bell Laboratories start-
ing from the late 1970s. R is not just another statistical environment (like
SPSS or SAS, for example), but a full fledged and self contained program-
ming language with a strong mathematical orientation. It is based on the
notion of vector that the many actuaries who have worked, or are still work-
ing, with APL or SAS IML have come to know and love.
R is under very active development and is constantly gaining “market
shares” in many fields related to statistics. One of its strengths is the
possibility to add functionalities to the base system by means of add-
on packages. Simply put, a package is a coherent set of functions and
data. The basic R systems consists of a dozen packages; hundreds of
contributed packages covering a wide array of modern statistical methods
are available from the Comprehensive R Archive Network (CRAN; http:
//cran.r-project.org).
The actuar project (Goulet, 2007) is a package of Actuarial Science func-
tions for R. Although various packages on CRAN provide functions that
may be of use to actuaries, actuar aims to serve as a central location for
more specifically actuarial functions and data sets. The project was offi-
cially launched in 2005 and is under active development.
This paper reviews, in non technical terms, the various features of the
current version of the actuar package.

2 Current status of the package

As of this writing, the version of actuar available on CRAN is 0.9-3. The
feature set of the package can be split in three main categories: loss distri-
butions modeling, risk theory and credibility theory.
As much as possible, the developers have tried to keep the “user in-
terface” of the various functions of the package consistent. Moreover, the
package follows the general R philosophy of working with model objects.
This means that instead of merely returning, say, a vector of probabilities,
many functions will return an object containing, among other things, the
said probabilities. The object can then be manipulated at one’s will using
various extraction, summary or plotting functions. This allows for a very
dynamic modeling–estimation–diagnosis–prediction process that few other
statistical packages provide.

1
 The package is released under the GNU General Public License (GPL),
version 2 or newer, thereby making it free software that anyone can use,
modify and redistribute, to the extent that the derivative work is also re-
leased under the GPL.

3 Documentation
It is a requirement of the R packaging system that every function and data
set in a package has a help page. The actuar package follows this require-
ment strictly. The help page of function foo is accessible by typing

> ?foo

or

> help("foo")

at the R command prompt. Most help pages provide usage examples.

In addition to the help pages, the package includes vignettes, longer PDF
documents on one or many topics. Running

> vignette(package = "actuar")

will give the list of available vignettes in the package.

Finally, one will find more comprehensive examples for the various fea-
tures of the package in the demo scripts (see ?demo). The list of demos
available in the package is given by

> demo(package = "actuar")

4 Loss distributions modeling features

Loss distributions is the subset of actuar containing the largest number
of functions. Some complement features of base R, while others provide
support for entirely untouched procedures common in Actuarial Science.
The following subsections detail the following actuar features:

1. introduction of 17 additional probability laws and functions to get the

kth true raw and limited moments;

2. fairly extensive support of grouped data;

3. calculation of the empirical raw and limited moments;

4. minimum distance estimation using three different measures;

5. treatment of coverage modifications (deductibles, limits, inflation, coin-

surance).

2
4.1 Probability laws
R already includes functions to compute the density function, cumulative
distribution function, quantile function of, and to generate variates from
a fair number of probability laws. For some root foo, the functions are
named dfoo, pfoo, qfoo and rfoo, respectively.
The actuar package provides d, p, q and r functions for all the probabil-
ity laws useful for loss severity modeling found in Appendix A of Klugman
et al. (2004) and not already present in base R, excluding the inverse Gaus-
sian and log-t but including the loggamma distribution (Hogg and Klugman,
1984). Among others, most welcome additions are functions for the Pareto
distribution.
Tables 1–3 list the supported distributions classified by families. Each
table details the names of the distributions as given in Klugman et al.
(2004), the root name of the R functions and the names of the arguments
corresponding to each parameter in the parametrization of Klugman et al.
(2004). One will note that by default all functions (except those for the
Pareto distribution) use a rate parameter equal to the inverse of the scale
parameter. This differs from Klugman et al. (2004) but is better in line with
the functions for the gamma, exponential and Weibull distributions in base
R.
All functions are written in C for speed. In almost every respect, they
behave just like the base R functions.
In addition to the d, p, q and r functions, the package provides m and
lev functions to compute the theoretical raw and limited moments, re-
spectively. All the probability laws of Tables 1– 3 are supported, plus the
following ones already in R: exponential, gamma, lognormal and Weibull.
The m and lev functions come especially useful with estimation methods
based on the matching of raw or limited moments. See Subsection 4.4 for
their empirical counterparts.

4.2 Grouped data

Grouped data is data represented in an interval-frequency manner. Typ-
ically, a grouped data set will report that there where nj claims in the
interval (cj−1 , cj ], j = 1, . . . , r . This representation is much more compact
than an individual data set (where the value of each claim is known), but
it also carries far less information. Now that storage space in computers
has almost become a non issue, grouped data has somewhat fallen out of
fashion, to the point that the material on grouped data present in the first
edition of Klugman et al. has been deleted in the second edition.
Still, grouped data remains in use in some fields of actuarial practice
and also of interest in teaching. For this reason, actuar provides facili-
ties to store, manipulate and summarize grouped data. A standard storage
method is needed since there are many ways to represent grouped data in
the computer: using a list or a matrix, aligning the nj s with the cj−1 s or

3
Table 1: Supported distributions from the Transformed Beta family, root
name of the R functions and argument name corresponding to each param-
eter in the parametrization of Klugman et al. (2004).
Distribution name Root (alias) Arguments
Transformed beta trbeta (pearson6) shape1 (α)
shape2 (γ)
shape3 (τ)
rate (λ = 1/θ)
scale (θ)
Burr burr shape1 (α)
shape2 (γ)
rate (λ = 1/θ)
scale (θ)
Loglogistic llogis shape (γ)
rate (λ = 1/θ)
scale (θ)
Paralogistic paralogis shape (α)
rate (λ = 1/θ)
scale (θ)
Generalized Pareto genpareto shape1 (α),
shape2 (τ),
rate (λ = 1/θ)
scale (θ)
Pareto pareto (pareto2) shape (α)
scale (θ)
Inverse Burr invburr shape1 (τ)
shape2 (γ)
rate (λ = 1/θ)
scale (θ)
Inverse Pareto invpareto shape (τ)
scale (θ)
Inverse paralogistic invparalogis shape (τ)
rate (λ = 1/θ)
scale (θ)

