Estimating Unpaid Claims Using Basic Techniques

Chapter 9 - Bornhuetter-Ferguson Technique

CHAPTER 9 – BORNHUETTER-FERGUSON TECHNIQUE

Actuaries rely on the Bornhuetter-Ferguson technique almost as often as they rely on the
development method. The Bornhuetter-Ferguson technique is essentially a blend of the
development and expected claims techniques. In the development technique, we multiply actual
claims by a cumulative claim development factor. This technique can lead to erratic, unreliable
projections when the cumulative development factor is large because a relatively small swing in
reported claims or the reporting of an unusually large claim could result in a very large swing in
projected ultimate claims. In the expected claims technique, the unpaid claim estimate is equal to
the difference between a predetermined estimate of expected claims and the actual payments.
This has the advantage of stability, but it completely ignores actual results as reported. The
Bornhuetter-Ferguson technique combines the two techniques by splitting ultimate claims into
two components: actual reported (or paid) claims and expected unreported (or unpaid) claims. As
experience matures, more weight is given to the actual claims and the expected claims become
gradually less important.

In the 1993 paper “Loss Development Using Credibility,”53 Eric Brosius described the
Bornhuetter-Ferguson method as a credibility weighting between the development method and
the expected claims method. In the development method, full credibility (i.e., Z = 1) is given to
actual claims experience; and in the expected claims method, no credibility (i.e., Z = 0) is given
to actual claims. In the Bornhuetter-Ferguson, credibility is equal to the percentage of claims
developed at a particular stage of maturity, which is a function of the cumulative claim
development factor (i.e., Z = 1.00 / cumulative development factor). Therefore, more weight is
given to the expected claims method in less mature years, and more weight is given to the
development method in more mature years of the experience period.

Key Assumptions

The key assumption of the Bornhuetter-Ferguson method is that unreported (or unpaid) claims
will develop based on expected claims. In other words, the claims reported to date contain no
informational value as to the amount of claims yet-to-be reported. This is different from the
development method where the primary assumption is that unreported (or unpaid) claims will
develop based on reported (or paid) claims to date.

The reporting and payment patterns used in the Bornhuetter-Ferguson methods are the same as
those selected in the development method. However, the application of the development factors
differs between the two methods. It is also important to note that the expected claims used in the
Bornhuetter-Ferguson method using reported claims are the same as those used in the
Bornhuetter-Ferguson method using paid claims.

Common Uses of the Bornhuetter-Ferguson Technique

The Bornhuetter-Ferguson technique is most frequently applied to reported and paid claims, yet it
can also be used with the number of claims and with ALAE. Actuaries use this technique with all
lines of insurance including short-tail lines and long-tail lines. Similar to the development

53
CAS Study Note, 1993

152
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

method, the Bornhuetter-Ferguson method is used with data organized in many different time
intervals including:

Accident year
Policy year
Underwriting year
Report year
Fiscal year

Actuaries also apply this technique to data organized by month, quarter, or half-year.

Mechanics of the Bornhuetter-Ferguson Technique

As indicated previously, the Bornhuetter-Ferguson technique is a blend of two other methods: the
development method and the expected claims method. The following two formulae represent the
reported and paid Bornhuetter-Ferguson methods, respectively:

Ultimate Claims = Actual Reported Claims + Expected Unreported Claims

Ultimate Claims = Actual Reported Claims + (Expected Claims) x (% Unreported)

Ultimate Claims = Actual Paid Claims + Expected Unpaid Claims

Ultimate Claims = Actual Paid Claims + (Expected Claims) x (% Unpaid)

Since actual reported and paid claims are both known quantities, the challenge of the
Bornhuetter-Ferguson method is to calculate the expected unreported and expected unpaid claims.
In order to complete the Bornhuetter-Ferguson method, the actuary must select claim
development patterns and develop an expected claims estimate.

In our step-by-step example of the Bornhuetter-Ferguson method, we use the cumulative claim
development patterns presented in Chapter 7 and the expected claims developed in Chapter 8.54 In
Exhibit I, Sheet 1, we present both the reported and paid Bornhuetter-Ferguson projections for
U.S. Industry Auto.

The second column of Exhibit I, Sheet 1, contains the expected claims developed in Chapter 8 for
U.S. Industry Auto. Columns (3) and (4) are the selected cumulative claim development factors
described in Chapter 7. We convert the cumulative claim development patterns to percentage
unreported and percentage unpaid in Columns (5) and (6), respectively. The percentage reported
is equal to the inverse of the cumulative reported claim development factor. Thus, the percentage
unreported is equal to 1.00 minus the inverse of the cumulative reported claim development
factor. Similarly, the percentage paid is equal to the inverse of the cumulative paid claim

54
Recall that expected claims are developed in Chapter 8 based on earned premiums and selected claim
ratios. We discussed the importance of adjusting premiums to an on-level basis when selecting expected
claim ratios. The purpose of adjusting premiums to an on-level basis is to develop a proxy for the
underlying exposures in each year of the experience period. An alternative to the use of premiums and
claim ratios for developing expected claims is exposures and pure premiums (also referred to as loss rates
or loss costs). Many actuaries who work with self-insurers rely on such an approach. Due to enhancements
in many insurers’ data systems, historical exposures may become more readily available to actuaries and
can thus be directly incorporated into the development of expected claims for the Bornhuetter-Ferguson
technique.

153
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

development factor; and the percentage unpaid is equal to 1.00 minus the inverse of the
cumulative paid claim development factor.

Once again, we summarize the selected claim development factors for reported and paid claims as
well as the associated reporting and payment patterns in Table 1 below.

Table 1 – U.S. Industry Auto

Selected Reporting and Payment Patterns
Reported Claims Paid Claims
Age CDF to % % CDF to % %
(Month) Ultimate Reported Unreported Ultimate Paid Unpaid
12 1.292 77.4% 22.6% 2.390 41.8% 58.2%
24 1.110 90.1% 9.9% 1.404 71.2% 28.8%
36 1.051 95.1% 4.9% 1.184 84.5% 15.5%
48 1.023 97.8% 2.2% 1.085 92.2% 7.8%
60 1.011 98.9% 1.1% 1.040 96.2% 3.8%
72 1.006 99.4% 0.6% 1.020 98.0% 2.0%
84 1.003 99.7% 0.3% 1.011 98.9% 1.1%
96 1.001 99.9% 0.1% 1.006 99.4% 0.6%
108 1.000 100.0% 0.0% 1.004 99.6% 0.4%
120 1.000 100.0% 0.0% 1.002 99.8% 0.2%

The primary assumption of the reported Bornhuetter-Ferguson method is that unreported claims
will emerge in accordance with expected claims. Thus, the next step of this method is to calculate
the expected unreported claims. In Column (7), we calculate expected unreported claims by
accident year. Expected unreported claims are equal to expected claims in Column (2) multiplied
by the percentage unreported in Column (5) for each year. Similarly, expected unpaid claims in
Column (8) are equal to expected claims from Column (2) multiplied by the percentage unpaid in
Column (6).

Returning to our original formulae for the Bornhuetter-Ferguson method,

Ultimate Claims = Actual Reported Claims + Expected Unreported Claims

Ultimate Claims = Actual Paid Claims + Expected Unpaid Claims

We can now calculate the projected ultimate claims. Using the reported Bornhuetter-Ferguson
method, projected ultimate claims in Column (11) are equal to the actual reported claims in
Column (9) plus the expected unreported claims in Column (7). The projected ultimate claims
based on the paid Bornhuetter-Ferguson method are shown in Column (12); they are equal to
actual paid claims in Column (10) plus expected unpaid claims in Column (8).

Unpaid Claim Estimate Based on Bornhuetter-Ferguson Technique

We follow a similar procedure for determining the unpaid claim estimate based on the
Bornhuetter-Ferguson technique (Exhibit I, Sheet 2) as presented in the prior chapters for the
development and expected claims techniques. Estimated IBNR is equal to projected ultimate

154
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

claims less reported claims55 and the total unpaid claim estimate is equal to the difference
between projected ultimate claims and paid claims.

Exhibit I, Sheet 2, presents the calculations of the unpaid claim estimate for U.S. Industry Auto.
Columns (2) and (3) contain reported and paid claims data as of December 31, 2007. The
projected ultimate claims, developed in Exhibit I, Sheet 1, are summarized in Columns (4) and
(5). Case outstanding, which are equal to the difference between reported claims and paid claims
as of December 31, 2007, are presented in Column (6). Estimated IBNR is equal to projected
ultimate claims minus reported claims. In Columns (7) and (8), we calculate estimated IBNR
based on the reported and paid Bornhuetter-Ferguson techniques, respectively. The total unpaid
claim estimate is equal to the sum of case outstanding and estimated IBNR. We present the
estimate of total unpaid claims in Columns (9) and (10) based on the reported and paid
Bornhuetter-Ferguson techniques, respectively.

When the Bornhuetter-Ferguson Technique Works and When it Does Not

An advantage of the Bornhuetter-Ferguson technique is that random fluctuations early in the life
of an accident year (or other defined time interval) do not significantly distort the projections. For
example, if several large and unusual claims are reported for an accident year, then the reported
claim development technique may produce overly conservative ultimate claims estimates. This
situation does not, however, seriously distort the Bornhuetter-Ferguson technique.

Actuaries frequently use the Bornhuetter-Ferguson method for long-tail lines of insurance,
particularly for the most immature years, due to the highly leveraged nature of claim development
factors for such lines. Actuaries may also use the Bornhuetter-Ferguson technique if the data is
extremely thin or volatile or both. For example, when an insurer has recently entered a new line
of business or a new territory and there is not yet a credible volume of historical claim
development experience, an actuary may use the Bornhuetter-Ferguson technique. In such
circumstances, the actuary would likely need to rely on benchmarks, either from similar lines at
the same insurer or insurance industry experience, for development patterns and expected claim
ratios (or pure premiums). (See previous comments about the use of industry benchmarks.)

In a discussion of when to use the Bornhuetter-Ferguson method in the paper “The Actuary and
IBNR,” the authors state: “It can be argued that the most prudent course is, when in doubt, to use
expected losses, in as much as it is certainly indicated for volatile lines, and in the case of a stable
line, the expected loss ratio should be predictable enough so that both techniques produce the
same result.”56

The Bornhuetter-Ferguson technique can be a useful method for very short-tail lines as well as
long-tail lines. For very short-tail lines, the IBNR can be set equal to a multiple of the last few
months’ earned premium; this is essentially an application of the Bornhuetter-Ferguson
technique.

55
Recall that the formula for the reported Bornhuetter-Ferguson method is:
Ultimate Claims = Actual Reported Claims + Expected Unreported Claims
Thus, for the reported Bornhuetter-Ferguson projection, the expected unreported claims are equal to the
estimated IBNR.
56
PCAS, 1972.

155
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

The Bornhuetter-Ferguson Method and Cumulative CDFs Less than 1.00

Downward development (i.e., cumulative development factors that are less than 1.00) does occur
for some insurers, for some lines of business. Automobile physical damage and property are
examples of coverages in which actuaries can observe this type of development experience. For
some insurers, salvage and subrogation recoveries lag the reporting and payment of claims, which
can result in report-to-report factors that are less than 1.00. For some insurers, a conservative
philosophy regarding case outstanding can also result in an observed downward development of
reported claims as payments for claims may be less than the case outstanding set by claims
adjusters. For those lines of business for which the actuary derives cumulative claim development
factors that are less than 1.00, we revisit the original premise of the Bornhuetter-Ferguson
method.

At the beginning of this chapter, we refer to Brosius’ description of the Bornhuetter-Ferguson

method as a credibility-weighting between the development method and the expected claims
method. Credibility is concerned with the combination of the projections from these two methods.
The basic formula for calculating the credibility-weighted projection is:

[(Z) x (development method)] + [(1 – Z) x (expected claims method)],

where,
0 Z 1,
Z is the credibility assigned to the development method, and
(1 – Z) is the complement of credibility assigned to the expected claims method.

As noted earlier, the credibility is equal to the percentage of claims developed at a particular stage
of maturity, which is a function of the cumulative claim development factor (Z = 1.000 /
cumulative development factor).

From a theoretical perspective, the credibility-weighting approach of the Bornhuetter-Ferguson

method does not hold true if the cumulative development factor is less than 1.00 since the value
assigned to credibility, Z, is then greater than 1.00. For example, if the cumulative development
factor is 0.93, then the credibility assigned to the development method is equal to 1.075 (1.00 /
0.93). However, as defined above, credibility must be a value between 0 and 1. Thus, a credibility
value of 1.075 is outside of the acceptable range.

While cumulative development factors that are less than 1.00 present a theoretical issue for the
use of the Bornhuetter-Ferguson method, in practice, many actuaries continue to use this method
with such factors. One solution to address this theoretical challenge is to limit the cumulative
development factors to a minimum value of 1.00 when applying the Bornhuetter-Ferguson
technique. (We follow this approach for the examples in this text.) Alternatively, and what
happens quite frequently in practice, is that the actuary will still calculate the reported
Bornhuetter-Ferguson projected ultimate claims using cumulative development factors that are
less than 1.00 but will rely on another technique to select ultimate claims for the year(s) in
question (i.e., years with cumulative development factors less than 1.00). As noted previously in
this chapter, actuaries frequently use the Bornhuetter-Ferguson method for long-tail lines of
insurance, particularly for the most immature years. Cumulative development factors for these
lines and years are typically much greater than 1.00. Nevertheless, it is worthwhile to note that
some actuaries continue to include the Bornhuetter-Ferguson method as part of their analyses
even in the presence of cumulative development factors that are less than 1.00.

