STAT3058/6058 Practice Questions Solutions

Topic 4 - Run-off Triangles

FOR THE CALCULATION-BASED QUESTIONS IN THIS TOPIC, A GOOD WAY TO PRAC-
TICE WILL BE TO DO THEM BOTH BY HAND AND VIA A SPREADSHEET.
1. Explain in words (i.e. without using mathematical symbols) the assumptions underlying the:

(i) basic chain ladder method.

(ii) inflation adjusted chain ladder method.
(iii) payments per claim incurred method.
(iv) Bornhuetter-Ferguson method.

Solution:
(i) Basic chain ladder assumptions

• There is a consistent delay pattern in the payment of claims between accident years.
• There is enough experience to date in the most recent accident years to allow development of that
experience to estimate outstanding claims.

• The first accident year is fully run off.

• Claims in each development year are a constant proportion (in monetary terms) of total claims in
each accident year.
• Inflation is not allowed for explicitly, rather it is allowed for implicitly as a weighted average of past
inflation.

(ii) Inflation-adjusted chain ladder assumptions

• For each accident year, the amount of claims paid, in real terms, in each development year, is a
constant proportion of the total claims, in real terms, from that accident year.

• The first accident year is fully run off.

• Inflation is allowed for explicitly and we assume that both the past and future inflation assumptions
are correct.

(iii) PPCI assumptions

• For each accident year, both the number and average amount of claims relating to each development
year are constant proportions of the totals from that accident year.

1
 • The first accident year is fully run off.

• Inflation is allowed for explicitly and we assume that both the past and future inflation assumptions
are correct.

(iv) BF assumptions

• The first accident year is fully run-off.

• The loss ratios are appropriate.
• The future claim development pattern will follow the pattern experienced for the more developed
accident years.

• Claims in each development year are a constant proportion (in monetary terms) of total claims in
each accident year.
• Inflation is not allowed for explicitly, rather it is allowed for implicitly as a weighted average of past
inflation.

2. The table below shows cumulative claims (not adjusted for inflation) from a portfolio of insurance policies
for 4 accident years.

Development Year
Accident Year 0 1 2 3
2014 2,047 3,141 3,209 3,310
2015 2,471 3,712 3,810
2016 2,388 3,750
2017 2,580

It may be assumed that payments are made in the middle of a calendar year. In addition, the discount rate
is assumed to be 4% per annum.

It is estimated that the inflation rate applicable to these data has been 5% per annum over the relevant period.

Use the inflation adjusted chain ladder method to estimate the total outstanding payments, up to the end of
Development Year 3, for Accident Year 2017 in mid-2017 prices.

Solution:
First we disaccumulate the amounts. This gives the following table:

Incremental Claim
Payments - Nominal
Development Year
Accident Year 0 1 2 3
2014 2,047 1,094 68 101
2015 2,471 1,241 98
2016 2,388 1,362
2017 2,580

2
Now we convert to 2017 prices, using a past inflation rate of 5%:

Incremental Claim
Payments - Real
Development Year
Accident Year 0 1 2 3
2014 2,369.6584 1,206.135 71.4 101
2015 2,724.2775 1,303.05 98
2016 2,507.40 1,362
2017 2,580

Now we find the cumulative claim payments in real terms:

Cumulative Claim
Payments - Real
Development Year
Accident Year 0 1 2 3
2014 2,369.6584 3,575.7934 3,647.1934 3,748.1934
2015 2,724.2775 4,027.3275 4,125.3275
2016 2,507.40 3,869.4
2017 2,580

Calculating the development factors in the usual way:

3, 575.7934 + 4, 027.3275 + 3, 869.4
M1 = = 1.509277
2, 369.6584 + 2, 724.2775 + 2, 507.40
3, 647.1934 + 4, 125.3275
M2 = = 1.022280
3, 575.7934 + 4, 027.3275
3, 748.1934
M3 = = 1.027693
3, 647.1934
We now project the cumulative claim payments (still working in 2017 values):

Projected Cumulative Claim

Payments - Real
Development Year
Accident Year 0 1 2 3
2014
2015 4,239.57
2016 3,955.61 4,065.15
2017 3,893.93 3,980.70 4,090.93

Disaccumulating the claim payments:

Projected Claim
Payments - Real
Development Year
Accident Year 0 1 2 3
2014
2015 114.24
2016 86.21 109.24
2017 1,313.93 86.76 110.24