4
Table 2: Supported distributions from the Transformed Gamma family,
root name of the R functions and argument name corresponding to each
parameter in the parametrization of Klugman et al. (2004).
Distribution name Root (alias) Arguments
Transformed gamma trgamma shape1 (α),
shape2 (τ)
rate (λ = 1/θ),
scale (θ),
Inverse transformed invtrgamma shape1 (α),
gamma shape2 (τ)
rate (λ = 1/θ),
scale (θ),
Inverse gamma invgamma shape (α),
rate (λ = 1/θ),
scale (θ)
Inverse Weibull invweibull shape (τ),
(lgompertz) rate (λ = 1/θ),
scale (θ)
Inverse exponential invexp rate (λ = 1/θ)
scale (θ)

Table 3: Other supported distributions, root name of the R functions and

argument name corresponding to each parameter in the parametrization of
Klugman et al. (2004).
Distribution name Root (alias) Arguments
1
Loggamma lgamma shapelog (α)
ratelog (λ)
Single parameter pareto1 shape (α)
Pareto min (θ)
Generalized beta genbeta shape1 (α)
shape2 (β)
shape3 (τ)
rate (λ = 1/θ)
scale (θ)
1
See Hogg and Klugman (1984).

5
with the cj s, omitting c0 or not, etc. Moreover, with appropriate extrac-
tion, replacement and summary functions, manipulation of grouped data
becomes similar to that of individual data.
First, function grouped.data creates a grouped data object similar to —
an inheriting from — a data frame. The input of the function is a vector of
group boundaries c0 , c1 , . . . , cr and one or more vectors of group frequen-
cies n1 , . . . , nr . Note that there should be one group boundary more than
group frequencies. Furthermore, the function assumes that the intervals
are contiguous. For example, the following data

Group Frequency (Line 1) Frequency (Line 2)

(0, 25] 30 26
(25, 50] 31 33
(50, 100] 57 31
(100, 150] 42 19
(150, 250] 65 16
(250, 500] 84 11

is entered and represented in R as

> x <- grouped.data(Group = c(0, 25, 50, 100, 150,
+ 250, 500), Line.1 = c(30, 31, 57, 42, 65, 84),
+ Line.2 = c(26, 33, 31, 19, 16, 11))
> x

Group Line.1 Line.2

1 (0, 25] 30 26
2 ( 25, 50] 31 33
3 ( 50, 100] 57 31
4 (100, 150] 42 19
5 (150, 250] 65 16
6 (250, 500] 84 11

> class(x)

[1] "grouped.data" "data.frame"

Second, the package supports the most common extraction and replace-
ment methods for "grouped.data" objects using the usual [ and [<- op-
erators. In particular, the following extraction operations are supported.
i) Extraction of the vector of group boundaries (the first column):

> x[, 1]

[1] 0 25 50 100 150 250 500

ii) Extraction of the vector or matrix of group frequencies (the second and
third columns):

6
 > x[, -1]

Line.1 Line.2
1 30 26
2 31 33
3 57 31
4 42 19
5 65 16
6 84 11

iii) Extraction of a subset of the whole object (first three lines):

> x[1:3, ]

Group Line.1 Line.2

1 (0, 25] 30 26
2 ( 25, 50] 31 33
3 ( 50, 100] 57 31

Notice how extraction results in a simple vector or matrix if either of the

group boundaries or the group frequencies are dropped.
As for replacement operations, the package implements the following.

i) Replacement of one or more group frequencies:

> x[1, 2] <- 22

> x

Group Line.1 Line.2

1 (0, 25] 22 26
2 ( 25, 50] 31 33
3 ( 50, 100] 57 31
4 (100, 150] 42 19
5 (150, 250] 65 16
6 (250, 500] 84 11

> x[1, c(2, 3)] <- c(22, 19)

> x

Group Line.1 Line.2

1 (0, 25] 22 19
2 ( 25, 50] 31 33
3 ( 50, 100] 57 31
4 (100, 150] 42 19
5 (150, 250] 65 16
6 (250, 500] 84 11

ii) Replacement of the boundaries of one or more groups:

7
 > x[1, 1] <- c(0, 20)
> x

Group Line.1 Line.2

1 (0, 20] 22 19
2 ( 20, 50] 31 33
3 ( 50, 100] 57 31
4 (100, 150] 42 19
5 (150, 250] 65 16
6 (250, 500] 84 11

> x[c(3, 4), 1] <- c(55, 110, 160)

> x

Group Line.1 Line.2

1 (0, 20] 22 19
2 ( 20, 55] 31 33
3 ( 55, 110] 57 31
4 (110, 160] 42 19
5 (160, 250] 65 16
6 (250, 500] 84 11

It is not possible to replace the boundaries and the frequencies simultane-

ously.
Finally, the package defines methods of a few existing summary func-
tions for grouped data objects. Computing the mean
r 
cj−1 + cj
X 
nj (1)
j=1
2

is made simple with a method for the mean function:

> mean(x)

Line.1 Line.2
188.0 108.2

Higher empirical moments can be computed with emm; see Subsection 4.4.
The R function hist splits individual data into groups and draws an
histogram of the frequency distribution. The package introduces a method
for already grouped data. Only the first frequencies column is considered
(see Figure 1 for the resulting graph):
> hist(x[, -3])
R has a function ecdf to compute the empirical cumulative distribution
function (cdf) of an individual data set,
n
1 X
Fn (x) = I{xj ≤ x}, (2)
n i=1

8
 Histogram of x[, −3]

0.003
0.002
Density

0.001
0.000

0 100 200 300 400 500

x[, −3]

Figure 1: Histogram of a grouped data object

where I{A} = 1 if A is true and I{A} = 0 otherwise. The design and

operation of ecdf is rather unusual, but very clever and handy: the function
does not return a vector of probabilities as one would expect, but rather a
function object to compute the value of Fn (x) in any x.
The approximation of the empirical cdf for grouped data is called an
ogive (Klugman et al., 1998; Hogg and Klugman, 1984). It is obtained by
joining the known values of Fn (x) at group boundaries with straight line
segments:

0,

 x ≤ c0
 (cj − x)Fn (cj−1 ) + (x − cj−1 )Fn (cj )

F̃n (x) = , cj−1 < x ≤ cj (3)


 cj − cj−1
1, x > cr .