156
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

XYZ Insurer

In Exhibit II, Sheets 1 and 2, we present the results of the reported and paid Bornhuetter-Ferguson
methods based on the expected claims developed in Chapter 8 for XYZ Insurer. The presentation
and calculations are identical to the previous example for U.S. Industry Auto (Exhibit I). We will
not examine the results of this projection in detail because we know that the expected claims
estimates underlying the projections are likely inaccurate. Remember that the primary assumption
of the development method does not hold true for XYZ Insurer as a result of the operational and
environmental changes that took place during the experience period. Nevertheless, we derive the
current estimates of expected claims using unadjusted reported and paid claim development
methods. We compare the results of the Bornhuetter-Ferguson method with the expected claims
method and the development method in Exhibit II, Sheet 3 (projected ultimate claims) and in
Exhibit II, Sheet 4 (estimated IBNR). In later chapters, we look at alternative methods that can be
used for developing expected claims for use in a revised Bornhuetter-Ferguson method.

Influence of a Changing Environment on the Bornhuetter-Ferguson Method57

In Chapters 7 and 8, we discuss the performance of the development technique and the expected
claims technique, respectively, during times of change. We continue with these examples using
the Bornhuetter-Ferguson technique. Since the Bornhuetter-Ferguson method is a combination of
the development method and the expected claims method, we will refer you to these prior
chapters for critical input. For example, refer to Chapter 7 for the reported and paid claim
development triangles and the selection of age-to-age factors, and refer to Chapter 8 for the
calculation of expected claims.

Scenario 1 – U.S. PP Auto Steady-State

For Scenarios 1 through 4, we use an expected claim ratio of 70%. Since the steady-state
environment also has a 70% ultimate claim ratio, the Bornhuetter-Ferguson technique generates
an accurate estimate of IBNR. We see in Chapters 7 and 8, that the development and expected
claims techniques also generate accurate IBNR values in a steady-state environment. Detailed
calculations are presented for the Bornhuetter-Ferguson method in the top section of Exhibit III,
Sheet 1.

57
We present the following examples to demonstrate the effect of not changing assumptions on the
resulting projections of ultimate claims and the estimate of unpaid claims. We recognize that the examples
are not necessarily representative of real-life applications of the Bornhuetter-Ferguson method since we
assume that there are no adjustments in expected claims in anticipation of the events that caused higher
claim ratios or changes in business mix. Most insurers have a feel for whether a market is getting softer or
harder, so they would have a sense as to the direction to adjust the expected claims, if not the absolute
amount of adjustment. In addition, actuaries typically use the Bornhuetter-Ferguson technique where
development data is sparse and erratic, which is exactly where the development approaches are very weak.
Hence, we note that the PP Auto examples are biased against a Bornhuetter-Ferguson approach.

157
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

Scenario 2 – U.S. PP Auto Increasing Claim Ratios

The weakness of the expected claims method is also a weakness of the Bornhuetter-Ferguson
method. Remember the original formulae for the reported and paid Bornhuetter-Ferguson
method:

Ultimate Claims = Actual Reported Claims + Expected Unreported Claims

Ultimate Claims = Actual Paid Claims + Expected Unpaid Claims

While projected ultimate claims are increasing between Scenarios 1 and 2, the increases are due
to higher values of actual reported and paid claims and not higher estimates of the expected
unreported and unpaid claims. Since the expected claims estimate does not change in this
example, the expected unreported and unpaid claims do not change in Scenario 2 from the steady-
state values of Scenario 1.

For the reported Bornhuetter-Ferguson technique, the estimated IBNR is identical between the
steady-state environment and the environment with increasing claim ratios. Without a deliberate
change in the expected claim ratio, this method will not respond to a situation with increasing
claim ratios. The paid Bornhuetter-Ferguson performs even worse than the reported Bornhuetter-
Ferguson technique for Scenario 2. The estimate of expected unpaid claims is understated to an
even greater degree than the expected unreported claims. This is due to the longer-term nature of
the payment pattern than the reporting pattern, which implies that the percentage unpaid cannot
be less than the percentage unreported at any age. (See Table 5 of Chapter 7, which summarizes
the reporting and payment patterns.)

Scenario 3 – U.S. PP Auto Increasing Case Outstanding Strength

We present the calculations for Scenario 3 in the top section of Exhibit III, Sheet 2. The reported
Bornhuetter-Ferguson technique produces an estimate of IBNR that is greater than the actual
IBNR for this scenario. Similar to the paid claim development technique, the paid Bornhuetter-
Ferguson method is unaffected by changes only in case outstanding strength.

In Chapter 7, we saw how increases in case outstanding strength led to increases in age-to-age
factors and in cumulative claim development factors. The cumulative claim development factors
are an important input to the Bornhuetter-Ferguson method. Thus, if the cumulative claim
development factors are changing due to increases in case outstanding strength, it will also have
an effect on the Bornhuetter-Ferguson projection. The expected claims, on the other hand, remain
unchanged.

The estimated IBNR, in this scenario, based on the reported Bornhuetter-Ferguson method is
greater than the actual IBNR requirement. However, the overstatement is less for the reported
Bornhuetter-Ferguson method than for the reported claim development method because we did
not increase the expected claims. In Chapter 7, we discuss how there are two forces that
contribute to the excessive estimate of IBNR in the development technique. First, age-to-age
factors increase due to the change in case outstanding adequacy. Second, we then multiply the
resulting higher cumulative claim development factors by the latest valuation of reported claims,
which contains higher reported values due to the increase in case outstanding strength. There is,
in essence, a leveraging effect of higher factors and higher claims in the development technique.
In the Bornhuetter-Ferguson method, the higher cumulative claim development factors result in

158
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

greater percentages of expected unreported claims. However, the same leveraging effect does not
exist since expected claims, not actual claims, are the basis for determining unreported claims in
the Bornhuetter-Ferguson method, and the expected claims did not change in our example.

Scenario 4 – U.S. PP Auto Increasing Claim Ratios and Case Outstanding Strength

We present the detailed calculations for Scenario 4 in the bottom section of Exhibit III, Sheet 2.
We see that the estimated IBNR based on the reported Bornhuetter-Ferguson method is
overstated while the paid Bornhuetter-Ferguson projection is understated, absent a change in the
expected claims assumption.

For both projections, the expected claims used in the example are too low. This is the reason that
the paid Bornhuetter-Ferguson method produces an estimate of IBNR that is $443,260 lower than
the actual IBNR. This is the same difference between estimated and actual IBNR that we saw in
Scenario 2, where claim ratios increased and case outstanding strength remained stable. Since the
payment pattern is unaffected by changes in case outstanding adequacy, there is no effect on the
paid Bornhuetter-Ferguson method. The sole reason for the inadequacy of the paid Bornhuetter-
Ferguson method is the understatement of expected claims.

In Scenario 2 (increasing claim ratios and stable case outstanding strength), we see that the
reported Bornhuetter-Ferguson technique produces an estimated IBNR that is lower than the
actual IBNR. In Scenario 3 (stable claim ratio and increasing case outstanding strength), the
estimated IBNR based on the reported Bornhuetter-Ferguson method is too high. These two
factors work in opposition in Scenario 4, in which both claim ratios and case outstanding strength
are increasing. Even though expected claims are too low for Scenario 4, there is more than an
offsetting effect from the higher cumulative development factors leading to an estimated IBNR
for the reported Bornhuetter-Ferguson technique that is $112,773 higher than the actual IBNR.

In Scenario 4, with increasing claim ratios and case outstanding strengthening, the difference
from the actual IBNR using the Bornhuetter-Ferguson method could be positive or negative
depending on the extent of case outstanding strengthening and deteriorating claim ratio.

U.S. Auto Steady-State (No Change in Product Mix)

In the last two examples, we present the projections for a combined portfolio of private passenger
and commercial automobile. In the top section of Exhibit IV, we summarize the calculations
assuming a steady-state (i.e., no change in product mix). Similar to our projections using the
claim development and expected claims techniques, we demonstrate in Exhibit IV, that the
Bornhuetter-Ferguson technique will generate the correct IBNR requirement if there is no change
in the product mix.

U.S. Auto Changing Product Mix

In the final example, we assume that the volume of commercial automobile insurance is
increasing at a greater rate than that of private passenger automobile insurance. In the bottom
section of Exhibit IV, we quickly observe that both the reported and paid Bornhuetter-Ferguson

159
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

methods produce estimated IBNR that is lower than the actual IBNR. This is due to the expected
claim ratio assumption that is unchanged from the U.S. Auto Steady-State.

Since the commercial automobile segment is growing at a greater rate than the private passenger
auto segment, and since commercial automobile has a higher ultimate claim ratio, the actuary
needs to modify the expected claim ratio assumption. Without such modification, the estimated
IBNR from both the expected claims and the Bornhuetter-Ferguson methods proves inadequate.
The reporting and payment patterns also require change. With an increasing proportion of
commercial automobile, the reporting and payment patterns lengthen, and thus result in the
requirement for a higher IBNR value.

Benktander Technique

An often-cited advantage of the Bornhuetter-Ferguson technique versus the development

technique is stability in the presence of sparse data. However, since the estimate of unpaid claims
for the most recent accident years using the Bornhuetter-Ferguson technique is heavily dependant
on the actuary’s judgment when determining the expected claims, actual claims emergence for
these years may be ignored to some extent.

The Benktander method, introduced in 1976, is a credibility-weighted average of the Bornhuetter-

Ferguson technique and the development technique. The advantage cited by the authors is that
this method will prove more responsive than the Bornhuetter-Ferguson technique and more stable
than the development technique. (For further information on the development of the technique
and underlying proofs of the methodology, see Thomas Mack’s 2000 ASTIN Bulletin paper
“Credible Claims Reserves: The Benktander Method.”)

The Benktander method is often considered an iterative Bornhuetter-Ferguson method. The only
difference in the two methods is the derivation of the expected claims. As we discuss in Chapter 8
– Expected Claims Technique, most insurers use an expected claim ratio and earned premium to
determine expected claims and many self-insurers use pure premiums and exposures. Such
expected claims become the input for the Bornhuetter-Ferguson technique. In the Benktander
technique, the expected claims are the projected ultimate claims from an initial Bornhuetter-
Ferguson projection – thus, the reference to the Benktander method as an iterative Bornhuetter-
Ferguson method. It is interesting to note that the Benktander projection of ultimate claims will
approach the projected ultimate claims produced by the development technique after sufficient
iterations. (See Thomas Mack’s 2000 ASTIN paper for the detailed proof.)

In Exhibits V and VI, we present the Benktander technique using our six examples of changing
environments. We follow the same exhibit format that was presented earlier in this chapter for the
Bornhuetter-Ferguson technique. The only difference between the Bornhuetter-Ferguson
projections in Exhibits III and IV and the Benktander projections in Exhibits V and VI are the
expected claims. In the Bornhuetter-Ferguson projections, we derive the expected claims based
on the initial expected claim ratio multiplied by the earned premium. In the Benktander
projections, the expected claims are based on the Bornhuetter-Ferguson projections (from
Exhibits III and IV).

In the following table, we summarize the differences from the true unpaid claims, in thousands of
dollars, based on the Bornhuetter-Ferguson technique and the Benktander technique for the six
examples related to changing environments.

160
Estimating Unpaid Claims Using Basic Techniques
Chapter 9 - Bornhuetter-Ferguson Technique

Difference from True IBNR ($000) Using

Bornhuetter-Ferguson Method Benktander Method
Example Name Reported Paid Reported Paid
U.S. PP Auto Steady-State 0 0 0 0
U.S. PP Auto Increasing Claim Ratios 163 443 29 196
U.S. PP Auto Increasing Case -205 0 -239 0
Outstanding Strength
U.S. PP Auto Increasing Claim Ratios -113 443 -300 196
and Case Outstanding Strength
U.S. Auto Steady-State 0 0 0 0
U.S. Auto Changing Product Mix 223 400 233 498

The Benktander technique is significantly more responsive to changes in the underlying claim
ratio but is less responsive to changes in the case outstanding adequacy. The Benktander
technique is also less responsive to changes in the product mix than the Bornhuetter-Ferguson
technique.

Note that the Benktander method always gives greater credibility to the development technique.
Thus, where there are no changes in the underlying claim development patterns, we expect the
Benktander method to be more responsive than the Bornhuetter-Ferguson method. Where claim
development patterns are changing, the Benktander method may not produce the most appropriate
estimate as seen in the examples with changing case outstanding adequacy and changes in
product mix. With the changing product mix, the Benktander method would have proven
responsive to the changing claim ratio but not to the changes in the underlying development
patterns.

161
Chapter 9 - Bornhuetter-Ferguson Technique Exhibit I
U.S. Industry Auto Sheet 1
Projection of Ultimate Claims Using Reported and Paid Claims ($000)

Projected Ultimate Claims

Accident Expected CDF to Ultimate Percentage Expected Claims Claims at 12/31/07 Using B-F Method with
Year Claims Reported Paid Unreported Unpaid Unreported Unpaid Reported Paid Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

1998 51,430,657 1.000 1.002 0.0% 0.2% 0 102,656 47,742,304 47,644,187 47,742,304 47,746,843
1999 51,408,736 1.000 1.004 0.0% 0.4% 0 204,816 51,185,767 51,000,534 51,185,767 51,205,350
2000 51,680,983 1.001 1.006 0.1% 0.6% 51,629 308,236 54,837,929 54,533,225 54,889,558 54,841,461
2001 54,408,716 1.003 1.011 0.3% 1.1% 162,738 591,984 56,299,562 55,878,421 56,462,300 56,470,405
2002 59,421,665 1.006 1.020 0.6% 2.0% 354,404 1,165,131 58,592,712 57,807,215 58,947,116 58,972,346
2003 56,318,302 1.011 1.040 1.1% 3.8% 612,761 2,166,089 57,565,344 55,930,654 58,178,105 58,096,743
2004 59,646,290 1.023 1.085 2.2% 7.8% 1,341,021 4,672,751 56,976,657 53,774,672 58,317,678 58,447,423
2005 61,174,953 1.051 1.184 4.9% 15.5% 2,968,528 9,506,918 56,786,410 50,644,994 59,754,938 60,151,912
2006 61,926,981 1.110 1.404 9.9% 28.8% 6,136,908 17,819,445 54,641,339 43,606,497 60,778,247 61,425,942
2007 61,864,556 1.292 2.390 22.6% 58.2% 13,981,773 35,979,805 48,853,563 27,229,969 62,835,336 63,209,774

Total 569,281,839 25,609,761 72,517,830 543,481,587 498,050,368 569,091,348 570,568,198

Column Notes:
(2) Developed in Chapter 8, Exhibit II, Sheet 1.
(3) and (4) Developed in Chapter 7, Exhibit I, Sheets 1 and 2.
(5) = [1.00 - (1.00 / (3))].
(6) = [1.00 - (1.00 / (4))].
(7) = [(2) x (5)].
(8) = [(2) x (6)].
(9) and (10) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(11) = [(7) + (9)].
(12) = [(8) + (10)].

