FiguRE 1: Sample Company Profile (amounts in millions)
net Earned
Premium
Expected
net losses
CV of net
losses
Booked
Reserves
170 100 0.10 100
130 80 0.15 150
70 50 0.15 150
Policy type
Auto
General Liability
Workers’
Compensation
Property
CV of
Reserves
0.15
0.10
0.18
240
160
0.44
50
0.14
FiguRE 2: Capital Allocation Based on Standard Deviation
(in millions)
Policy type
Auto
General Liability
Workers’ Compensation
Property
Standard
Deviation
18
19
27
70
Allocation
13.4%
14.2%
20.1%
52.2%
Allocated Capital
47
50
71
183
FiguRE 3: Capital Allocated According to Covariance (in millions)
Covariance
(trillions)
485
538
693
4,859
Policy type
Auto
General Liability
Workers’ Compensation
Property
Allocation
7.4%
8.2%
10.5%
73.9%
Allocated Capital
26
29
37
259
deviation are ease of calculation and familiarity to most senior management.
Drawbacks include its consideration of
both favorable and adverse deviations
from expectations and its lack of recognition of correlations among risks. Further,
the allocations derived from standard
deviation are not coherent: The sum of
the allocations for subsets of a segment
will equal the allocation for the segment
as a whole only by chance.
To illustrate this last drawback, we
can compare the surplus allocated when
only casualty and property are considered, rather than allocation among the
three casualty segments shown in Figure
2. The casualty standard deviation is $43
million, i.e., $21 million less than the sum
of the standard deviations for the three
segments. As such, if capital were allocated using only property and casualty,
$133 million would be allocated to casualty, with $217 million going to property.
The balance in allocated capital between
property and casualty changes significantly in this illustration because of the
assumption that the correlations among
the three casualty segments are less than
perfect (i.e., less than 100 percent).
Investment and credit risk are ignored
in this illustration.
Now, let’s take a look at how the different proportional capital allocation
approaches affect this company.
Standard Deviation
Standard deviation has long been a
common allocation basis in actuarial literature. There have been many papers
published discussing the use of standard
deviation as a basis for allocating capital
for pricing either explicitly or implicitly—e.g., in calculating increased limits
factors. With this approach, the standard
deviation of each segment is calculated
independently. Capital is allocated to
each segment in the same proportion as
its standard deviation bears to the sum
of the standard deviations.
The standard deviations by line for
the combinations of reserves and current accident-year results are shown in
Figure 2. Assuming the totality of capital is to be allocated among lines, the
percentage and dollar amounts of capital allocated are also shown. (There are
two aspects to this choice, both of which
are outside the scope of this article. The
first is whether to allocate the totality
of an insurer’s capital or just a portion
of it if the company perceives that it is
carrying excess capital of future opportunities. The second is whether
to allocate all capital to underwriting
and therefore include all of the related
investment risks with each line of business or to allocate capital separately to
underwriting, investment and any other significant sources of risk within the
organization.)
The primary advantages of standard
Covariance
Covariance is the expected value of the
product of the deviations of two variables from their means. In each iteration,
the difference between a segment’s value
and its mean is multiplied by the difference between the total value and the total
mean. The average of these products is
the covariance. Capital is then allocated
in proportion to the covariances. Figure
3 shows the covariances and allocated
capital for the sample company.
The advantages to covariance as an
allocation metric are:
It is fairly well understood, even out- ■ ■
side actuarial circles;
It is easy to calculate; ■ ■
It is coherent; ■ ■
It reflects the correlations among ■ ■
segments and with the total.