The package includes a function ogive that otherwise behaves exactly like
ecdf. In particular, methods for functions knots and plot allow, respec-
tively, to obtain the knots c0 , c1 , . . . , cr of the ogive and a graph (see Figure
2):
> Fnt <- ogive(x)

9
 ogive(x)

1.0
●

0.8

●
0.6
F(x)

●
0.4

●
0.2

●
0.0

0 100 200 300 400 500

Figure 2: Ogive of a grouped data object

> knots(Fnt)

[1] 0 20 55 110 160 250 500

> Fnt(knots(Fnt))

[1] 0.00000 0.07309 0.17608 0.36545 0.50498 0.72093 1.00000

> plot(Fnt)

4.3 Data sets

This is certainly not the most spectacular feature of actuar, but it remains
useful for illustrations and examples: the package includes the individual
dental claims and grouped dental claims data of Klugman et al. (2004):

> data(dental)
> dental

[1] 141 16 46 40 351 259 317 1511 107 567

10
> data(gdental)
> gdental

cj nj
1 (0, 25] 30
2 ( 25, 50] 31
3 ( 50, 100] 57
4 (100, 150] 42
5 (150, 250] 65
6 (250, 500] 84
7 (500, 1000] 45
8 (1000, 1500] 10
9 (1500, 2500] 11
10 (2500, 4000] 3

4.4 Calculation of empirical moments

The package provides two functions useful for estimation based on mo-
ments. First, function emm computes the kth empirical moment of a sample,
whether in individual or grouped data form:

> emm(dental, order = 1:3)

[1] 3.355e+02 2.931e+05 3.729e+08

> emm(gdental, order = 1:3)

[1] 3.533e+02 3.577e+05 6.586e+08

Second, in the same spirit as ecdf and ogive, function elev returns a
function to compute the empirical limited expected value — or first lim-
ited moment — of a sample for any limit. Again, there are methods for
individual and grouped data (see Figure 3 for the graphs):

> lev <- elev(dental)

> lev(knots(lev))

[1] 16.0 37.6 42.4 85.1 105.5 164.5 187.7 197.9 241.1
[10] 335.5

> plot(lev, type = "o", pch = 19)

> lev <- elev(gdental)
> lev(knots(lev))

[1] 0.00 24.01 46.00 84.16 115.77 164.85 238.26 299.77

[9] 324.90 347.39 353.34

> plot(lev, type = "o", pch = 19)

11
 elev(x = dental) elev(x = gdental)

● ● ●

300
●

300
●

250
●
●
Empirical LEV

Empirical LEV
200

200
●
●

● ●
150

100
100

●
●
●

50
●
50

●
● ●
● ●

0
0 500 1000 1500 0 1000 2000 3000 4000

x x

Figure 3: Empirical limited expected value function of an individual data

object (left) and a grouped data object (right)

4.5 Minimum distance estimation

Maximum likelihood estimation (for individual data) is well covered by
function fitdistr of package MASS. Package actuar provides function
mde, very similar in usage and inner working to fitdistr, to fit models
using three distance minimization techniques.

1. The Cramér-von Mises method (CvM) minimizes the squared difference

between the theoretical cdf and the empirical cdf or ogive at their
knots:
Xn
d(θ) = wj (F (xj ; θ) − Fn (xj ; θ))2 (4)
j=1

for individual data and

r
X
d(θ) = wj (F (cj ; θ) − F̃n (cj ; θ))2 (5)
j=1

for grouped data. Here, F (x) is the theoretical cdf of a parametric

family, Fn (x) is the empirical cdf, F̃n (x) is the ogive and w1 ≥ 0, w2 ≥
0, . . . are arbitrary weights (defaulting to 1).

2. The modified chi-square method (chi-square) applies to grouped

data only and minimizes the squared difference between the expected
and observed frequency within each group:
r
X
d(θ) = wj [n(F (cj ; θ) − F (cj−1 ; θ)) − nj ]2 , (6)
j=1

Pr
where n = j=1 nj . By default, wj = n−1
j .

12
 3. The layer average severity method (LAS) applies to grouped data only
and minimizes the squared difference between the theoretical and
empirical limited expected value within each group:
r
X
d(θ) = ˜ n (cj−1 , cj ; θ))2 ,
wj (LAS(cj−1 , cj ; θ) − LAS (7)
j=1

where LAS(x, y) = E[X ∧ y] − E[X ∧ x], LAS ˜ n (x, y) = Ẽn [X ∧ y] −

Ẽn [X ∧ x], E[X ∧ x] is the theoretical limited expected value of X at
x and Ẽn [X ∧ x] is its empirical counterpart for grouped data.

The arguments of mde are a data set, a function to compute F (x) or

E[X ∧ x], starting values for the optimization procedure and the name of
the method to use. The empirical functions are computed with ecdf, ogive
or elev.
The expressions below fit an exponential distribution to the grouped
dental data set, as per Example 2.21 of Klugman et al. (1998):

> mde(gdental, pexp, start = list(rate = 1/200),

+ measure = "CvM")

rate
0.003551

distance
0.002842

> mde(gdental, pexp, start = list(rate = 1/200),

+ measure = "chi-square")

rate
0.00364

distance
13.54

> mde(gdental, levexp, start = list(rate = 1/200),

+ measure = "LAS")

rate
0.002966

distance
694.5

It should be noted that optimization is not always that simple to achieve.

For example, consider the problem of fitting a Pareto distribution to the
same data set using the Cramér–von Mises method:

13
> mde(gdental, ppareto, start = list(shape = 3, scale = 600),
+ measure = "CvM")

Error in mde(gdental, ppareto, start = list(shape = 3, scale = 600),

measure = "CvM") :
optimization failed

Working in the log of the parameters often solves the problem since the
optimization routine can then flawlessly work with negative parameter val-
ues:

> pparetolog <- function(x, logshape, logscale) ppareto(x,

+ exp(logshape), exp(logscale))
> (p <- mde(gdental, pparetolog, start = list(logshape = log(3),
+ logscale = log(600)), measure = "CvM"))

logshape logscale
1.581 7.128

distance
0.0007905

The actual estimators of the parameters are obtained with

> exp(p$estimate)

logshape logscale
4.861 1246.485

4.6 Coverage modifications

Let X be the random variable of the actual claim amount for an insurance
policy and Y be the random variable of the amount of the claim as it ap-
pears in the insurer’s database. These two random variables will differ if
any of the following coverage modifications are present for the policy: an
ordinary or a franchise deductible, a limit, coinsurance, inflation (see Klug-
man et al., 2004, Chapter 5 for precise definitions of these terms).
Often, one will want to use data Y1 , . . . , Yn from the random variable
Y to fit a model on the unobservable random variable X. This requires
to express the probability density function (pdf) or cdf of Y in terms of
the pdf or cdf of X. Function coverage of actuar does just that: given a
pdf or cdf and any combination of the coverage modifications mentioned
above, coverage returns a function object to compute the pdf or cdf of the
modified random variable. The function can then be used in modeling like
any other d or p function.
For example, let Y represent the amount paid by an insurer for a policy
with an ordinary deductible d and a limit u − d (or maximum covered loss

14
of u). Then the definition of Y is

undefined,

 X<d
Y = X − d, d≤X≤u (8)

u − d,

X≥u

and its pdf is 

0,

 y =0
f (y + d)

X


, 0<y <u−d


− FX (d)
fY (y) = 11 −

FX (u) (9)
, y =u−d



1 − FX (d)




0, y > u − d.