162

Exhibits Combined.xls 9_1_1 04/03/2009 - 2:58 PM

Chapter 9 - Bornhuetter-Ferguson Technique Exhibit I
U.S. Industry Auto Sheet 2
Development of Unpaid Claim Estimate ($000)

Unpaid Claim Estimate at 12/31/07

Projected Ultimate Claims Case IBNR - Based on Total - Based on
Accident Claims at 12/31/07 Using B-F Method with Outstanding B-F Method with B-F Method with
Year Reported Paid Reported Paid at 12/31/07 Reported Paid Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

1998 47,742,304 47,644,187 47,742,304 47,746,843 98,117 0 4,539 98,117 102,656

1999 51,185,767 51,000,534 51,185,767 51,205,350 185,233 0 19,583 185,233 204,816
2000 54,837,929 54,533,225 54,889,558 54,841,461 304,704 51,629 3,532 356,333 308,236
2001 56,299,562 55,878,421 56,462,300 56,470,405 421,141 162,738 170,843 583,879 591,984
2002 58,592,712 57,807,215 58,947,116 58,972,346 785,497 354,404 379,634 1,139,901 1,165,131
2003 57,565,344 55,930,654 58,178,105 58,096,743 1,634,690 612,761 531,399 2,247,451 2,166,089
2004 56,976,657 53,774,672 58,317,678 58,447,423 3,201,985 1,341,021 1,470,766 4,543,006 4,672,751
2005 56,786,410 50,644,994 59,754,938 60,151,912 6,141,416 2,968,528 3,365,502 9,109,944 9,506,918
2006 54,641,339 43,606,497 60,778,247 61,425,942 11,034,842 6,136,908 6,784,603 17,171,750 17,819,445
2007 48,853,563 27,229,969 62,835,336 63,209,774 21,623,594 13,981,773 14,356,211 35,605,367 35,979,805

Total 543,481,587 498,050,368 569,091,348 570,568,198 45,431,219 25,609,761 27,086,611 71,040,980 72,517,830

Column Notes:
(2) and (3) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(4) and (5) Developed in Exhibit I, Sheet 1.
(6) = [(2) - (3)].
(7) = [(4) - (2)].
(8) = [(5) - (2)].
(9) = [(6) + (7)].
(10) = [(6) + (8)].

163

Exhibits Combined.xls 9_1_2 04/03/2009 - 2:58 PM

Chapter 9 - Bornhuetter-Ferguson Technique Exhibit II
XYZ Insurer - Auto BI Sheet 1
Projection of Ultimate Claims Using Reported and Paid Claims ($000)

Projected Ultimate Claims

Accident Expected CDF to Ultimate Percentage Expected Claims Claims at 12/31/08 Using B-F Method with
Year Claims Reported Paid Unreported Unpaid Unreported Unpaid Reported Paid Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

1998 15,660 1.000 1.010 0.0% 1.0% 0 155 15,822 15,822 15,822 15,977
1999 24,665 1.000 1.014 0.0% 1.4% 0 341 25,107 24,817 25,107 25,158
2000 35,235 1.000 1.031 0.0% 3.0% 0 1,059 37,246 36,782 37,246 37,841
2001 39,150 1.000 1.054 0.0% 5.1% 0 2,006 38,798 38,519 38,798 40,525
2002 47,906 1.003 1.116 0.3% 10.4% 143 4,980 48,169 44,437 48,312 49,417
2003 54,164 1.013 1.268 1.3% 21.1% 695 11,448 44,373 39,320 45,068 50,768
2004 86,509 1.064 1.525 6.0% 34.4% 5,204 29,782 70,288 52,811 75,492 82,593
2005 108,172 1.085 2.007 7.8% 50.2% 8,474 54,275 70,655 40,026 79,129 94,301
2006 70,786 1.196 3.160 16.4% 68.4% 11,600 48,386 48,804 22,819 60,404 71,205
2007 39,835 1.512 6.569 33.9% 84.8% 13,489 33,771 31,732 11,865 45,221 45,636
2008 39,433 2.551 21.999 60.8% 95.5% 23,975 37,640 18,632 3,409 42,607 41,049

Total 561,516 63,581 223,842 449,626 330,629 513,207 554,471

Column Notes:
(2) Developed in Chapter 8, Exhibit III, Sheet 1.
(3) and (4) Developed in Chapter 7, Exhibit II, Sheets 1 and 2, capped at a minimum of 1.00.
(5) = [1.00 - (1.00 / (3))].
(6) = [1.00 - (1.00 / (4))].
(7) = [(2) x (5)].
(8) = [(2) x (6)].
(9) and (10) Based on data from XYZ Insurer.
(11) = [(7) + (9)].
(12) = [(8) + (10)].

164

Exhibits Combined.xls 9_2_1 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit II
XYZ Insurer - Auto BI Sheet 2
Development of Unpaid Claim Estimate ($000)

Unpaid Claim Estimate at 12/31/08

Projected Ultimate Claims Case IBNR - Based on Total - Based on
Accident Claims at 12/31/08 Using B-F Method with Outstanding B-F Method with B-F Method with
Year Reported Paid Reported Paid at 12/31/08 Reported Paid Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

1998 15,822 15,822 15,822 15,977 0 0 155 0 155

1999 25,107 24,817 25,107 25,158 290 0 51 290 341
2000 37,246 36,782 37,246 37,841 465 0 595 465 1,059
2001 38,798 38,519 38,798 40,525 278 0 1,728 278 2,006
2002 48,169 44,437 48,312 49,417 3,731 143 1,248 3,875 4,980
2003 44,373 39,320 45,068 50,768 5,052 695 6,396 5,747 11,448
2004 70,288 52,811 75,492 82,593 17,477 5,204 12,305 22,681 29,782
2005 70,655 40,026 79,129 94,301 30,629 8,474 23,646 39,103 54,275
2006 48,804 22,819 60,404 71,205 25,985 11,600 22,401 37,585 48,386
2007 31,732 11,865 45,221 45,636 19,867 13,489 13,904 33,356 33,771
2008 18,632 3,409 42,607 41,049 15,223 23,975 22,417 39,198 37,640

Total 449,626 330,629 513,207 554,471 118,997 63,581 104,845 182,578 223,842

Column Notes:
(2) and (3) Based on data from XYZ Insurer.
(4) and (5) Developed in Exhibit II, Sheet 1.
(6) = [(2) - (3)].
(7) = [(4) - (2)].
(8) = [(5) - (2)].
(9) = [(6) + (7)].
(10) = [(6) + (8)].

165

Exhibits Combined.xls 9_2_2 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit II
XYZ Insurer - Auto BI Sheet 3
Summary of Ultimate Claims ($000)

Projected Ultimate Claims

Accident Claims at 12/31/08 Development Method Expected B-F Method
Year Reported Paid Reported Paid Claims Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8)

1998 15,822 15,822 15,822 15,980 15,660 15,822 15,977

1999 25,107 24,817 25,082 25,164 24,665 25,107 25,158
2000 37,246 36,782 36,948 37,922 35,235 37,246 37,841
2001 38,798 38,519 38,487 40,600 39,150 38,798 40,525
2002 48,169 44,437 48,313 49,592 47,906 48,312 49,417
2003 44,373 39,320 44,950 49,858 54,164 45,068 50,768
2004 70,288 52,811 74,787 80,537 86,509 75,492 82,593
2005 70,655 40,026 76,661 80,333 108,172 79,129 94,301
2006 48,804 22,819 58,370 72,108 70,786 60,404 71,205
2007 31,732 11,865 47,979 77,941 39,835 45,221 45,636
2008 18,632 3,409 47,530 74,995 39,433 42,607 41,049

Total 449,626 330,629 514,929 605,030 561,516 513,207 554,471

Column Notes:
(2) and (3) Based on data from XYZ Insurer.
(4) and (5) Developed in Chapter 7, Exhibit II, Sheet 3.
(6) Developed in Chapter 8, Exhibit III, Sheet 1.
(7) and (8) Developed in Exhibit II, Sheet 1.

166

Exhibits Combined.xls 9_2_3 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit II
XYZ Insurer - Auto BI Sheet 4
Summary of IBNR ($000)

Case Estimated IBNR

Accident Outstanding Development Method Expected B-F Method
Year at 12/31/08 Reported Paid Claims Reported Paid
(1) (2) (3) (4) (5) (6) (7)

1998 0 0 158 - 162 0 155

1999 290 - 25 58 - 442 0 51
2000 465 - 298 676 - 2,011 0 595
2001 278 - 310 1,802 352 0 1,728
2002 3,731 145 1,423 - 262 143 1,248
2003 5,052 577 5,485 9,791 695 6,396
2004 17,477 4,498 10,249 16,221 5,204 12,305
2005 30,629 6,006 9,678 37,517 8,474 23,646
2006 25,985 9,566 23,304 21,982 11,600 22,401
2007 19,867 16,247 46,209 8,103 13,489 13,904
2008 15,223 28,898 56,363 20,801 23,975 22,417

Total 118,997 65,303 155,405 111,890 63,581 104,845

Column Notes:
(2) Based on data from XYZ Insurer.
(3) and (4) Estimated in Chapter 7, Exhibit II, Sheet 4.
(5) Estimated in Chapter 8, Exhibit III, Sheet 3.
(6) and (7) Estimated in Exhibit II, Sheet 2.

167

Exhibits Combined.xls 9_2_4 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit III
Impact of Changing Conditions Sheet 1
U.S. PP Auto - Development of Unpaid Claim Estimate

Age of Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Expected Claims at 12/31/08 CDF to Ultimate Expected Percentage Using B-F Method with Using B-F Method with Actual Using B-F Method with
Year at 12/31/08 Claims Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Steady-State
1999 120 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 884,463 857,661 1.010 1.042 1.0% 4.0% 893,397 893,397 8,934 8,934 8,934 0 0
2005 48 938,067 919,306 863,022 1.020 1.087 2.0% 8.0% 938,067 938,067 18,761 18,761 18,761 0 0
2006 36 984,970 935,722 827,375 1.053 1.190 5.0% 16.0% 984,970 984,970 49,249 49,249 49,249 0 -0
2007 24 1,034,219 930,797 734,295 1.111 1.408 10.0% 29.0% 1,034,219 1,034,219 103,422 103,422 103,422 0 0
2008 12 1,085,930 836,166 456,090 1.299 2.381 23.0% 58.0% 1,085,930 1,085,930 249,764 249,764 249,764 0 -0

Total 8,804,525 8,365,887 7,573,548 8,804,525 8,804,525 438,638 438,638 438,638 0 0

Increasing Claim Ratios

1999 120 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 1,010,815 980,184 1.010 1.042 1.0% 4.0% 1,019,749 1,015,920 8,934 5,105 10,210 1,276 5,105
2005 48 938,067 1,116,300 1,047,955 1.020 1.087 2.0% 8.0% 1,135,061 1,123,000 18,761 6,700 22,782 4,020 16,081
2006 36 984,970 1,203,071 1,063,768 1.053 1.190 5.0% 16.0% 1,252,319 1,221,363 49,249 18,292 63,320 14,071 45,027
2007 24 1,034,219 1,263,224 996,544 1.111 1.408 10.0% 29.0% 1,366,646 1,296,467 103,422 33,243 140,358 36,936 107,116
2008 12 1,085,930 1,194,523 651,558 1.299 2.381 23.0% 58.0% 1,444,287 1,281,397 249,764 86,874 356,805 107,042 269,931

Total 8,804,525 9,647,366 8,575,112 10,086,004 9,806,090 438,638 158,724 601,984 163,346 443,260

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) See Chapter 8, Exhibit IV, Sheet 1.
(4) and (5) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit III, Sheets 2 through 5.
(6) and (7) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit III, Sheets 2 through 5.
(8) = [1.00 - (1.00 / (6))].
(9) = [1.00 - (1.00 / (7))].
(10) = [((3) x (8)) + (4)].
(11) = [((3) x (9)) + (5)].
(12) = [(10) - (4)].
(13) = [(11) - (4)].
(14) Developed in Chapter 7, Exhibit III, Sheet 1.
(15) = [(14) - (12)].
(16) = [(14) - (13)].
168

Exhibits Combined.xls 9_3_1 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit III
Impact of Changing Conditions Sheet 2
U.S. PP Auto - Development of Unpaid Claim Estimate

Age of Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Expected Claims at 12/31/08 CDF to Ultimate Expected Percentage Using B-F Method with Using B-F Method with Actual Using B-F Method with
Year at 12/31/08 Claims Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Increasing Case Outstanding Strength

1999 120 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 884,463 857,661 1.010 1.042 1.0% 4.0% 893,397 893,397 8,934 8,934 8,934 0 0
2005 48 938,067 933,377 863,022 1.020 1.087 1.9% 8.0% 951,395 938,067 18,018 4,690 4,690 - 13,328 0
2006 36 984,970 962,808 827,375 1.055 1.190 5.2% 16.0% 1,013,733 984,970 50,925 22,162 22,162 - 28,763 -0
2007 24 1,034,219 979,922 734,295 1.119 1.408 10.6% 29.0% 1,089,655 1,034,219 109,733 54,296 54,296 - 55,437 0
2008 12 1,085,930 931,185 456,090 1.318 2.381 24.1% 58.0% 1,193,385 1,085,930 262,200 154,745 154,745 - 107,455 -0