3
 Now we discount the claim payments at a rate of 4% p.a.:

Discounted Projected Claim

Payments - Real
Development Year
Accident Year 0 1 2 3
2014
2015 112.02
2016 84.54 103.28
2017 1,288.42 81.80 99.94

Hence, before the discounting, the total amount of outstanding payments for 2017 at mid-2017 prices is:

1, 313.93 + 86.76 + 110.24 = 1510.93

After the discounting, the total amount of outstanding payments for 2017 at mid-2017 prices is:

1, 288.42 + 81.80 + 99.94 = 1470.16

Note that this question asked us to find the outstanding amount in 2017 prices. If we had been asked to
find the outstanding amount in nominal (actual) prices, we would have had to apply one further stage -
adjusting for future inflation.

3. The table below shows the claim payments made by a general insurer in each year for a particular type of
insurance.
Claim Payments
Made during
Year ($’000s)
Development Year
Accident Year 0 1 2 3
2012 10 50 50 30
2013 50 70 30
2014 40 30
2015 90

(i) What was the total amount paid during the 2015 calendar year?

(ii) Calculate the development factors from development years 1 to 2 for each of the 2012 and 2013 accident
years.

(iii) Given three reasons why it may not be appropriate to use the basic chain ladder method to project the
claim payments for this portfolio, using figures from the table to support your comments.

Solution:
(i) The payments made during 2015 correspond to the figures in the longest diagonal, which total:

90 + 30 + 30 + 30 = 180

4
 So the total amount paid during 2015 was $180,000.

(ii) The development factor for Accident Year 2012, Development Year 2 is:
10 + 50 + 50 110
= = 1.83̇
10 + 50 60
The development factor for Accident Year 2013, Development Year 2 is:
50 + 70 + 30 150
= = 1.25
50 + 70 120

(iii) Reason why the basic chain ladder method may not be appropriate here include:

1. The first accident year may not be fully run-off.

2. The development ratios do not appear to be constant for each origin year. This is illustrated by
the ratios calculated in part (ii).
3. The total payment amounts are relatively small (e.g. a total of $180,000 paid in 2015). There may
not be enough policies for the statistical model to apply, since the figures may be dominated by the
random errors eij .
4. The basic chain ladder does not explicitly model inflation but assumes that a weighted average
of past inflation will apply in the future. If inflation for calendar years 2012 to 2015 has varied
significantly, a chain ladder calculation will not give reliable results. The figures in the table are
too erratic (e.g. the values in Development Year 0 vary from $10,000 to $90,000) to judge whether
this is the case.

4. The tables below show the cumulative cost of incurred claims and the number of claims reported each year for
a certain cohort of insurance policies. The claims are assumed to be fully run-off at the end of Development
Year 2.

Cumulative cost of incurred claims:

Development Year
Accident Year 0 1 2
0 288 634 893
1 465 980
2 773

The numbers of claims reported in each year are:

Development Year
Accident Year 0 1 2
0 110 85 55
1 167 113
2 285

Given that the discounting rate is 4% p.a., and the total amount paid in claims to date, relating to accident
years 0, 1 and 2, is $2,750, calculate the outstanding claims reserve using the PPCI method.

5
Solution:
(a) Method I: This is the method discussed in the course.
Based on the cumulative numbers of claims reported, we estimate the development factors and fill the
lower triangle:

Development Year
0:1 1:2
Mj∗ 1.715 1.282

Projected Cumulative Claim

Development Year
Accident Year 0 1 2
0
1 358.974
2 488.718 626.562

Next we disaccumulate the claim payments:

Claim Payments
Development Year
Accident Year 0 1 2
0 288 346 259
1 465 515
2 773

Then we calculate the PPCIs (assume the average PPCIs are the selected “representative” PPCIs):

Payments per Claim Incurred

Development Year
Accident Year 0 1 2
0 1.1520 1.3840 1.0360
1 1.2954 1.4346
2 1.2337

Average 1.2270 1.4093 1.0360

Now we fill the lower triangle:

Projected Payments
per Claim Incurred
Development Year
Accident Year 0 1 2
0
1 1.0360
2 1.4093 1.0360

Then we project the future claim payments:

6
 Projected Claim Payments
Development Year
Accident Year 0 1 2
0
1 372
2 883 649

After making 4% p.a. discounting adjustment, we obtain:

Discounted Projected
Claim Payments
Development Year
Accident Year 0 1 2
0
1 365
2 866 612

Hence, with the incremental past data and the discounted projection results, the total expected ultimate
loss will be 4,489. Moreover, since the claims paid to date are 2,750, the outstanding claims reserve is
1,739.