Assume X has a gamma distribution. Then an R function to compute this

pdf in any y for a deductible d = 1 and a limit u = 10 is obtained with
coverage as follows:

> f <- coverage(dgamma, pgamma, deductible = 1, limit = 10)

> f(0, shape = 5, rate = 1)

[1] 0

> f(5, shape = 5, rate = 1)

[1] 0.1343

> f(9, shape = 5, rate = 1)

[1] 0.02936

> f(12, shape = 5, rate = 1)

[1] 0

See the "coverage" vignette for the detailed pdf and cdf formulas un-
der various combinations of coverage modifications.

5 Risk theory
The current version of actuar addresses only one risk theory problem with
two user visible functions: the calculation of the aggregate claim amount
distribution of an insurance portfolio using the classical collective model
of risk theory (Klugman et al., 2004; Gerber, 1979; Denuit and Charpentier,
2004; Kaas et al., 2001). That said, the package offers five different cal-
culation or approximation methods of the distribution and four different
techniques to discretize a continuous loss variable. Moreover, we feel the

15
implementation described below makes R shine as a computing and mod-
eling platform for such problems.
Let the random variable S represent the aggregate claim amount (or total
amount of claims) of a portfolio of independent risks, random variable N
represent the number of claims (or frequency) in the portfolio and random
variable Cj the amount of claim j (or severity). Then, we have the random
sum
S = C1 + · · · + CN , (10)
where we assume that C1 , C2 , . . . are mutually independent and identically
distributed random variables each independent from N. The task at hand
consists in calculating numerically the cdf of S, given by

FS (x) = Pr[S ≤ x]
∞
X
= Pr[S ≤ x|N = n]pn
n=0
X∞
= FC∗n (x)pn , (11)
n=0

where FC (x) = Pr[C ≤ x] is the common cdf of C1 , . . . , Cn , pn = Pr[N = n]

and FC∗n (x) = Pr[C1 + · · · + Cn ≤ x] is the n-fold convolution of FC (·). If
C is discrete on 0, 1, 2, . . . , one has

I{x ≥ 0}, k=0


∗k
FC (x) = FC (x), k=1 (12)
Px F ∗(k−1) (x − y)f (y), k = 2, 3, . . .


y=0 C C

5.1 Discretization of claim amount distributions

Some numerical techniques to compute the aggregate claim amount dis-
tribution (see Subsection 5.2) require a discrete arithmetic claim amount
distribution; that is, a distribution defined on 0, h, 2h, . . . for some step (or
span, or lag) h. The package provides function discretize to discretize a
continuous distribution. (The function can also be used to modify the sup-
port of an already discrete distribution, but this requires additional care.)
Let F (x) denote the cdf of the distribution to discretize on some interval
(a, b), E[X ∧ x] the limited expected value of X at x and fx the probability
mass at x in the discretized distribution. Currently, discretize supports
the following four discretization methods.

1. Upper discretization, or forward difference of F (x):

fx = F (x + h) − F (x) (13)

for x = a, a + h, . . . , b − h. The discretized cdf is always above the

true cdf.

16
 2. Lower discretization, or backward difference of F (x):
(
F (a), x=a
fx = (14)
F (x) − F (x − h), x = a + h, . . . , b.

The discretized cdf is always under the true cdf.

3. Rounding of the random variable, or the midpoint method:

(
F (a + h/2), x=a
fx = (15)
F (x + h/2) − F (x − h/2), x = a + h, . . . , b − h.

The true cdf passes exactly midway through the steps of the dis-
cretized cdf.

4. Unbiased, or local matching of the first moment method:

E[X ∧ a] − E[X ∧ a + h]

+ 1 − F (a), x=a



h



2E[X ∧ x] − E[X ∧ x − h] − E[X ∧ x + h]

fx = , a<x<b (16)

 h
 E[X ∧ b] − E[X ∧ b − h]



 − 1 + F (b), x = b.
h
The discretized and the true distributions have the same total proba-
bility and expected value on (a, b).

Figure 4 illustrates the four methods. It should be noted that although

very close in this example, the rounding and unbiased methods are not
identical.
Usage of discretize is similar to R’s plotting function curve. The
cdf to discretize and, for the unbiased method only, the limited expected
value function are passed to discretize as expressions in x. The other
arguments are the upper and lower bounds of the discretization interval,
the step h and the discretization method. For example, upper and unbiased
discretizations of a Gamma(2, 1) distribution on (0, 17) with a step of 0.5
are achieved with, respectively,

> fx <- discretize(pgamma(x, 2, 1), method = "upper",

+ from = 0, to = 17, step = 0.5)
> fx <- discretize(pgamma(x, 2, 1), method = "unbiased",
+ lev = levgamma(x, 2, 1), from = 0, to = 17,
+ step = 0.5)

Function discretize is written in a modular fashion making it simple

to add other discretization methods if needed.

17
 ● ●
●
● ●
●
● ●
0.8 ●

● ●

●
●
0.6
plnorm(x)

● ●
0.4

●
● ● upper
0.2

● lower
● rounding
● unbiased
0.0

0 1 2 3 4 5

Figure 4: Comparison of four discretization methods

5.2 Calculation of the aggregate claim amount distribution

Function aggregateDist serves as a unique front end for various methods
to compute or approximate the cdf of the aggregate claim amount random
variable S. Currently, five methods are supported.

1. Recursive calculation using the well known algorithm of Panjer (1981).

This requires the severity distribution to be discrete arithmetic on
0, 1, 2, . . . , m for some monetary unit and the frequency distribution
to be a member of either the (a, b, 0) or (a, b, 1) family of distribu-
tions (Klugman et al., 2004).

2. Exact calculation by numerical convolutions using (11) and (12). Hence,

this also requires a discrete severity distribution. However, there is no
restriction on the shape of the frequency distribution. The package
merely implements the sum (11), the convolutions being computed
with R’s function convolve. This approach is practical for small prob-
lems only, even on today’s fast computers.

18
 3. Normal approximation of the cdf, that is
S − µS
 
FS (x) ≈ Φ , (17)
σS

where µS = E[S] and σS2 = Var[S]. For most realistic models, this
approximation is rather crude in the tails of the distribution.