Total 8,804,525 8,551,189 7,573,548 9,009,508 8,804,525 458,319 253,336 253,336 - 204,983 0

Increasing Claim Ratios and Case Outstanding Strength

1999 120 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 1,010,815 980,184 1.010 1.042 1.0% 4.0% 1,019,749 1,015,920 8,934 5,105 10,210 1,276 5,105
2005 48 938,067 1,133,386 1,047,955 1.019 1.087 1.9% 8.0% 1,151,324 1,123,000 17,938 - 10,386 5,695 - 12,243 16,081
2006 36 984,970 1,237,897 1,063,768 1.055 1.190 5.2% 16.0% 1,289,001 1,221,363 51,105 - 16,533 28,494 - 22,611 45,027
2007 24 1,034,219 1,329,895 996,544 1.120 1.408 10.7% 29.0% 1,440,327 1,296,467 110,432 - 33,427 73,688 - 36,744 107,116
2008 12 1,085,930 1,330,264 651,558 1.320 2.381 24.3% 58.0% 1,593,780 1,281,397 263,516 - 48,867 221,064 - 42,452 269,931

Total 8,804,525 9,901,689 8,575,112 10,362,123 9,806,090 460,434 - 95,600 347,660 - 112,773 443,260

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) See Chapter 8, Exhibit IV, Sheet 2.
(4) and (5) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit III, Sheets 6 through 9.
(6) and (7) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit III, Sheets 6 through 9.
(8) = [1.00 - (1.00 / (6))].
(9) = [1.00 - (1.00 / (7))].
(10) = [((3) x (8)) + (4)].
(11) = [((3) x (9)) + (5)].
(12) = [(10) - (4)].
(13) = [(11) - (4)].
(14) Developed in Chapter 7, Exhibit III, Sheet 1.
(15) = [(14) - (12)].
(16) = [(14) - (13)].
169

Exhibits Combined.xls 9_3_2 04/03/2009 - 2:58 PM

 Chapter 9 - Bornhuetter-Ferguson Technique Exhibit IV
Impact of Change in Product Mix Example
U.S. Auto - Development of Unpaid Claim Estimate

Age of Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Expected Claims at 12/31/08 CDF to Ultimate Expected Percentage Using B-F Method with Using B-F Method with Actual Using B-F Method with
Year at 12/31/08 Claims Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)

Steady-State (No Change in Product Mix)

1999 120 1,500,000 1,500,000 1,500,000 1.000 1.000 0.0% 0.0% 1,500,000 1,500,000 0 0 0 0 0
2000 108 1,575,000 1,575,000 1,566,600 1.000 1.005 0.0% 0.5% 1,575,000 1,575,000 0 0 0 0 0
2001 96 1,653,750 1,653,750 1,628,393 1.000 1.016 0.0% 1.5% 1,653,750 1,653,750 0 0 0 0 0
2002 84 1,736,438 1,736,438 1,700,551 1.000 1.021 0.0% 2.1% 1,736,438 1,736,438 0 0 0 0 0
2003 72 1,823,259 1,814,751 1,757,622 1.005 1.037 0.5% 3.6% 1,823,259 1,823,259 8,509 8,509 8,509 0 0
2004 60 1,914,422 1,885,068 1,786,794 1.016 1.071 1.5% 6.7% 1,914,422 1,914,422 29,354 29,354 29,354 0 0
2005 48 2,010,143 1,948,499 1,742,124 1.032 1.154 3.1% 13.3% 2,010,143 2,010,143 61,644 61,644 61,644 0 0
2006 36 2,110,651 1,937,577 1,581,581 1.089 1.335 8.2% 25.1% 2,110,651 2,110,651 173,073 173,073 173,073 0 0
2007 24 2,216,183 1,852,729 1,277,999 1.196 1.734 16.4% 42.3% 2,216,183 2,216,183 363,454 363,454 363,454 0 0
2008 12 2,326,992 1,568,393 729,124 1.484 3.191 32.6% 68.7% 2,326,992 2,326,992 758,599 758,599 758,599 0 0

Total 18,866,839 17,472,204 15,270,788 18,866,839 18,866,839 1,394,634 1,394,634 1,394,634 0 0

Changing Product Mix

1999 120 1,500,000 1,500,000 1,500,000 1.000 1.000 0.0% 0.0% 1,500,000 1,500,000 0 0 0 0 0
2000 108 1,575,000 1,575,000 1,566,600 1.000 1.005 0.0% 0.5% 1,575,000 1,575,000 0 0 0 0 0
2001 96 1,653,750 1,653,750 1,628,393 1.000 1.016 0.0% 1.5% 1,653,750 1,653,750 0 0 0 0 0
2002 84 1,736,438 1,736,438 1,700,551 1.000 1.021 0.0% 2.1% 1,736,438 1,736,438 0 0 0 0 0
2003 72 1,823,259 1,814,751 1,757,622 1.005 1.037 0.5% 3.6% 1,823,259 1,823,259 8,509 8,509 8,509 0 0
2004 60 1,914,422 1,885,068 1,786,794 1.016 1.071 1.5% 6.7% 1,914,422 1,914,422 29,354 29,354 29,354 0 0
2005 48 2,249,446 2,193,545 1,951,435 1.032 1.154 3.1% 13.3% 2,262,528 2,251,361 68,983 57,816 71,855 2,872 14,039
2006 36 2,673,012 2,471,446 1,983,482 1.090 1.336 8.3% 25.2% 2,692,025 2,656,353 220,579 184,907 239,057 18,478 54,150
2007 24 3,211,085 2,680,487 1,766,164 1.200 1.750 16.7% 42.9% 3,216,658 3,142,865 536,171 462,378 596,924 60,753 134,547
2008 12 3,897,387 2,556,695 1,097,644 1.503 3.273 33.5% 69.4% 3,860,964 3,804,378 1,304,270 1,247,684 1,445,385 141,115 197,702

Total 22,233,799 20,067,179 16,738,684 22,235,045 22,057,826 2,167,866 1,990,647 2,391,084 223,219 400,438

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) See Chapter 8, Exhibit V.
(4) and (5) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit IV, Sheets 2 through 5.
(6) and (7) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit IV, Sheets 2 through 5.
(8) = [1.00 - (1.00 / (6))].
(9) = [1.00 - (1.00 / (7))].
(10) = [((3) x (8)) + (4)].
(11) = [((3) x (9)) + (5)].
(12) = [(10) - (4)].
(13) = [(11) - (4)].
(14) Developed in Chapter 7, Exhibit IV, Sheet 1.
(15) = [(14) - (12)].
(16) = [(14) - (13)].
170

Exhibits Combined.xls 9_4 04/03/2009 - 2:58 PM

Chapter 9 - Bornhuetter-Ferguson Technique Exhibit V
Impact of Changing Conditions Sheet 1
U.S. PP Auto - Development of Unpaid Claim Estimate Using Gunnar Benktander Method

Age of Expected Ultimate Claims Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Using B-F Method with Claims at 12/31/08 CDF to Ultimate Expected Percentage Using G-B Method with Using G-B Method with Actual Using G-B Method with
Year at 12/31/08 Reported Paid Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)

Steady-State
1999 120 700,000 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 893,397 884,463 857,661 1.010 1.042 1.0% 4.0% 893,397 893,397 8,934 8,934 8,934 0 0
2005 48 938,067 938,067 919,306 863,022 1.020 1.087 2.0% 8.0% 938,067 938,067 18,761 18,761 18,761 0 0
2006 36 984,970 984,970 935,722 827,375 1.053 1.190 5.0% 16.0% 984,970 984,970 49,249 49,249 49,249 0 -0
2007 24 1,034,219 1,034,219 930,797 734,295 1.111 1.408 10.0% 29.0% 1,034,219 1,034,219 103,422 103,422 103,422 0 0
2008 12 1,085,930 1,085,930 836,166 456,090 1.299 2.381 23.0% 58.0% 1,085,930 1,085,930 249,764 249,764 249,764 0 -0

Total 8,804,525 8,804,525 8,365,887 7,573,548 8,804,525 8,804,525 438,638 438,638 438,638 0 -0

Increasing Claim Ratios

1999 120 700,000 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 1,019,749 1,015,920 1,010,815 980,184 1.010 1.042 1.0% 4.0% 1,021,012 1,020,821 10,197 10,006 10,210 13 204
2005 48 1,135,061 1,123,000 1,116,300 1,047,955 1.020 1.087 2.0% 8.0% 1,139,001 1,137,795 22,701 21,495 22,782 80 1,286
2006 36 1,252,319 1,221,363 1,203,071 1,063,768 1.053 1.190 5.0% 16.0% 1,265,687 1,259,186 62,616 56,115 63,320 704 7,204
2007 24 1,366,646 1,296,467 1,263,224 996,544 1.111 1.408 10.0% 29.0% 1,399,889 1,372,519 136,665 109,295 140,358 3,694 31,064
2008 12 1,444,287 1,281,397 1,194,523 651,558 1.299 2.381 23.0% 58.0% 1,526,709 1,394,768 332,186 200,245 356,805 24,620 156,560

Total 10,086,004 9,806,090 9,647,366 8,575,112 10,220,240 10,053,031 572,874 405,665 601,984 29,110 196,319

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) and (4) Developed in Exhibit III, Sheet 1.
(5) and (6) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit III, Sheets 2 through 5.
(7) and (8) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit III, Sheets 2 through 5.
(9) = [1.00 - (1.00 / (7))].
(10) = [1.00 - (1.00 / (8))].
(11) = [((3) x (9)) + (5)].
(12) = [((4) x (10)) + (6)].
(13) = [(11) - (5)].
(14) = [(12) - (5)].
(15) Developed in Chapter 7, Exhibit III, Sheet 1.
(16) = [(15) - (13)].
(17) = [(15) - (14)].

171

Exhibits Combined.xls 9_5_1 04/03/2009 - 2:58 PM

Chapter 9 - Bornhuetter-Ferguson Technique Exhibit V
Impact of Changing Conditions Sheet 2
U.S. PP Auto - Development of Unpaid Claim Estimate Using Gunnar Benktander Method

Age of Expected Ultimate Claims Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Using B-F Method with Claims at 12/31/08 CDF to Ultimate Expected Percentage Using G-B Method with Using G-B Method with Actual Using G-B Method with
Year at 12/31/08 Reported Paid Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)

Increasing Case Outstanding Strength

1999 120 700,000 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 893,397 893,397 884,463 857,661 1.010 1.042 1.0% 4.0% 893,397 893,397 8,934 8,934 8,934 0 0
2005 48 951,395 938,067 933,377 863,022 1.020 1.087 1.9% 8.0% 951,651 938,067 18,274 4,690 4,690 - 13,584 0
2006 36 1,013,733 984,970 962,808 827,375 1.055 1.190 5.2% 16.0% 1,015,221 984,970 52,412 22,162 22,162 - 30,250 -0
2007 24 1,089,655 1,034,219 979,922 734,295 1.119 1.408 10.6% 29.0% 1,095,537 1,034,219 115,615 54,296 54,296 - 61,319 0
2008 12 1,193,385 1,085,930 931,185 456,090 1.318 2.381 24.1% 58.0% 1,219,330 1,085,930 288,146 154,745 154,745 - 133,401 -0

Total 9,009,508 8,804,525 8,551,189 7,573,548 9,043,078 8,804,525 491,890 253,336 253,336 - 238,553 -0

Increasing Claim Ratios and Case Outstanding Strength

1999 120 700,000 700,000 700,000 700,000 1.000 1.000 0.0% 0.0% 700,000 700,000 0 0 0 0 0
2000 108 735,000 735,000 735,000 735,000 1.000 1.000 0.0% 0.0% 735,000 735,000 0 0 0 0 0
2001 96 771,750 771,750 771,750 764,033 1.000 1.010 0.0% 1.0% 771,750 771,750 0 0 0 0 0
2002 84 810,338 810,338 810,338 802,234 1.000 1.010 0.0% 1.0% 810,338 810,338 0 0 0 0 0
2003 72 850,854 850,854 842,346 833,837 1.010 1.020 1.0% 2.0% 850,854 850,854 8,509 8,509 8,509 0 0
2004 60 1,019,749 1,015,920 1,010,815 980,184 1.010 1.042 1.0% 4.0% 1,021,012 1,020,821 10,197 10,006 10,210 13 204
2005 48 1,151,324 1,123,000 1,133,386 1,047,955 1.019 1.087 1.9% 8.0% 1,155,402 1,137,795 22,016 4,409 5,695 - 16,321 1,286
2006 36 1,289,001 1,221,363 1,237,897 1,063,768 1.055 1.190 5.2% 16.0% 1,304,776 1,259,186 66,879 21,289 28,494 - 38,385 7,204
2007 24 1,440,327 1,296,467 1,329,895 996,544 1.120 1.408 10.7% 29.0% 1,483,691 1,372,519 153,796 42,625 73,688 - 80,108 31,064
2008 12 1,593,780 1,281,397 1,330,264 651,558 1.320 2.381 24.3% 58.0% 1,717,017 1,394,768 386,753 64,504 221,064 - 165,689 156,560

Total 10,362,123 9,806,090 9,901,689 8,575,112 10,549,840 10,053,031 648,150 151,342 347,660 - 300,490 196,319

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) and (4) Developed in Exhibit III, Sheet 2.
(5) and (6) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit III, Sheets 6 through 9.
(7) and (8) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit III, Sheets 6 through 9.
(9) = [1.00 - (1.00 / (7))].
(10) = [1.00 - (1.00 / (8))].
(11) = [((3) x (9)) + (5)].
(12) = [((4) x (10)) + (6)].
(13) = [(11) - (5)].
(14) = [(12) - (5)].
(15) Developed in Chapter 7, Exhibit III, Sheet 1.
(16) = [(15) - (13)].
(17) = [(15) - (14)].