(b) Method II: This is another way to do the PPCI method for your information.
First we need the cumulative numbers of claims reported:

Cumulative Claims Reported

Development Year
Accident Year 0 1 2
0 110 195 250
1 167 280
2 285

We now divide the cumulative cost by the cumulative claim numbers to get the average cost per claim:

Payments per Claim Incurred

Development Year
Accident Year 0 1 2
0 2.6182 3.2513 3.5720
1 2.7844 3.5000
2 2.7123

Assuming that Accident Year 0 is fully run off, we divide the numbers in the top row by 3.572 to get the
percentages 73.297%, 91.021%, 100%.
So the ultimate average cost per claim figure for Accident Year 1 is:
3.5
= 3.84525
0.91021

We can now calculate the percentage figures for Accident Year 1 as 72.412%, 91.021%.

7
 For accident Year 2, we take the average of the two previous figures for Development Year 0:
1
× (72.412 + 73.297) = 72.855%
2

So the ultimate figure for Accident Year 2 is:

2.71228
= 3.72286
0.72854

Using the same approach for the claim number figures, we get:

Cumulative Claims Reported

Development Year
Accident Year 0 1 2
0 110 195 250
44% 78% 100%
1 167 280
46.52% 78%
2 285
45.26%

These give ultimate claim number values of 250, 358.974 and 629.685.

So after the 4% p.a. discounting adjustment, the estimated total expected ultimate loss will be:

250 × 3.5720 + 358.97 × 3.8453 × 1.04−0.5 + 629.69 × 3.7229 × 1.04−1.5 = 4, 456.88.

Since the claims paid to date are 2,750, the outstanding claims reserve is 1,706.88.

5. The following table shows cumulative incurred claims data, by year of accident and reporting development,
for a portfolio of domestic household insurance policies:

Cumulative
Incurred Claims
($’000s)
Development Year
Accident Year 0 1 2 3
2011 829 917 978 1,020
2012 926 987 1,053
2013 997 1,098
2014 1,021

The corresponding cumulative number of reported claims, by year of accident and reporting development, are
as follows:

8
 Cumulative
Number of
Reported Claims
Development Year
Accident Year 0 1 2 3
2011 63 68 70 74
2012 65 69 72
2013 71 76
2014 70

Use the average cost per claim method with simple average grossing up factors to calculate an estimate of the
outstanding claim amount for these policies for claims arising during these accident years. The claims paid to
date are $3,640,000. In addition, the discount rate is assumed to be 4% per annum. State any assumptions
used.

Solution:
(a) Method I: This is the method discussed in the course.
First estimate the development factors:

Development Year
0:1 1:2 2:3
Mj∗ 1.070 1.036 1.057

Then project the cumulative claims:

Projected Cumulative Claim

Development Year
Accident Year 0 1 2 3
2011
2012 76
2013 79 83
2014 75 78 82

Next disaccumulate the claim payments:

Claim Payments ($’000s)

Development Year
Accident Year 0 1 2 3
2011 829 88 61 42
2012 926 61 66
2013 997 101
2014 1,021

Calculate the PPCIs (assume the average PPCIs are the selected “representative” PPCIs):

9
 Payments per Claim Incurred
Development Year
Accident Year 0 1 2 3
2011 11,202.70 1,189.19 824.32 567.57
2012 12,165.92 801.43 867.12
2013 11,972.37 1,212.85
2014 12,436.54

Average 11,944.38 1,067.82 845.72 567.57

Then fill the lower triangle:

Projected Payments per Claim Incurred

Development Year
Accident Year 0 1 2 3
2011
2012 567.57
2013 845.72 567.57
2014 1,067.82 845.72 567.57

Then project the claim payments:

Projected Claim Payments ($’000s)

Development Year
Accident Year 0 1 2 3
2011
2012 43.20
2013 70.43 47.26
2014 87.66 69.43 46.60

Next discount the results at a rate of 4% p.a.:

Discounted Projected
Claim Payments ($’000s)
Development Year
Accident Year 0 1 2 3
2011
2012 42.36
2013 69.06 44.56
2014 85.96 65.46 42.24

Hence, with the incremental past data and the discounted projection results, the ultimate claim incurred
amount from accident years 2011 to 2014 is 4,541.64. The claim paid to date (from accident years 2011
to 2014) amount to 3,640, resulting in a total outstanding claim amount of 901.64, i.e. $901,640.
(b) Method II: This is another way to do the PPCI method.
Dividing each cell in the first table by the corresponding cell in the second table gives the cumulative
average incurred cost per claim, by year of accident and reporting development:

10
 Payments per Claim Incurred
Development Year
Accident Year 0 1 2 3
2011 13.1587 13.4853 13.9714 13.7838
2012 14.2462 14.3043 14.6250
2013 14.0423 14.4474
2014 14.5857

Completing the ultimate cumulative number of claims reported table and the ultimate cumulative average
incurred cost per claim table using the chain ladder technique with simple grossing up factors gives:

Cumulative Claims Reported

Development Year
Accident Year 0 1 2 3
2011 63 68 70 74
2012 65 69 72
2013 71 76
2014 70

Projected Cumulative Claim

Development Year
Accident Year 0 1 2 3
2011
2012 76.114
2013 83.267
2014 82.095

The simple average grossing up factors used are:

• Development Year 2 to 3: 94.595%

• Development Year 1 to 3: 91.273% (average of 91.892% and 90.653%)

• Development Year 0 to 3: 85.267% (average of 85.135%, 85.398% and 85.268%)

Payments per Claim Incurred

Development Year
Accident Year 0 1 2 3
2011 13.159 13.485 13.971 13.784
2012 14.246 14.304 14.625
2013 14.042 14.447
2014 14.586

11
 Projected Payments per
Claim Incurred ($’000s)
Development Year
Accident Year 0 1 2 3
2011
2012 14.429
2013 14.669
2014 15.093

The simple average grossing up factors used are:

• Development Year 2 to 3: 101.361%

• Development Year 1 to 3: 98.487% (average of 97.834% and 99.139%)

• Development Year 0 to 3: 96.642% (average of 95.465%, 98.736% and 95.725%)

Therefore, with the incremental past data and the discounted projection results, the estimated ultimate
claims incurred amount from accident years 2011 to 2014 is:

1, 020 + 76.114 × 14.429 × 1.04−0.5 + 83.267 × 14.669 × 1.04−1.5 + 82.095 × 15.093 × 1.04−2.5 = 4, 372

The claim paid to date (from accident years 2011 to 2014) amount to 3,640, resulting in a total outstand-
ing claim amount of 732, i.e. $732,000.

Assumptions:

• For each accident year, both the number and average amount of claims relating to each development
year are constant proportions of the totals from that accident year.
• The first accident year is fully run off.

• Inflation is allowed for explicitly and we assume that both the past and future inflation assumptions
are correct.

6. Using the claim payment triangle given in the lecture notes and a discount rate of 4% p.a., apply the IACL
method so as to obtain an estimate of the total outstanding claims amount.

Claims Payments ($’000)

Development Year
Accident Year 0 1 2 3 4
2002 786 624 806 224 79
2003 904 671 940 281
2004 995 819 1,066
2005 1,220 922
2006 1,182

The inflation rates for claims costs in consecutive years are as follows.

12
 Years 2002/03 2003/04 2004/05 2005/06
Rate (%) 6 6 8 8

Years 2006/07 2007/08 2008/09 2009/10

Projected Rate (%) 7 7 7 7

Solution:
Inflate:

Claim Payments in 2006 $s

Development Year
Accident Year 0 1 2 3 4
2002 1,030.11 771.50 940.12 241.92 79.00
2003 1,117.69 782.65 1,015.20 281.00
2004 1,160.57 884.52 1,066.00
2005 1,317.60 922.00
2006 1,182.00

Accumulate:

Cumulative Claim Payments ($’000)

Development Year
Accident Year 0 1 2 3 4
2002 1,030.11 1,801.61 2,741.73 2,983.65 3,062.65
2003 1,117.69 1,900.35 2,915.55 3,196.55
2004 1,160.57 2,045.09 3,111.09
2005 1,317.60 2,239.60
2006 1,182.00

Then we calculate the ratios:

Development Year
0:1 1:2 2:3 3:4
Mj∗ 1.73 1.53 1.09 1.03

Use these “average” ratios as development factors.