4. Normal Power II approximation:

s !
3 9 6 x − µS
F (x) ≈ Φ − + +1+ , (18)
γS γS2 γS σS
3/2
where γS = E[(S − µS )3 ]/σS . The approximation is valid for x > µS
only and performs reasonably well when γS < 1. See Daykin et al.
(1994) for details.

5. Simulation of a random sample from S and approximation of FS (x)

by the empirical cdf (2). The simulation itself is done with function
simpf (see Subsection 6.2). This function admits very general hierar-
chical models for both the frequency and the severity components.
Here also, adding other methods to aggregateDist is simple due to its
modular conception.
The arguments of aggregateDist differ depending on the calculation
method; see the help page for details. One interesting argument to note
is x.scale to specify the monetary unit of the severity distribution. This
way, one does not have to mentally do the conversion between the support
of 0, 1, 2, . . . assumed by the recursive and convolution methods and the
true support of S.
Function aggregateDist returns a function object to compute the value
of FS (x) in any x. Moreover, the package defines a few summary functions
to extract information from this object.
For illustration purposes, consider the following model: the distribution
of S is a compound Poisson with parameter λ = 10 and severity distribu-
tion Gamma(2, 1). To obtain an approximation of the cdf of S we first
discretize the gamma distribution on (0, 22) with a step of 0.5 and then
use the recursive method:
> fx <- discretize(pgamma(x, 2, 1), from = 0, to = 22,
+ step = 0.5, method = "unbiased", lev = levgamma(x,
+ 2, 1))
> Fs <- aggregateDist("recursive", model.freq = "poisson",
+ model.sev = fx, lambda = 10, x.scale = 0.5)
> summary(Fs)

Aggregate Claim Amount Empirical CDF:

Min. 1st Qu. Median Mean 3rd Qu. Max.
0.0 14.5 19.5 20.0 25.0 71.0

19
Hence, object Fs contains an empirical cdf with support

> knots(Fs)

[1] 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
[12] 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5
[23] 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 15.5 16.0
[34] 16.5 17.0 17.5 18.0 18.5 19.0 19.5 20.0 20.5 21.0 21.5
[45] 22.0 22.5 23.0 23.5 24.0 24.5 25.0 25.5 26.0 26.5 27.0
[56] 27.5 28.0 28.5 29.0 29.5 30.0 30.5 31.0 31.5 32.0 32.5
[67] 33.0 33.5 34.0 34.5 35.0 35.5 36.0 36.5 37.0 37.5 38.0
[78] 38.5 39.0 39.5 40.0 40.5 41.0 41.5 42.0 42.5 43.0 43.5
[89] 44.0 44.5 45.0 45.5 46.0 46.5 47.0 47.5 48.0 48.5 49.0
[100] 49.5 50.0 50.5 51.0 51.5 52.0 52.5 53.0 53.5 54.0 54.5
[111] 55.0 55.5 56.0 56.5 57.0 57.5 58.0 58.5 59.0 59.5 60.0
[122] 60.5 61.0 61.5 62.0 62.5 63.0 63.5 64.0 64.5 65.0 65.5
[133] 66.0 66.5 67.0 67.5 68.0 68.5 69.0 69.5 70.0 70.5 71.0

A nice graph of this function is obtained with plot (see Figure 5):

> plot(Fs, do.points = FALSE, verticals = TRUE, xlim = c(0,

+ 60))

Finally, one can easily compute the mean and obtain the quantiles of the
approximate distribution as follows:

> mean(Fs)

[1] 20

> quantile(Fs)

25% 50% 75% 90% 95% 97.5% 99% 99.5%

14.5 19.5 25.0 30.5 34.0 37.0 41.0 43.5

> quantile(Fs, 0.999)

99.9%
49.5

To conclude on the subject, Figure 6 shows the cdf of S using five com-
binations of discretization and calculation method supported by actuar.
Other combinations are possible.

6 Credibility theory
The credibility theory facilities of actuar consist of one data set and three
main functions:

20
 Aggregate Claim Amount Distribution
Recursive method approximation

1.0
0.8
0.6
FS(x)

0.4
0.2
0.0

0 10 20 30 40 50 60

Figure 5: Graphic of the empirical cdf of S obtained with the recursive

method

1. matrix hachemeister containing the famous data set of Hachemeis-

ter (1975);

2. function simpf to simulate data from compound hierarchical models;

3. function cm to fit linear hierarchical credibility models;

4. function bstraub, a faster and simpler version of cm to fit Bühlmann

and Bühlmann–Straub models.

6.1 Hachemeister data set

The data set of Hachemeister (1975) consists of average claim amounts for
private passenger bodily injury insurance for five U.S. states over 12 quar-
ters between July 1970 and June 1973 and the corresponding number of
claims. The data set is included in the package in the form of a matrix with
5 rows and 25 columns. The first column contains a state index, columns
2–13 contain the claim averages and columns 14–25 contain the claim num-
bers:

21
 Aggregate Claim Amount Distribution

1.0
0.8
0.6
FS(x)

0.4

recursive + unbiased
recursive + upper
recursive + lower
0.2

simulation
normal approximation
0.0

0 10 20 30 40 50 60

Figure 6: Comparison between the empirical or approximate cdf of S ob-

tained with five different methods

> data(hachemeister)
> hachemeister

state ratio.1 ratio.2 ratio.3 ratio.4 ratio.5 ratio.6

[1,] 1 1738 1642 1794 2051 2079 2234
[2,] 2 1364 1408 1597 1444 1342 1675
[3,] 3 1759 1685 1479 1763 1674 2103
[4,] 4 1223 1146 1010 1257 1426 1532
[5,] 5 1456 1499 1609 1741 1482 1572
ratio.7 ratio.8 ratio.9 ratio.10 ratio.11 ratio.12
[1,] 2032 2035 2115 2262 2267 2517
[2,] 1470 1448 1464 1831 1612 1471
[3,] 1502 1622 1828 2155 2233 2059
[4,] 1953 1123 1343 1243 1762 1306
[5,] 1606 1735 1607 1573 1613 1690
weight.1 weight.2 weight.3 weight.4 weight.5 weight.6
[1,] 7861 9251 8706 8575 7917 8263
[2,] 1622 1742 1523 1515 1622 1602
[3,] 1147 1357 1329 1204 998 1077

22
[4,] 407 396 348 341 315 328
[5,] 2902 3172 3046 3068 2693 2910
weight.7 weight.8 weight.9 weight.10 weight.11 weight.12
[1,] 9456 8003 7365 7832 7849 9077
[2,] 1964 1515 1527 1748 1654 1861
[3,] 1277 1218 896 1003 1108 1121
[4,] 352 331 287 384 321 342
[5,] 3275 2697 2663 3017 3242 3425

6.2 Portfolio simulation

Function simpf simulates portfolios of data following compound models
of the form (10) where both the frequency and the severity components
can have a hierarchical structure. The main characteristic of hierarchical
models is to have the probability law at some level in the classification
structure be conditional on the outcome in previous levels. For example,
consider the following three-level compound hierarchical model:

Sijt = Cijt1 + · · · + CijtNijt , (19)

for i = 1, . . . , I, j = 1, . . . , Ji , t = 1, . . . , nij and with

Nijt |Λij , Φi ∼ Poisson(wijt Λij ) Cijtu |Θij , Ψi ∼ Lognormal(Θij , 1)

Λij |Φi ∼ Gamma(Φi , 1) Θij |Ψi ∼ N(Ψi , 1) (20)
Φi ∼ Exponential(2) Ψi ∼ N(2, 0.1).