172

Exhibits Combined.xls 9_5_2 04/03/2009 - 2:58 PM

Chapter 9 - Bornhuetter-Ferguson Technique Exhibit VI
Impact of Change in Product Mix Example
U.S. Auto - Development of Unpaid Claim Estimate Using Gunnar Benktander Method

Age of Expected Ultimate Claims Projected Ultimate Claims Estimated IBNR Diff from Actual IBNR
Accident Accident Year Using B-F Method with Claims at 12/31/08 CDF to Ultimate Expected Percentage Using G-B Method with Using G-B Method with Actual Using G-B Method with
Year at 12/31/08 Reported Paid Reported Paid Reported Paid Unreported Unpaid Reported Paid Reported Paid IBNR Reported Paid
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)

Steady-State (No Change in Product Mix)

1999 120 1,500,000 1,500,000 1,500,000 1,500,000 1.000 1.000 0.0% 0.0% 1,500,000 1,500,000 0 0 0 0 0
2000 108 1,575,000 1,575,000 1,575,000 1,566,600 1.000 1.005 0.0% 0.5% 1,575,000 1,575,000 0 0 0 0 0
2001 96 1,653,750 1,653,750 1,653,750 1,628,393 1.000 1.016 0.0% 1.5% 1,653,750 1,653,750 0 0 0 0 0
2002 84 1,736,438 1,736,438 1,736,438 1,700,551 1.000 1.021 0.0% 2.1% 1,736,438 1,736,438 0 0 0 0 0
2003 72 1,823,259 1,823,259 1,814,751 1,757,622 1.005 1.037 0.5% 3.6% 1,823,259 1,823,259 8,509 8,509 8,509 0 0
2004 60 1,914,422 1,914,422 1,885,068 1,786,794 1.016 1.071 1.5% 6.7% 1,914,422 1,914,422 29,354 29,354 29,354 0 0
2005 48 2,010,143 2,010,143 1,948,499 1,742,124 1.032 1.154 3.1% 13.3% 2,010,143 2,010,143 61,644 61,644 61,644 0 0
2006 36 2,110,651 2,110,651 1,937,577 1,581,581 1.089 1.335 8.2% 25.1% 2,110,651 2,110,651 173,073 173,073 173,073 0 0
2007 24 2,216,183 2,216,183 1,852,729 1,277,999 1.196 1.734 16.4% 42.3% 2,216,183 2,216,183 363,454 363,454 363,454 0 0
2008 12 2,326,992 2,326,992 1,568,393 729,124 1.484 3.191 32.6% 68.7% 2,326,992 2,326,992 758,599 758,599 758,599 -0 0

Total 18,866,839 18,866,839 17,472,204 15,270,788 18,866,839 18,866,839 1,394,634 1,394,634 1,394,634 0 0

Changing Product Mix

1999 120 1,500,000 1,500,000 1,500,000 1,500,000 1.000 1.000 0.0% 0.0% 1,500,000 1,500,000 0 0 0 0 0
2000 108 1,575,000 1,575,000 1,575,000 1,566,600 1.000 1.005 0.0% 0.5% 1,575,000 1,575,000 0 0 0 0 0
2001 96 1,653,750 1,653,750 1,653,750 1,628,393 1.000 1.016 0.0% 1.5% 1,653,750 1,653,750 0 0 0 0 0
2002 84 1,736,438 1,736,438 1,736,438 1,700,551 1.000 1.021 0.0% 2.1% 1,736,438 1,736,438 0 0 0 0 0
2003 72 1,823,259 1,823,259 1,814,751 1,757,622 1.005 1.037 0.5% 3.6% 1,823,259 1,823,259 8,509 8,509 8,509 0 0
2004 60 1,914,422 1,914,422 1,885,068 1,786,794 1.016 1.071 1.5% 6.7% 1,914,422 1,914,422 29,354 29,354 29,354 0 0
2005 48 2,262,528 2,251,361 2,193,545 1,951,435 1.032 1.154 3.1% 13.3% 2,262,929 2,251,616 69,384 58,071 71,855 2,470 13,784
2006 36 2,692,025 2,656,353 2,471,446 1,983,482 1.090 1.336 8.3% 25.2% 2,693,594 2,652,159 222,148 180,713 239,057 16,909 58,344
2007 24 3,216,658 3,142,865 2,680,487 1,766,164 1.200 1.750 16.7% 42.9% 3,217,588 3,113,616 537,101 433,129 596,924 59,823 163,795
2008 12 3,860,964 3,804,378 2,556,695 1,097,644 1.503 3.273 33.5% 69.4% 3,848,776 3,739,784 1,292,081 1,183,089 1,445,385 153,304 262,296

Total 22,235,045 22,057,826 20,067,179 16,738,684 22,225,757 21,960,044 2,158,578 1,892,866 2,391,084 232,507 498,219

Column Notes:
(2) Age of accident year at December 31, 2008.
(3) and (4) Developed in Exhibit IV.
(5) and (6) From last diagonal of reported and paid claim triangles in Chapter 7, Exhibit IV, Sheets 2 through 5.
(7) and (8) CDF based on 5-year simple average age-to-age factors presented in Chapter 7, Exhibit IV, Sheets 2 through 5.
(9) = [1.00 - (1.00 / (7))].
(10) = [1.00 - (1.00 / (8))].
(11) = [((3) x (9)) + (5)].
(12) = [((4) x (10)) + (6)].
(13) = [(11) - (5)].
(14) = [(12) - (5)].
(15) Developed in Chapter 7, Exhibit IV, Sheet 1.
(16) = [(15) - (13)].
(17) = [(15) - (14)].

173

Exhibits Combined.xls 9_6 04/03/2009 - 2:58 PM

Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

CHAPTER 10 – CAPE COD TECHNIQUE

The Cape Cod method, also known as the Stanard-Buhlmann method, is similar to the
Bornhuetter-Ferguson technique. As in the Bornhuetter-Ferguson technique, the Cape Cod
method splits ultimate claims into two components: actual reported (or paid) and expected
unreported (or unpaid). As an accident year (or other time interval) matures, the actual reported
claims replace the expected unreported claims and the initial expected claims assumption
becomes gradually less important. The primary difference between the two methods is the
derivation of the expected claim ratio. In the Cape Cod technique, the expected claim ratio is
obtained from the reported claims experience instead of an independent and often judgmental
selection as in the Bornhuetter-Ferguson technique.

Key Assumptions

The key assumption of the Cape Cod method is that unreported claims will develop based on
expected claims, which are derived using reported (or paid) claims and earned premium. Both the
Cape Cod and Bornhuetter-Ferguson methods differ from the development method where the
primary assumption is that unreported claims will develop based on reported claims to date (not
expected claims).

Common Uses of the Cape Cod Technique

Reinsurers are among the most frequent users of the Cape Cod technique. Actuaries generally use
the Cape Cod method in a reported claims application, but they can also use it with paid claims.
The technique is appropriate for all lines of insurance including short-tail lines and long-tail lines.
Similar to the development and Bornhuetter-Ferguson methods, actuaries using the Cape Cod
method can organize data in a variety of different time intervals:

Accident year
Policy year
Underwriting year
Report year
Fiscal year

Actuaries can also apply this technique with monthly, quarterly, or semiannual data.

Mechanics of the Cape Cod Technique

Similar to the Bornhuetter-Ferguson technique, the Cape Cod method is a blend of two other
methods: the claim development method and the expected claims method. We restate below the
formula of the reported Bornhuetter-Ferguson method, which is the same as the Cape Cod
method:

Ultimate Claims = Actual Reported Claims + Expected Unreported Claims

174
Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

Again, the major difference between the Cape Cod technique and the Bornhuetter-Ferguson is the
source of the expected claims. In “Reinsurance” Patrik states:

The key innovation of the SB (Stanard-Buhlmann) Method is that the ultimate

expected loss ratio for all years combined is estimated from the overall reported
claims experience, instead of being selected judgmentally, as in the BF
(Bornhuetter-Ferguson) Method. A problem with the SB Method is that the IBNR
by year is highly dependent upon the rate level adjusted premium by year. The
user must adjust each year’s premium to reflect the rate level cycle on a relative
basis. But this is also a problem with the BF Method.58

In our step-by step example of the Cape Cod method, we use the cumulative claim development
patterns presented in Chapter 7. We begin in Exhibit I, Sheet 1, with the development of the
estimated claim ratio. In our U.S. Industry Auto example, we do not have details of historical rate
level changes. Thus, in both the Bornhuetter-Ferguson method and the Cape Cod method, we rely
on unadjusted earned premium data.

In Column (2) of Exhibit I, Sheet 1, we summarize the unadjusted earned premiums by year.
Column (3) contains the age of each accident year as of the latest valuation date, December 31,
2007. The reported claims in Column (4) are the latest diagonal in the reported claim
development triangle presented in Chapter 7. We also derive the claim development factor to
ultimate, Column (5), in Chapter 7. In Column (6), we show the reporting pattern. The percentage
reported is equal to the inverse of the cumulative reported claim development factor.

A new concept of the Cape Cod method is the “used-up premium.” The used-up premium is the
denominator in our determination of the expected claim ratio. This allocation of premium
represents the premium corresponding to the claims that are expected to be reported through the
valuation date. The used-up premium in Column (7) is equal to the earned premium in Column
(2) multiplied by the percentage of claims reported in Column (6). Reinsurers often use ultimate
premiums in Column (2) instead of earned premium. In Column (8), we calculate estimated claim
ratios, by accident year, by dividing actual reported claims from Column (4) by the used-up
premium in Column (7). (An alternative to the use of premium and claim ratios is exposures and
pure premiums. Instead of calculating used-up premium, the actuary could calculate used-up
exposures and calculate estimated pure premiums instead of estimated claim ratios for each year
in the experience period.)

In our U.S. Industry Auto example, we observe a change in the claim ratios for the latest accident
years when compared with the earliest years (i.e., 1998 through 2002). The average estimated
claim ratio for accident years 1998 through 2002 is 75.2%. For this period of time, the claim
ratios vary from a low of 69.6% to a high of 79.7%. We contrast this with the more recent years’
experience, which has an average claim ratio of 64.8%. For each year, 2003 through 2007, the
estimated claim ratio is less than 67.5%. In the expected claims technique and the Bornhuetter-
Ferguson technique, we rely on different claim ratios for the earlier years and the latest years
in the experience period to best reflect our expectation of expected claims for each year. In
contrast, the Cape Cod method requires the use of the weighted average claim ratio from all
years. Thus, one can distinguish the mechanical approach of developing expected claims in the
Cape Cod method from the Bornhuetter-Ferguson method in which actuarial judgment plays an
important role in the development of the a priori expected claim estimate.

58
Foundations of Casualty Actuarial Science, 2001. We refer the reader to “Reinsurance” (Chapter 7) for
Patrik’s complete development of the formulae underlying the Cape Cod technique.

175
Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

Unpaid Claim Estimate Based on Cape Cod Technique

We follow a similar procedure for determining the unpaid claim estimate based on the Cape Cod
technique as presented in the prior chapters. Estimated IBNR is equal to projected ultimate claims
less reported claims and the total unpaid claim estimate is equal to the difference between
projected ultimate claims and paid claims.

Exhibit I, Sheet 3 displays the calculations for the estimated unpaid claims of U.S. Industry Auto.
Columns (2) and (3) contain reported and paid claims data as of December 31, 2007. We
summarize the projected ultimate claims from Exhibit I, Sheet 2 in Column (4). Case outstanding,
which are equal to the difference between reported claims and paid claims as of December 31,
2007, are presented in Column (5). Estimated IBNR is equal to projected ultimate claims minus
reported claims. We calculate the estimated IBNR based on the Cape Cod technique in Column
(6). The total unpaid claim estimate (Column (7)) is equal to the sum of case outstanding and
estimated IBNR.

When the Cape Cod Technique Works and When it Does Not

Similar comments apply to the Cape Cod method as to the Bornhuetter-Ferguson technique. The
only difference between the two methods is the derivation of the expected claims. Thus, an
advantage of the Cape Cod method, when compared to the development technique, is that it may
not be distorted by random fluctuations early in the development of an accident year (or other
time interval). A determining factor influencing the fluctuations, in either the Bornhuetter-
Ferguson or Cape Cod methods, is the extent to which actual claims for the most recent years
affect the derivation of expected claims for such years.

The Cape Cod method is not necessarily as appropriate as the Bornhuetter-Ferguson technique if
the data is extremely thin or volatile or both. Since the expected claims are based on reported
claims to date, there must be a sufficient volume of credible reported claims in order to derive a
reliable expected claims estimate.

It is worthwhile to note that in an ideal situation, the actuary would have the history of rate level
changes and would be able to adjust historical premiums to an on-level basis for both the Cape
Cod and Bornhuetter-Ferguson projections. The actuary would also adjust claims for trend,
benefit-level changes, and other similar factors. From a theoretical perspective, these methods
require such adjustment. From a practical perspective, however, such information is often
unavailable. In these situations, many actuaries continue to use both the Bornhuetter-Ferguson
and Cape Cod methods for the purpose of developing the unpaid claim estimate without the
adjustment of premiums or claims. Under such circumstances, it would be prudent for the
actuary, when evaluating the results of various techniques and selecting final ultimate claims
values, to take into consideration where simplifying assumptions (such as not adjusting premium
for rate level changes) were required.

XYZ Insurer

In Exhibit II, Sheets 1 through 3, we use the Cape Cod technique for XYZ Insurer. There are
weaknesses in this projection technique due to the uncertainty in the selected development
patterns for reported claims. Due to the numerous changes the insurer has faced, we are uncertain

176
Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

as to the applicability of historical claim development patterns. Since the Cape Cod method uses
claim development patterns to calculate the used-up premium, which is a critical component in
the expected claim ratio determination, this method may not be appropriate for this example.
(Similar to the Bornhuetter-Ferguson method, we limit the reported cumulative claim
development factors to a minimum of 1.00 for the Cape Cod technique.)