Project cumulative claim payments:

Projected Cumulative Claim Payments ($’000)

Development Year
Accident Year 0 1 2 3 4
2002
2003 3,281.18
2004 3,398.66 3,488.64
2005 3,417.00 3,732.84 3,831.68
2006 2,040.70 3,113.53 3,401.33 3,491.39

Disaccumulate:

13
 Projected Claim Payments ($’000)
Development Year
Accident Year 0 1 2 3 4
2002
2003 84.64
2004 287.57 89.99
2005 1,177.40 315.84 98.84
2006 858.70 1,072.83 287.79 90.06

Inflate:

Inflated Projected Claim Payments ($’000)

Development Year
Accident Year 0 1 2 3 4
2002
2003 90.56
2004 307.70 103.03
2005 1,259.81 361.61 121.08
2006 918.81 1,228.28 352.56 118.05

Discount:

Discounted Projected Claim Payments ($’000)

Development Year
Accident Year 0 1 2 3 4
2002
2003 88.80
2004 301.72 97.14
2005 1,235.35 340.95 109.77
2006 900.97 1,158.11 319.63 102.91

Finally, we get the outstanding claims reserve:

(900.97 + 1235.35 + 1158.11 + · · · + 102.91) × 1, 000 = $4, 655, 350.

[compare with $4,659,000 using BCL]

7. Consider the following run-off triangle which shows the (inflation-adjusted) amounts, in units of $10,000, that
were paid for industrial accidents by the Rand Insurance Corporation in the years prior to 1 January 1994.

Projected No. of
Development Year Claims Reported
Accident Year 0 1 2 in Accident Year
1991 6.3 5.8 2.0 340
1992 8.6 9.1 466
1993 24.0 872

For example, $91,000 was paid in 1993 for accidents that occurred in 1992.

14
Use the PPCI method to estimate the total amount that is yet to be paid by Rand for all accidents which
occurred in 1992 and 1993. Assume an annual claims inflation rate of 20% from 1993 to 1994 and 10% from
1994 to 1995, and a discount rate of 17%.

Solution:
Calculate the PPCIs:

Payments per Claim Incurred

Development Year
Accident Year 0 1 2
1991 185.2941 170.5882 58.8235
1992 184.5494 195.2790
1993 275.2294

Average 215.02 182.93 58.82

Assume that the average PPCIs are representative and fill in the lower triangle:

Projected Payments
per Claim Incurred
Development Year
Accident Year 0 1 2
1991
1992 58.82
1993 182.93 58.82

Project future claim payments:

Projected Claim
Payments ($’0,000s)
Development Year
Accident Year 0 1 2
1991
1992 2.74
1993 15.95 5.13

After inflating and discounting the results, we obtain:

Discounted Projected
Claim Payments ($’0,000s)
Development Year
Accident Year 0 1 2
1991
1992 3.04
1993 17.70 5.35

Finally, we calculate the outstanding claims estimate:

(17.70 + 3.04 + 5.35) × 10, 000 = $260, 900.

15
8. The table below shows the numbers of household insurance claims reported in each development year.

Claims Reported
Development Year
Accident Year 0 1 2 3
1989 17,500 5,000 2,250 750
1990 21,000 6,200 2,750
1991 18,800 5,500
1992 21,300

Use the basic chain ladder method to estimate the total ultimate number of claims arising from accidents
occurring between 1 January 1989 and 31 December 1992.

Solution:
Accumulate the results:

Cumulative Claim Reported

Development Year
Accident Year 0 1 2 3
1989 17,500 22,500 24,750 25,500
1990 21,000 27,200 29,950
1991 18,800 24,300
1992 21,300

Then we calculate the ratios and obtain the development factors:

Development Year
0:1 1:2 2:3
Mj∗ 1.291 1.101 1.030

After using the average development ratios as the selected development ratios, we fill the lower triangle:

Projected Cumulative Claims

Development Year
Accident Year 0 1 2 3
1989
1990 30,857.58
1991 26,744.67 27,555.11
1992 27,507.85 30,275.24 31,192.67

Finally, we get the projected ultimate number of claims:

25, 500 + 30, 857.58 + 27, 555.11 + 31, 192.67 = 115, 105.36 ≈ 115, 105.