The random variables Φi , Λij , Ψi and Θij are generally seen as risk parame-
ters in the actuarial literature. The wijt s are known weights.
Function simpf is presented in the credibility theory section because it
was originally written in this context, but it has much wider applications.
For instance, as mentioned in Subsection 5.2, it is used by aggregateDist
for the approximation of the cdf of S by simulation.
Goulet and Pouliot (2007) describe in detail the model specification
method used in simpf. For the sake of completeness, we briefly outline
this method here.
A hierarchical model is completely specified by the number of nodes at
each level (I, J1 , . . . , JI and n11 , . . . , nIJ , above) and by the probability laws
at each level. The number of nodes is passed to simpf by means of a named
list where each element is a vector of the number of nodes at a given level.
Vectors are recycled when the number of nodes is the same throughout a
level. Probability models are expressed in a semi-symbolic fashion using an
object of mode "expression". Each element of the object must be named
— with names matching those of the number of nodes list — and should
be a complete call to an existing random number generation function, with
the number of variates omitted. Now, hierarchical models are achieved by
replacing one or more parameters of a distribution at a given level by any

23
combination of the names of the levels above. If no mixing is to take place
at a level, the model for this level can be NULL.
Function simpf also supports usage of weights, or volumes, in models.
These usually modify the frequency parameters to take into account the
“size” of an insurance contract. The weights will be used in simulation
wherever the name weights appears in a model.
Hence, function simpf has four main arguments: 1) nodes for the num-
ber of nodes list; 2) model.freq for the frequency model; 3) model.sev for
the severity model; 4) weights for the vector of weights in lexicographic
order, that is all weights of contract 1, then all weights of contract 2, and
so on.
For example, assuming that I = 2, J1 = 4, J2 = 3, n11 = · · · = n14 = 4
and n21 = n22 = n23 = 5 in model (20) above, and that weights are simply
simulated from a uniform distribution on (0.5, 2.5), then simulation of a
data set with simpf is achieved with:

> wijt <- runif(31, 0.5, 2.5)

> nodes <- list(class = 2, contract = c(4, 3), year = c(4,
+ 4, 4, 4, 5, 5, 5))
> mf <- expression(class = rexp(2), contract = rgamma(class,
+ 1), year = rpois(weights * contract))
> ms <- expression(class = rnorm(2, sqrt(0.1)), contract = rnorm(class,
+ 1), year = rlnorm(contract, 1))
> pf <- simpf(nodes = nodes, model.freq = mf, model.sev = ms,
+ weights = wijt)

The function returns the variates in a two-dimension list containing all

the individual claim amounts for each contract. Such an object can be seen
as a three-dimension array with a third dimension of potentially varying
length. The function also returns a matrix of integers giving the classifi-
cation indexes of each contract in the portfolio (subscripts i and j in the
notation above). Displaying the complete content of the object returned by
simpf can be impractical. For this reason, only the simulation model and
the number of claims in each node is printed:

> pf

Portfolio of claim amounts

Frequency model
class ~ rexp(2)
contract ~ rgamma(class, 1)
year ~ rpois(weights * contract)
Severity model
class ~ rnorm(2, sqrt(0.1))
contract ~ rnorm(class, 1)
year ~ rlnorm(contract, 1)

24
 Number of claims per node:

class contract year.1 year.2 year.3 year.4 year.5

[1,] 1 1 2 2 1 0 NA
[2,] 1 2 0 0 0 0 NA
[3,] 1 3 0 3 0 2 NA
[4,] 1 4 0 1 1 1 NA
[5,] 2 1 0 1 1 1 2
[6,] 2 2 0 0 0 0 0
[7,] 2 3 3 4 2 2 0

The package contains three functions to easily compute the matrices

of variates of the aggregate random variable S and variates of the fre-
quency random variable N, and to rearrange the list returned by simpf
into a matrix of the variates of the severity random variables Cj . These are,
respectively, aggregate, frequency and severity. In addition, function
weights extracts the weights matrix from a simulated data set.
By default, aggregate returns the values of Sijt in a regular matrix
(subscripts i and j in the rows, subscript t in the columns). The method
has a by argument to get statistics for other groupings and a FUN argument
to get statistics other than the sum:

> aggregate(pf)

class contract year.1 year.2 year.3 year.4 year.5

[1,] 1 1 31.37 7.521 11.383 0.000 NA
[2,] 1 2 0.00 0.000 0.000 0.000 NA
[3,] 1 3 0.00 72.706 0.000 23.981 NA
[4,] 1 4 0.00 98.130 50.622 55.705 NA
[5,] 2 1 0.00 11.793 2.253 2.397 10.48
[6,] 2 2 0.00 0.000 0.000 0.000 0.00
[7,] 2 3 44.81 88.737 57.593 14.589 0.00

> aggregate(pf, by = c("class", "year"), FUN = mean)

class year.1 year.2 year.3 year.4 year.5

[1,] 1 15.69 29.73 31.00 26.562 NA
[2,] 2 14.94 20.11 19.95 5.662 5.238

Function frequency returns the values of Nijt . It is a wrapper for

aggregate with the default sum statistic replaced by length. Hence, ar-
guments by and FUN remain available:

> frequency(pf)

class contract year.1 year.2 year.3 year.4 year.5

[1,] 1 1 2 2 1 0 NA

25
[2,] 1 2 0 0 0 0 NA
[3,] 1 3 0 3 0 2 NA
[4,] 1 4 0 1 1 1 NA
[5,] 2 1 0 1 1 1 2
[6,] 2 2 0 0 0 0 0
[7,] 2 3 3 4 2 2 0

> frequency(pf, by = "class")

class freq
[1,] 1 17
[2,] 2 16

Finally, function severity returns the individual variates Cijtu in a ma-

trix similar to those above, but with a number of columns equal to the
maximum number of observations per contract,
nij
X
max Nijt .
i,j
t=1