We have detailed rate change information for XYZ Insurer as well as information regarding the
effect of legal reform on the insurance product. We incorporate this information into the Cape
Cod projection method presented in Exhibit II. The first adjustment is to restate earned premium
for each accident year as if it were at the 2008 rate level. These calculations are contained within
Columns (2) through (4) of Exhibit II, Sheet 1. In Columns (6) through (9), we adjust the current
reported claims for the influences of inflation (through claims trend factors) and tort reform. Once
we have on-level earned premium and adjusted claims, we proceed to calculate estimated claim
ratios as described in the previous example for U.S. Industry Auto. We divide the adjusted claims
by used-up, on-level premium to derive the claim ratios shown in Column (13). We use the label
“Estimated Adjusted Claim Ratios” to indicate that the reported claims are adjusted for inflation
and tort reform. We rely on the claim ratio for all years combined, 70.8%, from Column (13)
(also shown in Column (14) for each year) as our starting point for developing estimated
unadjusted claim ratios in Column (15). These claim ratios, which are adjusted back to the rate
level, inflationary level, and tort environment for each accident year, become our starting point
for projecting expected claims in Exhibit II, Sheet 2.

We follow the same format as the example for U.S. Industry Auto in Exhibit II, Sheets 2 and 3.
We compare the results of the Cape Cod method with the Bornhuetter-Ferguson method, the
expected claims method, and the claim development method in Exhibit II, Sheet 4 (projected
ultimate claims) and in Exhibit II, Sheet 5 (estimated IBNR).

Influence of a Changing Environment on the Cape Cod Method59

In prior chapters, we discuss the performance of each of the estimation techniques during times of
change. We continue these examples using the Cape Cod method.

Scenario 1 – U.S. PP Auto Steady-State

We see in Chapters 7 through 9 that the development technique, expected claims technique, and
Bornhuetter-Ferguson techniques all generate an accurate IBNR value in a steady-state
environment. It is not surprising to find that the Cape Cod method also generates the actual IBNR
in a steady-state environment. The top section of Exhibit III, Sheets 1 and 3, contains detailed
calculations for the Cape Cod method.

59
We present the following examples to demonstrate the effect of not changing assumption on the resulting
projections of ultimate claims and the estimate of unpaid claims. We recognize that the examples are not
necessarily representative of real-life applications of the Cape Cod method since we assume that there are
no adjustments in expected claims in anticipation of the events that caused higher claim ratios or changes in
business mix. Most insurers have a feel for whether a market is getting softer or harder, so they would have
a sense as to the direction to adjust the expected claims, if not the absolute amount of adjustment.

177
Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

Scenario 2 – U.S. PP Auto Increasing Claim Ratios

Recall that the weakness of the expected claims method, which is the lack of responsiveness to
actual emerging claims, is also a weakness of the Bornhuetter-Ferguson method. The Cape Cod
method, which derives the expected claim ratio based on reported claims through the valuation
date, does not have this same weakness. In Exhibit III, Sheet 1, we see that the estimated claim
ratios in Column (8) respond to the changing environment in claims experience. The total all
years combined estimated claim ratio is 80.7% for this scenario; this compares to the 70%
expected claim ratio for the steady-state.

In the Bornhuetter-Ferguson reported claim projection, there is no change in the estimated IBNR
of $438,638 between Scenario 1 and Scenario 2 since the expected claim ratio does not change.
However, using the Cape Cod method, the estimated IBNR is $505,828 for Scenario 2. While this
value is still short of the actual IBNR requirements of $601,984, the Cape Cod technique is more
responsive than the Bornhuetter-Ferguson method when the claim ratios are increasing.

Scenario 3 – U.S. PP Auto Increasing Case Outstanding Strength

We present the calculations for Scenario 3 in the top section of Exhibit III, Sheets 2 and 4. In this
example, we see that the Cape Cod method results in an estimated IBNR that overstates the actual
IBNR by an even greater amount than the reported Bornhuetter-Ferguson technique. In the
previous chapters, we discuss how the increase in case outstanding strength leads to an increase in
the cumulative claim development factors. Whereas the expected claims for Scenario 3 of the
Bornhuetter-Ferguson method remain unchanged, the expected claims increase using the Cape
Cod method because the method reflects the higher level of reported claims. The projected
ultimate claims are increasing for the Cape Cod method under Scenario 3 due to both increasing
expected claims and higher claim development factors to ultimate.

Scenario 4 – U.S. PP Auto Increasing Claim Ratios and Case Outstanding Strength

In times of increasing claim ratios and increasing case outstanding strength, the Cape Cod method
can overstate the actual IBNR. In this example, the method responds effectively to the change in
claim ratios, however it overreacts to the change in case outstanding adequacy. In our example, the
Cape Cod method significantly overstates the actual IBNR needed, indicating that the effect of
increasing case outstanding strength exceeds the influence of increasing claim ratios. The estimated
claim ratios are driven higher than their true values by the combined effects of both increasing
claims and greater adequacy in case outstanding. We present the detailed calculations for Scenario
4 in the bottom section of Exhibit III, Sheets 2 and 4.

U.S. Auto Steady-State (No Change in Product Mix)

In the last two examples, we present projections for a combined portfolio of private passenger and
commercial automobile. In the top section of Exhibit IV, Sheets 1 and 2, we summarize the
calculations for the steady-state environment where there is no change in product mix. Similar to
our projections using the development and expected claims techniques, we demonstrate in Exhibit
IV, Sheet 2 that the Cape Cod technique generates the correct IBNR requirement when there is no
change in the product mix.

178
Estimating Unpaid Claims Using Basic Techniques
Chapter 10 - Cape Cod Technique

U.S. Auto Changing Product Mix

In the final example, we assume that the volume of commercial automobile insurance is
increasing at a greater rate than private passenger automobile insurance. In the bottom section of
Exhibit IV, Sheet 2, we observe that the Cape Cod method produces estimated IBNR that is lower
than the actual IBNR. Even though reported claims are increasing in this scenario when compared
to the prior scenario, there are also changes in the reporting pattern. Thus, the Cape Cod method
is not responding appropriately to the changing product mix. Detailed calculations are contained
within the bottom section of Exhibit IV, Sheets 1 and 2.

179
 Chapter 10 - Cape Cod Technique Exhibit I
U.S. Industry Auto Sheet 1
Development of Expected Claim Ratio

Age of Reported Reported % of Estimated

Accident Earned Accident Year Claims at CDF to Ultimate Used Up Claim
Year Premium at 12/31/07 12/31/07 Ultimate Reported Premium Ratios
(1) (2) (3) (4) (5) (6) (7) (8)

1998 68,574,209 120 47,742,304 1.000 100.0% 68,574,209 69.6%

1999 68,544,981 108 51,185,767 1.000 100.0% 68,544,981 74.7%
2000 68,907,977 96 54,837,929 1.001 99.9% 68,839,138 79.7%
2001 72,544,955 84 56,299,562 1.003 99.7% 72,327,971 77.8%
2002 79,228,887 72 58,592,712 1.006 99.4% 78,756,349 74.4%
2003 86,643,542 60 57,565,344 1.011 98.9% 85,700,833 67.2%
2004 91,763,523 48 56,976,657 1.023 97.8% 89,700,413 63.5%
2005 94,115,312 36 56,786,410 1.051 95.1% 89,548,346 63.4%
2006 95,272,279 24 54,641,339 1.110 90.1% 85,830,882 63.7%
2007 95,176,240 12 48,853,563 1.292 77.4% 73,665,820 66.3%

Total 820,771,905 543,481,587 781,488,943 69.5%

Column and Line Notes:

(2) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(3) Age of accident year in (1) at December 31, 2007.
(4) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(5) Developed in Chapter 7, Exhibit I, Sheet 1.
(6) = [1.00 / (5)].
(7) = [(2) x (6)].
(8) = [(4) / (7)].

180

Exhibits Combined.xls 10_1_1 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit I
U.S. Industry Auto Sheet 2
Projection of Ultimate Claims Using Reported Claims ($000)

Expected Estimated Reported Expected Reported Projected

Accident Earned Claim Expected CDF to Percentage Unreported Claims at Ultimate
Year Premium Ratio Claims Ultimate Unreported Claims 12/31/07 Claims
(1) (2) (3) (4) (5) (6) (7) (8) (9)

1998 68,574,209 69.5% 47,689,504 1.000 0.0% 0 47,742,304 47,742,304

1999 68,544,981 69.5% 47,669,177 1.000 0.0% 0 51,185,767 51,185,767
2000 68,907,977 69.5% 47,921,621 1.001 0.1% 47,874 54,837,929 54,885,803
2001 72,544,955 69.5% 50,450,934 1.003 0.3% 150,900 56,299,562 56,450,462
2002 79,228,887 69.5% 55,099,233 1.006 0.6% 328,624 58,592,712 58,921,336
2003 86,643,542 69.5% 60,255,708 1.011 1.1% 655,601 57,565,344 58,220,945
2004 91,763,523 69.5% 63,816,367 1.023 2.2% 1,434,777 56,976,657 58,411,434
2005 94,115,312 69.5% 65,451,904 1.051 4.9% 3,176,068 56,786,410 59,962,478
2006 95,272,279 69.5% 66,256,509 1.110 9.9% 6,565,960 54,641,339 61,207,299
2007 95,176,240 69.5% 66,189,720 1.292 22.6% 14,959,286 48,853,563 63,812,849

Total 820,771,905 570,800,677 27,319,090 543,481,587 570,800,677

Column Notes:
(2) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(3) Based on total weighted estimated claim ratios developed in Exhibit I, Sheet 1.
(4) = [(2) x (3)].
(5) Developed in Chapter 7, Exhibit I, Sheet 1.
(6) = [1.00 - (1.00 / (5))].
(7) = [(4) x (6)].
(8) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(9) = [(7) + (8)].

181

Exhibits Combined.xls 10_1_2 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit I
U.S. Industry Auto Sheet 3
Development of Unpaid Claim Estimate ($000)

Projected Case Unpaid Claim Estimate

Accident Claims at 12/31/07 Ultimate Outstanding Based on Cape Cod Method
Year Reported Paid Claims at 12/31/07 IBNR Total
(1) (2) (3) (4) (5) (6) (7)

1998 47,742,304 47,644,187 47,742,304 98,117 0 98,117

1999 51,185,767 51,000,534 51,185,767 185,233 0 185,233
2000 54,837,929 54,533,225 54,885,803 304,704 47,874 352,578
2001 56,299,562 55,878,421 56,450,462 421,141 150,900 572,041
2002 58,592,712 57,807,215 58,921,336 785,497 328,624 1,114,121
2003 57,565,344 55,930,654 58,220,945 1,634,690 655,601 2,290,291
2004 56,976,657 53,774,672 58,411,434 3,201,985 1,434,777 4,636,762
2005 56,786,410 50,644,994 59,962,478 6,141,416 3,176,068 9,317,484
2006 54,641,339 43,606,497 61,207,299 11,034,842 6,565,960 17,600,802
2007 48,853,563 27,229,969 63,812,849 21,623,594 14,959,286 36,582,880

Total 543,481,587 498,050,368 570,800,677 45,431,219 27,319,090 72,750,309

Column Notes:
(2) and (3) Based on Best's Aggregates & Averages U.S. private passenger automobile experience.
(4) Developed in Exhibit I, Sheet 2.
(5) = [(2) - (3)].
(6) = [(4) - (2)].
(7) = [(5) + (6)].

182

Exhibits Combined.xls 10_1_3 04/03/2009 - 2:58 PM

Chapter 10 - Cape Cod Technique Exhibit II
XYZ Insurer - Auto BI Sheet 1
Development of Expected Claim Ratio

On-Level Age of Reported Pure Tort Adjusted Reported % of Used Up Claim Ratios
Accident Earned On-Level Earned Accident Year Claims at Premium Reform Claims at CDF to Ultimate On-Level Estimated Selected Estimated
Year Premium Adjustment Premium at 12/31/08 12/31/08 Trend Factors 12/31/08 Ultimate Reported Premium Adjusted Adjusted Unadjusted
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

1998 20,000 0.989 19,783 132 15,822 1.400 0.670 14,846 1.000 100.0% 19,783 75.0% 70.8% 74.6%
1999 31,500 0.970 30,548 120 25,107 1.354 0.670 22,777 1.000 100.0% 30,548 74.6% 70.8% 75.7%
2000 45,000 0.951 42,784 108 37,246 1.309 0.670 32,671 1.000 100.0% 42,784 76.4% 70.8% 76.7%
2001 50,000 0.932 46,606 96 38,798 1.266 0.670 32,905 1.000 100.0% 46,606 70.6% 70.8% 77.8%
2002 61,183 0.914 55,911 84 48,169 1.224 0.670 39,500 1.003 99.7% 55,744 70.9% 70.8% 78.9%
2003 69,175 0.870 60,204 72 44,373 1.183 0.670 35,182 1.013 98.7% 59,432 59.2% 70.8% 77.7%
2004 99,322 0.810 80,411 60 70,288 1.144 0.670 53,884 1.064 94.0% 75,574 71.3% 70.8% 74.7%
2005 138,151 0.704 97,258 48 70,655 1.106 0.670 52,371 1.085 92.2% 89,639 58.4% 70.8% 67.2%
2006 107,578 0.640 68,850 36 48,804 1.070 0.750 39,153 1.196 83.6% 57,567 68.0% 70.8% 56.5%
2007 62,438 0.800 49,950 24 31,732 1.034 1.000 32,819 1.512 66.1% 33,036 99.3% 70.8% 54.7%
2008 47,797 1.000 47,797 12 18,632 1.000 1.000 18,632 2.551 39.2% 18,737 99.4% 70.8% 70.8%

Total 732,144 600,103 449,626 374,739 529,449 70.8%

Column and Line Notes:

(2) Based on data from insurer.
(3) For 2002 and after, based on Chapter 8, Exhibit III, Sheet 2. For 1998-2001, assume a 2% rate change per annum.
(4) = [(2) x (3)].
(5) Age of accident year in (1) at December 31, 2008.
(6) Based on data from insurer.
(7) Assume an annual pure premium trend rate of 3.425%.
(8) Based on independent analysis of tort reform.
(9) = [(6) x (7) x (8)].
(10) Developed in Chapter 7, Exhibit II, Sheet 1, in which the CDF are limited to a minimum of 1.00.
(11) = [1.00 / (10)].
(12) = [(4) x (11)].
(13) = [(9) / (12)].
(14) = [Total in (13)].
(15) = [(14) x (3) / (7) / (8)].

183

Exhibits Combined.xls 10_2_1 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit II
XYZ Insurer - Auto BI Sheet 2
Projection of Ultimate Claims Using Reported Claims ($000)

Expected Estimated Reported Expected Reported Projected

Accident Earned Claim Expected CDF to Percentage Unreported Claims at Ultimate
Year Premium Ratio Claims Ultimate Unreported Claims 12/31/08 Claims
(1) (2) (3) (4) (5) (6) (7) (8) (9)

1998 20,000 74.6% 14,924 1.000 0.0% 0 15,822 15,822

1999 31,500 75.7% 23,833 1.000 0.0% 0 25,107 25,107
2000 45,000 76.7% 34,523 1.000 0.0% 0 37,246 37,246
2001 50,000 77.8% 38,895 1.000 0.0% 0 38,798 38,798
2002 61,183 78.9% 48,259 1.003 0.3% 144 48,169 48,313
2003 69,175 77.7% 53,744 1.013 1.3% 690 44,373 45,062
2004 99,322 74.7% 74,241 1.064 6.0% 4,466 70,288 74,754
2005 138,151 67.2% 92,871 1.085 7.8% 7,276 70,655 77,931
2006 107,578 56.5% 60,743 1.196 16.4% 9,955 48,804 58,759
2007 62,438 54.7% 34,184 1.512 33.9% 11,575 31,732 43,307
2008 47,797 70.8% 33,830 2.551 60.8% 20,569 18,632 39,201

Total 732,144 510,046 54,674 449,626 504,300

Column Notes:
(2) Based on data from XYZ Insurer.
(3) Selected based on estimated claim ratios developed in Exhibit II, Sheet 1.
(4) = [(2) x (3)].
(5) Developed in Chapter 7, Exhibit II, Sheet 1, limited to a minimum of 1.00.
(6) = [1.00 - (1.00 / (5))].
(7) = [(4) x (6)].
(8) Based on data from insurer.
(9) = [(7) + (8)].

184

Exhibits Combined.xls 10_2_2 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit II
XYZ Insurer - Auto BI Sheet 3
Development of Unpaid Claim Estimate ($000)

Projected Case Unpaid Claim Estimate

Accident Claims at 12/31/08 Ultimate Outstanding Based on Cape Cod Method
Year Reported Paid Claims at 12/31/08 IBNR Total
(1) (2) (3) (4) (5) (6) (7)

1998 15,822 15,822 15,822 0 0 0

1999 25,107 24,817 25,107 290 0 290
2000 37,246 36,782 37,246 465 0 465
2001 38,798 38,519 38,798 278 0 278
2002 48,169 44,437 48,313 3,731 144 3,876
2003 44,373 39,320 45,062 5,052 690 5,742
2004 70,288 52,811 74,754 17,477 4,466 21,943
2005 70,655 40,026 77,931 30,629 7,276 37,904
2006 48,804 22,819 58,759 25,985 9,955 35,940
2007 31,732 11,865 43,307 19,867 11,575 31,442
2008 18,632 3,409 39,201 15,223 20,569 35,792

Total 449,626 330,629 504,300 118,997 54,674 173,671

Column Notes:
(2) and (3) Based on data from XYZ Insurer.
(4) Developed in Exhibit II, Sheet 2.
(5) = [(2) - (3)].
(6) = [(4) - (2)].
(7) = [(4) - (3)].

185

Exhibits Combined.xls 10_2_3 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit II
XYZ Insurer - Auto BI Sheet 4
Summary of Ultimate Claims ($000)

Projected Ultimate Claims

Accident Claims at 12/31/08 Development Method Expected B-F Method Cape
Year Reported Paid Reported Paid Claims Reported Paid Cod
(1) (2) (3) (4) (5) (6) (7) (8) (9)

1998 15,822 15,822 15,822 15,980 15,660 15,822 15,977 15,822

1999 25,107 24,817 25,082 25,164 24,665 25,107 25,158 25,107
2000 37,246 36,782 36,948 37,922 35,235 37,246 37,841 37,246
2001 38,798 38,519 38,487 40,600 39,150 38,798 40,525 38,798
2002 48,169 44,437 48,313 49,592 47,906 48,312 49,417 48,313
2003 44,373 39,320 44,950 49,858 54,164 45,068 50,768 45,062
2004 70,288 52,811 74,787 80,537 86,509 75,492 82,593 74,754
2005 70,655 40,026 76,661 80,333 108,172 79,129 94,301 77,931
2006 48,804 22,819 58,370 72,108 70,786 60,404 71,205 58,759
2007 31,732 11,865 47,979 77,941 39,835 45,221 45,636 43,307
2008 18,632 3,409 47,530 74,995 39,433 42,607 41,049 39,201

Total 449,626 330,629 514,929 605,030 561,516 513,207 554,471 504,300

Column Notes:
(2) and (3) Based on data from XYZ Insurer.
(4) and (5) Developed in Chapter 7, Exhibit II, Sheet 3.
(6) Developed in Chapter 8, Exhibit III, Sheet 1.
(7) and (8) Developed in Chapter 9, Exhibit II, Sheet 1.
(9) Developed in Exhibit II, Sheet 2.

186

Exhibits Combined.xls 10_2_4 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit II
XYZ Insurer - Auto BI Sheet 5
Summary of IBNR ($000)

Case Estimated IBNR

Accident Outstanding Development Method Expected B-F Method Cape
Year at 12/31/08 Reported Paid Claims Reported Paid Cod
(1) (2) (3) (4) (5) (6) (7) (8)

1998 0 0 158 - 162 0 155 0

1999 290 - 25 58 - 442 0 51 0
2000 465 - 298 676 - 2,011 0 595 0
2001 278 - 310 1,802 352 0 1,728 0
2002 3,731 145 1,423 - 262 143 1,248 144
2003 5,052 577 5,485 9,791 695 6,396 690
2004 17,477 4,498 10,249 16,221 5,204 12,305 4,466
2005 30,629 6,006 9,678 37,517 8,474 23,646 7,276
2006 25,985 9,566 23,304 21,982 11,600 22,401 9,955
2007 19,867 16,247 46,209 8,103 13,489 13,904 11,575
2008 15,223 28,898 56,363 20,801 23,975 22,417 20,569

Total 118,997 65,303 155,405 111,890 63,581 104,845 54,674

Column Notes:
(2) Based on data from XYZ Insurer.
(3) and (4) Estimated in Chapter 7, Exhibit II, Sheet 4.
(5) Estimated in Chapter 8, Exhibit III, Sheet 3.
(6) and (7) Estimated in Chapter 9, Exhibit II, Sheet 2.
(8) Estimated in Exhibit II, Sheet 3.

187

Exhibits Combined.xls 10_2_5 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit III
Impact of Changing Conditions Sheet 1
Scenarios 1 and 2 - Development of Expected Claim Ratio

Age of Reported Reported % of Estimated

Accident Earned Accident Year Claims at CDF to Ultimate Used Up Claim
Year Premium at 12/31/08 12/31/08 Ultimate Reported Premium Ratios
(1) (2) (3) (4) (5) (6) (7) (8)

Steady-State
1999 1,000,000 120 700,000 1.000 100.0% 1,000,000 70.0%
2000 1,050,000 108 735,000 1.000 100.0% 1,050,000 70.0%
2001 1,102,500 96 771,750 1.000 100.0% 1,102,500 70.0%
2002 1,157,625 84 810,338 1.000 100.0% 1,157,625 70.0%
2003 1,215,506 72 842,346 1.010 99.0% 1,203,351 70.0%
2004 1,276,282 60 884,463 1.010 99.0% 1,263,519 70.0%
2005 1,340,096 48 919,306 1.020 98.0% 1,313,294 70.0%
2006 1,407,100 36 935,722 1.053 95.0% 1,336,745 70.0%
2007 1,477,455 24 930,797 1.111 90.0% 1,329,710 70.0%
2008 1,551,328 12 836,166 1.299 77.0% 1,194,523 70.0%

Total 12,577,893 8,365,887 11,951,267 70.0%

Increasing Claim Ratios

1999 1,000,000 120 700,000 1.000 100.0% 1,000,000 70.0%
2000 1,050,000 108 735,000 1.000 100.0% 1,050,000 70.0%
2001 1,102,500 96 771,750 1.000 100.0% 1,102,500 70.0%
2002 1,157,625 84 810,338 1.000 100.0% 1,157,625 70.0%
2003 1,215,506 72 842,346 1.010 99.0% 1,203,351 70.0%
2004 1,276,282 60 1,010,815 1.010 99.0% 1,263,519 80.0%
2005 1,340,096 48 1,116,300 1.020 98.0% 1,313,294 85.0%
2006 1,407,100 36 1,203,071 1.053 95.0% 1,336,745 90.0%
2007 1,477,455 24 1,263,224 1.111 90.0% 1,329,710 95.0%
2008 1,551,328 12 1,194,523 1.299 77.0% 1,194,523 100.0%

Total 12,577,893 9,647,366 11,951,267 80.7%

Column Notes:
(2) Assume $1,000,000 for first year in experience period (1999) and 5% annual increase thereafter.
(3) Age of accident year at December 31, 2008.
(4) From last diagonal of reported claim triangles in Chapter 7, Exhibit III, Sheets 2 and 4.
(5) Developed in Chapter 7, Exhibit III, Sheets 2 and 4.
(6) = [1.00 / (5)].
(7) = [(2) x (6)].
(8) = [(4) / (7)].
188

Exhibits Combined.xls 10_3_1 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit III
Impact of Changing Conditions Sheet 2
Scenarios 3 and 4 - Development of Expected Claim Ratio

Age of Reported Reported % of Estimated

Accident Earned Accident Year Claims at CDF to Ultimate Used Up Claim
Year Premium at 12/31/08 12/31/08 Ultimate Reported Premium Ratios
(1) (2) (3) (4) (5) (6) (7) (8)

Increasing Case Outstanding Strength

1999 1,000,000 120 700,000 1.000 100.0% 1,000,000 70.0%
2000 1,050,000 108 735,000 1.000 100.0% 1,050,000 70.0%
2001 1,102,500 96 771,750 1.000 100.0% 1,102,500 70.0%
2002 1,157,625 84 810,338 1.000 100.0% 1,157,625 70.0%
2003 1,215,506 72 842,346 1.010 99.0% 1,203,351 70.0%
2004 1,276,282 60 884,463 1.010 99.0% 1,263,519 70.0%
2005 1,340,096 48 933,377 1.020 98.1% 1,314,355 71.0%
2006 1,407,100 36 962,808 1.055 94.8% 1,334,350 72.2%
2007 1,477,455 24 979,922 1.119 89.4% 1,320,694 74.2%
2008 1,551,328 12 931,185 1.318 75.9% 1,176,756 79.1%

Total 12,577,893 8,551,189 11,923,151 71.7%

Increasing Claim Ratios and Case Outstanding Strength

1999 1,000,000 120 700,000 1.000 100.0% 1,000,000 70.0%
2000 1,050,000 108 735,000 1.000 100.0% 1,050,000 70.0%
2001 1,102,500 96 771,750 1.000 100.0% 1,102,500 70.0%
2002 1,157,625 84 810,338 1.000 100.0% 1,157,625 70.0%
2003 1,215,506 72 842,346 1.010 99.0% 1,203,351 70.0%
2004 1,276,282 60 1,010,815 1.010 99.0% 1,263,519 80.0%
2005 1,340,096 48 1,133,386 1.019 98.1% 1,314,470 86.2%
2006 1,407,100 36 1,237,897 1.055 94.8% 1,334,094 92.8%
2007 1,477,455 24 1,329,895 1.120 89.3% 1,319,695 100.8%
2008 1,551,328 12 1,330,264 1.320 75.7% 1,174,877 113.2%

Total 12,577,893 9,901,689 11,920,130 83.1%

Column Notes:
(2) Assume $1,000,000 for first year in experience period (1999) and 5% annual increase thereafter.
(3) Age of accident year at December 31, 2008.
(4) From last diagonal of reported claim triangles in Chapter 7, Exhibit III, Sheets 6 and 8.
(5) Developed in Chapter 7, Exhibit III, Sheets 6 and 8.
(6) = [1.00 / (5)].
(7) = [(2) x (6)].
(8) = [(4) / (7)].
189

Exhibits Combined.xls 10_3_2 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit III
Impact of Changing Conditions Sheet 3
U.S. PP Auto - Development of Unpaid Claim Estimate

Expected Estimated Reported Expected Reported Projected Difference

Accident Earned Claim Expected CDF to Percentage Unreported Claims at Ultimate Estimated Actual from
Year Premium Ratio Claims Ultimate Unreported Claims 12/31/08 Claims IBNR IBNR Actual IBNR
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Steady-State
1999 1,000,000 70.0% 700,000 1.000 0.0% 0 700,000 700,000 0 0 0
2000 1,050,000 70.0% 735,000 1.000 0.0% 0 735,000 735,000 0 0 0
2001 1,102,500 70.0% 771,750 1.000 0.0% 0 771,750 771,750 0 0 0
2002 1,157,625 70.0% 810,338 1.000 0.0% 0 810,338 810,338 0 0 0
2003 1,215,506 70.0% 850,854 1.010 1.0% 8,509 842,346 850,854 8,509 8,509 0
2004 1,276,282 70.0% 893,397 1.010 1.0% 8,934 884,463 893,397 8,934 8,934 0
2005 1,340,096 70.0% 938,067 1.020 2.0% 18,761 919,306 938,067 18,761 18,761 0
2006 1,407,100 70.0% 984,970 1.053 5.0% 49,249 935,722 984,970 49,249 49,249 0
2007 1,477,455 70.0% 1,034,219 1.111 10.0% 103,422 930,797 1,034,219 103,422 103,422 0
2008 1,551,328 70.0% 1,085,930 1.299 23.0% 249,764 836,166 1,085,930 249,764 249,764 0