Thus, the original period of observation (subscript t) and the identifier of

the severity within the period (subscript u) are lost and each variate now
constitute a “period” of observation. For this reason, the function provides
an argument splitcol in case one would like to extract separately the
individual severities of one or more periods:
> severity(pf)

$first
class contract claim.1 claim.2 claim.3 claim.4 claim.5
[1,] 1 1 7.974 23.401 3.153 4.368 11.383
[2,] 1 2 NA NA NA NA NA
[3,] 1 3 3.817 41.979 26.910 4.903 19.078
[4,] 1 4 98.130 50.622 55.705 NA NA
[5,] 2 1 11.793 2.253 2.397 9.472 1.004
[6,] 2 2 NA NA NA NA NA
[7,] 2 3 14.322 11.522 18.966 33.108 15.532
claim.6 claim.7 claim.8 claim.9 claim.10 claim.11
[1,] NA NA NA NA NA NA
[2,] NA NA NA NA NA NA
[3,] NA NA NA NA NA NA
[4,] NA NA NA NA NA NA
[5,] NA NA NA NA NA NA
[6,] NA NA NA NA NA NA
[7,] 14.99 25.11 40.15 17.44 4.426 10.16

$last
NULL

26
> severity(pf, splitcol = 1)

$first
class contract claim.1 claim.2 claim.3 claim.4 claim.5
[1,] 1 1 3.153 4.368 11.383 NA NA
[2,] 1 2 NA NA NA NA NA
[3,] 1 3 3.817 41.979 26.910 4.903 19.078
[4,] 1 4 98.130 50.622 55.705 NA NA
[5,] 2 1 11.793 2.253 2.397 9.472 1.004
[6,] 2 2 NA NA NA NA NA
[7,] 2 3 33.108 15.532 14.990 25.107 40.150
claim.6 claim.7 claim.8
[1,] NA NA NA
[2,] NA NA NA
[3,] NA NA NA
[4,] NA NA NA
[5,] NA NA NA
[6,] NA NA NA
[7,] 17.44 4.426 10.16

$last
class contract claim.1 claim.2 claim.3
[1,] 1 1 7.974 23.40 NA
[2,] 1 2 NA NA NA
[3,] 1 3 NA NA NA
[4,] 1 4 NA NA NA
[5,] 2 1 NA NA NA
[6,] 2 2 NA NA NA
[7,] 2 3 14.322 11.52 18.97

Finally, the weights matrix of the portfolio is

> weights(pf)

class contract year.1 year.2 year.3 year.4 year.5

[1,] 1 1 0.8361 2.115 1.2699 1.1555 NA
[2,] 1 2 1.7042 1.709 0.7493 1.0892 NA
[3,] 1 3 1.6552 1.762 1.5240 1.5100 NA
[4,] 1 4 1.5681 1.614 2.2358 2.1594 NA
[5,] 2 1 0.7229 1.907 2.2950 1.0595 0.9564
[6,] 2 2 0.5307 0.758 0.6868 0.9738 2.0823
[7,] 2 3 1.6995 2.320 1.6208 2.0114 1.2583

Function simpf was used to simulate the data in Forgues et al. (2006).

27
6.3 Fitting of hierarchical credibility models
The linear model fitting function of base R is named lm. Since credibility
models are very close in many respects to linear models, and since the
credibility model fitting function of actuar borrows much of its interface
from lm, we named the credibility function cm.
We hope that, in the long term, cm can act as a unified interface for most
credibility models. Currently, it supports the models of Bühlmann (1969)
and Bühlmann and Straub (1970), as well as the hierarchical model of Jewell
(1975). The last model includes the first two as special cases. (According
to our current counting scheme, a Bühlmann-Straub model is a two-level
hierarchical model.)
There are some variations in the formulas of the hierarchical model
in the literature. We estimate the structure parameters as indicated in
Goovaerts and Hoogstad (1987) but compute the credibility premiums as
given in Bühlmann and Jewell (1987) or Bühlmann and Gisler (2005); Goulet
(1998) has all the appropriate formulas for our implementation. For in-
stance, for a three-level hierarchical model like (19)-(20), the best linear
prediction of the ratio Xij,nij +1 = Sij,nij +1 /wij,nij +1 is

π̂ij = zij Xijw + (1 − zij )π̂i

(21)
π̂i = zi Xizw + (1 − zi )m
with
nij
wij Σ X wijt
zij = , Xijw = Xijt
wij Σ + s 2 /a t=1
wij Σ
Ji
ziΣ X zij
zi = , Xizw = Xijw .
ziΣ + a/b z
j=1 iΣ

The (pseudo-)estimators of the structure parameters s 2 , a, b and m are,

respectively,
Ji n
I X ij
2 1 X X
ŝ = PI PJi wijt (Xijt − Xijw )2
i=1 j=1 (nij − 1) i=1 j=1 t=1
Ji
I X
1 X
â = PI zij (Xijw − Xizw )2
i=1 (Ji − 1) i=1 j=1
I
1 X
b̂ = zi (Xizw − Xzzw )2
I − 1 i=1

and
I
X zi
m̂ = Xzzw = Xizw .
z
i=1 Σ

28
 Function cm takes in argument a formula describing the hierarchical
interactions in the data set, a data set containing the variables referenced in
the formula and the names of the columns where the ratios and the weights
are to be found in the data set. The latter should be a matrix or a data frame
with one column of indexes (numeric or character) for each hierarchical
interaction, at least two nodes in each level and more than one period of
experience for at least one contract. Missing values are represented by NAs.
There can be contracts with no experience (complete lines of NAs).
The function returns a fitted model object containing the estimators of
the structure parameters. To compute the credibility premiums, one calls
function predict with the said object in argument. One can also obtain a
nicely formated view of the most important results with a call to summary.
These two functions can report for the whole portfolio or for a subset of
levels only by means of an argument "levels".
In order to give an easily reproducible example, we group states 1 and
3 of the Hachemeister data set into one class and states 2, 4 and 5 into
another. This also shows that data does not have to be sorted by level. The
fitted model is:

> X <- cbind(class = c(1, 2, 1, 2, 2), hachemeister)

> fit <- cm(~class + class:state, data = X, ratios = ratio.1:ratio.12,
+ weights = weight.1:weight.12)
> fit

Call: cm(formula = ~class + class:state, data = X, ratios = ratio.1:ratio.12,

Structure Parameters Estimators

Collective premium: 1746

Between class variance: 88981
Within class/Between state variance: 10952
Within state variance: 139120026

The key results of this fit are obtained with summary:

> summary(fit)

Call: cm(formula = ~class + class:state, data = X, ratios = ratio.1:ratio.12,

Structure Parameters Estimators

Collective premium: 1746

Between class variance: 88981
Within class/Between state variance: 10952
Within state variance: 139120026

Detailed premiums

29
 Level: class
class Ind. premium Weight Cred. factor Cred. premium
1 1967 1.407 0.9196 1949
2 1528 1.596 0.9284 1543

Level: state
class state Ind. premium Weight Cred. factor
1 1 2061 100155 0.8874
2 2 1511 19895 0.6103
1 3 1806 13735 0.5195
2 4 1353 4152 0.2463
2 5 1600 36110 0.7398
Cred. premium
2048
1524
1875
1497
1585

Function predict returns only the credibility premiums per level:

> predict(fit)

$class
[1] 1949 1543

$state
[1] 2048 1524 1875 1497 1585

Finally, one can obtain the results above for the class level only as follows:
> summary(fit, levels = "class")

Call: cm(formula = ~class + class:state, data = X, ratios = ratio.1:ratio.12,

Structure Parameters Estimators

Collective premium: 1746

Between class variance: 88981
Within class variance: 10952

Detailed premiums

Level: class
class Ind. premium Weight Cred. factor Cred. premium
1 1967 1.407 0.9196 1949
2 1528 1.596 0.9284 1543

30
> predict(fit, levels = "class")

$class
[1] 1949 1543

The results above differ from those of Goovaerts and Hoogstad (1987)
for the same example because the formulas for the credibility premiums
are different.

6.4 Bühlmann and Bühlmann–Straub models

Function bstraub is a faster and simpler alternative to cm for Bühlmann
and Bühlmann–Straub models. There is no formula argument in this func-
tion and data is passed by means of a matrix of ratios and a matrix of
weights. Furthermore, this function has the option to compute the unbi-
ased estimator of the between variance
 
I
wΣΣ X
2 2
â = 2 PI  wiΣ (Xiw − Xww ) − (I − 1)ŝ  .
wΣΣ − i=1 wi2Σ i=1

Otherwise, usage of summary and predict for models fitted with bstraub
is identical to models fitted with cm.

7 Conclusion
The paper presented the facilities of the R package actuar version 0.9-3 in
the fields of loss distribution modeling, risk theory and credibility theory.
We feel this version of the package covers most of the basics needs in these
areas. In the future we plan to improve the functions currently available —
especially speed wise — but also to start adding more advanced features.
For example, future versions of the package should include support for
dependence models in risk theory and regression credibility models.
Obviously, the package left many other fields of Actuarial Science un-
touched so far. For this situation to change, we hope that experts in their
field will join their efforts to ours and contribute code to the actuar project.
This project intends to continue to grow and improve by and for the com-
munity of developers and users.
Finally, if you use R or actuar for actuarial analysis, please cite the soft-
ware in publications. Use

> citation()

or

> citation("actuar")

for information on how to cite the software.

31
Acknowledgments
The package would not be at this stage of development without the stim-
ulating contribution of the following students: Sébastien Auclair, Mathieu
Pigeon, Louis-Philippe Pouliot and Tommy Ouellet.
This research benefited from financial support from the Natural Sci-
ences and Engineering Research Council of Canada and from the Chaire
d’actuariat (Actuarial Science Chair) of Université Laval.

References
Bühlmann, H., 1969. Experience rating and credibility. ASTIN Bulletin 5,
157–165.

Bühlmann, H., Gisler, A., 2005. A course in credibility theory and its appli-
cations. Springer.

Bühlmann, H., Jewell, W. S., 1987. Hierarchical credibility revisited. Bulletin

of the Swiss Association of Actuaries 87, 35–54.

Bühlmann, H., Straub, E., 1970. Glaubgwürdigkeit für Schadensätze. Bul-

letin of the Swiss Association of Actuaries 70, 111–133.

Daykin, C., Pentikäinen, T., Pesonen, M., 1994. Practical Risk Theory for
Actuaries. Chapman & Hall, London.

Denuit, M., Charpentier, A., 2004. Mathématiques de l’assurance non-vie.

Vol. 1, Principes fondamentaux de théorie du risque. Economica, Paris.

Forgues, A., Goulet, V., Lu, J., 2006. Credibility for severity revisited. North
American Actuarial Journal 10 (1), 49–62.

Gerber, H. U., 1979. An Introduction to Mathematical Risk Theory. Huebner

Foundation, Philadelphia.

Goovaerts, M. J., Hoogstad, W. J., 1987. Credibility theory. No. 4 in Surveys

of actuarial studies. Nationale-Nederlanden N.V., Netherlands.

Goulet, V., 1998. Principles and application of credibility theory. Journal of

Actuarial Practice 6, 5–62.

Goulet, V., 2007. actuar: An R Package for Actuarial Science, version 0.9-3.
École d’actuariat, Université Laval.
URL http://www.actuar-project.org

Goulet, V., Pouliot, L.-P., 2007. Simulation of hierarchical models in R. Jour-

nal of Statistical Software Submitted for publication.

32
Hachemeister, C. A., 1975. Credibility for regression models with appli-
cation to trend. In: Credibility, theory and applications. Proceedings of
the Berkeley actuarial research conference on credibility. Academic Press,
New York.

Hogg, R. V., Klugman, S. A., 1984. Loss Distributions. Wiley, New York.

Jewell, W. S., 1975. The use of collateral data in credibility theory: a hierar-
chical model. Giornale dell’Istituto Italiano degli Attuari 38, 1–16.

Kaas, R., Goovaerts, M., Dhaene, J., Denuit, M., 2001. Modern actuarial risk
theory. Kluwer Academic Publishers, Dordrecht.

Klugman, S. A., Panjer, H. H., Willmot, G., 1998. Loss Models: From Data to
Decisions. Wiley, New York.

Klugman, S. A., Panjer, H. H., Willmot, G., 2004. Loss Models: From Data to
Decisions, 2nd Edition. Wiley, New York.

Panjer, H. H., 1981. Recursive evaluation of a family of compound distribu-

tions. Astin Bulletin 12, 22–26.

R Development Core Team, 2007. R: A language and environment for statis-

tical computing. R Foundation for Statistical Computing, Vienna, Austria.
URL http://www.r-project.org

Vincent Goulet
École d’actuariat
Pavillon Alexandre-Vachon, bureau 1620
Université Laval
Québec (QC) G1K 7P4
Canada
E-mail: [email protected]
URL: http://vgoulet.act.ulaval.ca/

33