Total 12,577,893 8,804,525 438,638 8,365,887 8,804,525 438,638 438,638 0

Increasing Claim Ratios

1999 1,000,000 80.7% 807,225 1.000 0.0% 0 700,000 700,000 0 0 0
2000 1,050,000 80.7% 847,587 1.000 0.0% 0 735,000 735,000 0 0 0
2001 1,102,500 80.7% 889,966 1.000 0.0% 0 771,750 771,750 0 0 0
2002 1,157,625 80.7% 934,464 1.000 0.0% 0 810,338 810,338 0 0 0
2003 1,215,506 80.7% 981,188 1.010 1.0% 9,812 842,346 852,158 9,812 8,509 - 1,303
2004 1,276,282 80.7% 1,030,247 1.010 1.0% 10,302 1,010,815 1,021,117 10,302 10,210 - 92
2005 1,340,096 80.7% 1,081,759 1.020 2.0% 21,635 1,116,300 1,137,935 21,635 22,782 1,146
2006 1,407,100 80.7% 1,135,847 1.053 5.0% 56,792 1,203,071 1,259,863 56,792 63,320 6,527
2007 1,477,455 80.7% 1,192,640 1.111 10.0% 119,264 1,263,224 1,382,488 119,264 140,358 21,094
2008 1,551,328 80.7% 1,252,272 1.299 23.0% 288,022 1,194,523 1,482,545 288,022 356,805 68,783

Total 12,577,893 10,153,194 505,828 9,647,366 10,153,194 505,828 601,984 96,155

Column Notes:
(2) Assume $1,000,000 for first year in experience period (1999) and 5% annual increase thereafter.
(3) Selected based on estimated overall claim ratio developed in Exhibit III, Sheet 1.
(4) = [(2) x (3)].
(5) Developed in Chapter 7, Exhibit III, Sheets 2 and 4.
(6) = [1.00 - (1.00 / (5))].
(7) = [(4) x (6)].
(8) From last diagonal of reported claim triangles in Chapter 7, Exhibit III, Sheets 2 and 4.
(9) = [(7) + (8)].
(10) = [(9) - (8)].
(11) Developed in Chapter 7, Exhibit III, Sheet 1.
(12) = [(11) - (10)]. 190

Exhibits Combined.xls 10_3_3 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit III
Impact of Changing Conditions Sheet 4
U.S. PP Auto - Development of Unpaid Claim Estimate

Expected Estimated Reported Expected Reported Projected Difference

Accident Earned Claim Expected CDF to Percentage Unreported Claims at Ultimate Estimated Actual from
Year Premium Ratio Claims Ultimate Unreported Claims 12/31/08 Claims IBNR IBNR Actual IBNR
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Increasing Case Outstanding Strength

1999 1,000,000 71.7% 717,192 1.000 0.0% 0 700,000 700,000 0 0 0
2000 1,050,000 71.7% 753,052 1.000 0.0% 0 735,000 735,000 0 0 0
2001 1,102,500 71.7% 790,704 1.000 0.0% 0 771,750 771,750 0 0 0
2002 1,157,625 71.7% 830,239 1.000 0.0% 0 810,338 810,338 0 0 0
2003 1,215,506 71.7% 871,751 1.010 1.0% 8,718 842,346 851,063 8,718 8,509 - 209
2004 1,276,282 71.7% 915,339 1.010 1.0% 9,153 884,463 893,617 9,153 8,934 - 219
2005 1,340,096 71.7% 961,106 1.020 1.9% 18,461 933,377 951,838 18,461 4,690 - 13,771
2006 1,407,100 71.7% 1,009,161 1.055 5.2% 52,176 962,808 1,014,984 52,176 22,162 - 30,014
2007 1,477,455 71.7% 1,059,619 1.119 10.6% 112,428 979,922 1,092,350 112,428 54,296 - 58,132
2008 1,551,328 71.7% 1,112,600 1.318 24.1% 268,640 931,185 1,199,825 268,640 154,745 - 113,895

Total 12,577,893 9,020,764 469,576 8,551,189 9,020,764 469,576 253,336 - 216,240

Increasing Claim Ratios and Case Outstanding Strength

1999 1,000,000 83.1% 830,670 1.000 0.0% 0 700,000 700,000 0 0 0
2000 1,050,000 83.1% 872,203 1.000 0.0% 0 735,000 735,000 0 0 0
2001 1,102,500 83.1% 915,813 1.000 0.0% 0 771,750 771,750 0 0 0
2002 1,157,625 83.1% 961,604 1.000 0.0% 0 810,338 810,338 0 0 0
2003 1,215,506 83.1% 1,009,684 1.010 1.0% 10,097 842,346 852,443 10,097 8,509 - 1,588
2004 1,276,282 83.1% 1,060,168 1.010 1.0% 10,602 1,010,815 1,021,417 10,602 10,210 - 391
2005 1,340,096 83.1% 1,113,177 1.019 1.9% 21,287 1,133,386 1,154,673 21,287 5,695 - 15,591
2006 1,407,100 83.1% 1,168,835 1.055 5.2% 60,644 1,237,897 1,298,541 60,644 28,494 - 32,150
2007 1,477,455 83.1% 1,227,277 1.120 10.7% 131,047 1,329,895 1,460,941 131,047 73,688 - 57,359
2008 1,551,328 83.1% 1,288,641 1.320 24.3% 312,707 1,330,264 1,642,971 312,707 221,064 - 91,642

Total 12,577,893 10,448,073 546,383 9,901,689 10,448,073 546,383 347,660 - 198,721

Column Notes:
(2) Assume $1,000,000 for first year in experience period (1999) and 5% annual increase thereafter.
(3) Selected based on estimated overall claim ratio developed in Exhibit III, Sheet 2.
(4) = [(2) x (3)].
(5) Developed in Chapter 7, Exhibit III, Sheets 6 and 8.
(6) = [1.00 - (1.00 / (5))].
(7) = [(4) x (6)].
(8) From last diagonal of reported claim triangles in Chapter 7, Exhibit III, Sheets 6 and 8.
(9) = [(7) + (8)].
(10) = [(9) - (8)].
(11) Developed in Chapter 7, Exhibit III, Sheet 1.
(12) = [(11) - (10)]. 191

Exhibits Combined.xls 10_3_4 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit IV
Impact of Change in Product Mix Example Sheet 1
Scenarios 5 and 6 - Development of Expected Claim Ratio

Age of Reported Reported % of Estimated

Accident Earned Accident Year Claims at CDF to Ultimate Used Up Claim
Year Premium at 12/31/08 12/31/08 Ultimate Reported Premium Ratios
(1) (2) (3) (4) (5) (6) (7) (8)

Steady-State (No Change in Product Mix)

1999 2,000,000 120 1,500,000 1.000 100.0% 2,000,000 75.0%
2000 2,100,000 108 1,575,000 1.000 100.0% 2,100,000 75.0%
2001 2,205,000 96 1,653,750 1.000 100.0% 2,205,000 75.0%
2002 2,315,250 84 1,736,438 1.000 100.0% 2,315,250 75.0%
2003 2,431,013 72 1,814,751 1.005 99.5% 2,419,668 75.0%
2004 2,552,563 60 1,885,068 1.016 98.5% 2,513,424 75.0%
2005 2,680,191 48 1,948,499 1.032 96.9% 2,597,999 75.0%
2006 2,814,201 36 1,937,577 1.089 91.8% 2,583,436 75.0%
2007 2,954,911 24 1,852,729 1.196 83.6% 2,470,306 75.0%
2008 3,102,656 12 1,568,393 1.484 67.4% 2,091,190 75.0%

Total 25,155,785 17,472,204 23,296,273 75.0%

Changing Product Mix

1999 2,000,000 120 1,500,000 1.000 100.0% 2,000,000 75.0%
2000 2,100,000 108 1,575,000 1.000 100.0% 2,100,000 75.0%
2001 2,205,000 96 1,653,750 1.000 100.0% 2,205,000 75.0%
2002 2,315,250 84 1,736,438 1.000 100.0% 2,315,250 75.0%
2003 2,431,013 72 1,814,751 1.005 99.5% 2,419,668 75.0%
2004 2,552,563 60 1,885,068 1.016 98.5% 2,513,424 75.0%
2005 2,999,262 48 2,193,545 1.032 96.9% 2,907,284 75.4%
2006 3,564,016 36 2,471,446 1.090 91.7% 3,269,911 75.6%
2007 4,281,446 24 2,680,487 1.200 83.3% 3,566,552 75.2%
2008 5,196,516 12 2,556,695 1.503 66.5% 3,457,489 73.9%

Total 29,645,066 20,067,179 26,754,578 75.0%

Column and Line Notes:

(2) For no change scenario, assume $2,000,000 for first year in experience period (1999) and 5%
annual increase thereafter. For change scenario, assume annual increase of 30% for commercial
auto beginning in 2005.
(3) Age of accident year at December 31, 2008.
(4) From last diagonal of reported claim triangles in Chapter 7, Exhibit IV, Sheets 2 and 4.
(5) Developed in Chapter 7, Exhibit IV, Sheets 2 and 4.
(6) = [1.00 / (5)].
(7) = [(2) x (6)].
(8) = [(4) / (7)]. 192

Exhibits Combined.xls 10_4_1 04/03/2009 - 2:58 PM

 Chapter 10 - Cape Cod Technique Exhibit IV
Impact of Change in Product Mix Example Sheet 2
U.S. Auto - Development of Unpaid Claim Estimate

Expected Estimated Reported Expected Reported Projected Difference

Accident Earned Claim Expected CDF to Percentage Unreported Claims at Ultimate Estimated Actual from
Year Premium Ratio Claims Ultimate Unreported Claims 12/31/08 Claims IBNR IBNR Actual IBNR
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Steady-State (No Change in Product Mix)

1999 2,000,000 75.0% 1,500,000 1.000 0.0% 0 1,500,000 1,500,000 0 0 0
2000 2,100,000 75.0% 1,575,000 1.000 0.0% 0 1,575,000 1,575,000 0 0 0
2001 2,205,000 75.0% 1,653,750 1.000 0.0% 0 1,653,750 1,653,750 0 0 0
2002 2,315,250 75.0% 1,736,438 1.000 0.0% 0 1,736,438 1,736,438 0 0 0
2003 2,431,013 75.0% 1,823,259 1.005 0.5% 8,509 1,814,751 1,823,259 8,509 8,509 0
2004 2,552,563 75.0% 1,914,422 1.016 1.5% 29,354 1,885,068 1,914,422 29,354 29,354 0
2005 2,680,191 75.0% 2,010,143 1.032 3.1% 61,644 1,948,499 2,010,143 61,644 61,644 0
2006 2,814,201 75.0% 2,110,651 1.089 8.2% 173,073 1,937,577 2,110,651 173,073 173,073 0
2007 2,954,911 75.0% 2,216,183 1.196 16.4% 363,454 1,852,729 2,216,183 363,454 363,454 0
2008 3,102,656 75.0% 2,326,992 1.484 32.6% 758,599 1,568,393 2,326,992 758,599 758,599 0

Total 25,155,785 18,866,839 1,394,634 17,472,204 18,866,839 1,394,634 1,394,634 0

Changing Product Mix

1999 2,000,000 75.0% 1,500,093 1.000 0.0% 0 1,500,000 1,500,000 0 0 0
2000 2,100,000 75.0% 1,575,098 1.000 0.0% 0 1,575,000 1,575,000 0 0 0
2001 2,205,000 75.0% 1,653,853 1.000 0.0% 0 1,653,750 1,653,750 0 0 0
2002 2,315,250 75.0% 1,736,545 1.000 0.0% 0 1,736,438 1,736,438 0 0 0
2003 2,431,013 75.0% 1,823,373 1.005 0.5% 8,509 1,814,751 1,823,260 8,509 8,509 -1
2004 2,552,563 75.0% 1,914,541 1.016 1.5% 29,356 1,885,068 1,914,424 29,356 29,354 -2
2005 2,999,262 75.0% 2,249,586 1.032 3.1% 68,987 2,193,545 2,262,532 68,987 71,855 2,867
2006 3,564,016 75.0% 2,673,178 1.090 8.3% 220,593 2,471,446 2,692,039 220,593 239,057 18,464
2007 4,281,446 75.0% 3,211,284 1.200 16.7% 536,204 2,680,487 3,216,691 536,204 596,924 60,720
2008 5,196,516 75.0% 3,897,629 1.503 33.5% 1,304,351 2,556,695 3,861,045 1,304,351 1,445,385 141,035

Total 29,645,066 22,235,179 2,168,000 20,067,179 22,235,179 2,168,000 2,391,084 223,083

Column Notes:
(2) For no change scenario, assume $2,000,000 for first year in experience period (1999) and 5% annual increase thereafter.
For change scenario, assume annual increase of 30% for commercial auto beginning in 2005.
(3) Selected based on estimated overall claim ratios developed in Exhibit IV, Sheet 1.
(4) = [(2) x (3)].
(5) Developed in Chapter 7, Exhibit IV, Sheets 2 and 4.
(6) = [1.00 - (1.00 / (5))].
(7) = [(4) x (6)].
(8) From last diagonal of reported claim triangles in Chapter 7, Exhibit IV, Sheets 2 and 4.
(9) = [(7) + (8)].
(10) = [(9) - (8)].
(11) Developed in Chapter 7, Exhibit IV, Sheet 1.
(12) = [(11) - (10)].
193

Exhibits Combined.xls 10_4_2 04/03/2009 - 2:58 PM