9. An insurance company has paid the following claim amounts (in $’000s):

16
 Development Year
Accident Year 1 2 3 4 5
1 2,800 1,400 987 322 57
2 3,260 2,004 1,017 421
3 3,854 1,978 857
4 3,722 2,114
5 4,627
The earned premium in each year is 6,727 for Accident Year 1, 8,289 for Accident Year 2, 9,627 for Accident
Year 3, 9,928 for Accident Year 4 and 10,004 for Accident Year 5. In addition, the discount rate is assumed
to be 4% per annum.

Apply the Bornhuetter-Ferguson method to estimate the amount of claims yet to be paid, stating any as-
sumptions that you make.

Solution:
We need the cumulative claims data:

Cumulative Claim Payments ($’000s)

Development Year
Accident Year 1 2 3 4 5
1 2,800 4,200 5,187 5,509 5,566
2 3,260 5,264 6,281 6,702
3 3,854 5,832 6,689
4 3,722 5,836
5 4,627

The ultimate loss ratio is 5,566

6,727 = 0.827412.

Next we calculate the expected end of year figures (the initial ultimate liability):

• 2: 0.827412 × 8, 289 = 6, 858.4

• 3: 0.827412 × 9, 627 = 7, 965.5

• 4: 0.827412 × 9, 928 = 8, 214.5
• 5: 0.827412 × 10, 004 = 8, 277.4

The development factors are:

• Year 4 to Year 5: 5,566

5,509 = 1.010347

• Year 3 to Year 4: 5,509+6,702

5,187+6,281 = 1.064789

• Year 2 to Year 3: 5,187+6,281+6,689

4,200+5,264+5,832 = 1.187042

• Year 1 to Year 2: 4,200+5,264+5,832+5,836

2,800+3,260+3,854+3,722 = 1.549721

The emerging liabilities for each year are:

• 2: 6, 858.4 × 1 − 1


1.010347 = 70.2

17
  
• 3: 7, 965.5 × 1 − 1
1.010347×1.064789 = 561.3
 
• 4: 8, 214.5 × 1 − 1
1.010347×1.064789×1.187042 = 1, 782
 
• 5: 8, 277.4 × 1 − 1
1.010347×1.064789×1.187042×1.549721 = 4, 094.9

Given that these are claims paid (rather than incurred), we don’t need to calculate the revised ultimate
liability to get the reserve. We can just total up the emerging liabilities (with 4% p.a. discounting
adjustment) and obtain the estimation amount of $6,262,000.

The assumptions are:

• The first accident year is fully run-off.

• The loss ratios are appropriate.

• The future claim development pattern will follow the pattern experienced for the more developed
accident years.
• Claims in each development year are a constant proportion (in monetary terms) of total claims in
each accident year.

• Inflation is not allowed for explicitly, rather it is allowed for implicitly as a weighted average of past
inflation.

10. The table below shows the cumulative cost of incurred claims (in $’000s). The claims are assumed to be fully
run-off at the end of Development Year 2.

Development Year
Accident Year 0 1 2
2011 2,670 3,290 4,310
2012 2,850 3,420
2013 3,030

Annual premiums written were:

Year Premiums ($’000s)

2011 5,390
2012 5,600
2013 6,030

The ultimate loss ratio has been estimated at 80% and the total amount of claims paid to date is $5,720,000.
In addition, the discount rate is assumed to be 4% per annum. Calculate the outstanding claims reserve using
the Bornhuetter-Ferguson method.

Solution:
First wee calculate the expected end of year figures (the initial ultimate liability):

18
 • 2011: 0.8 × 5, 390 = 4, 312

• 2012: 0.8 × 5, 600 = 4, 480

• 2013: 0.8 × 6, 030 = 4, 824

Now we calculate the development factors:

3, 290 + 3, 420 6, 710
= = 1.215580
2, 670 + 2, 850 5, 520
4, 310
= 1.310030
3, 290
The emerging liabilities for each year are:

• 2011: 4, 312 × (1 − 11 ) = 0
• 2012: 4, 480 × (1 − 1
1.310030 ) = 1, 060.2
• 2013: 4, 824 × (1 − 1
1.215580×1.310030 ) = 1, 794.7

So after 4% p.a. discounting adjustment (for both past and future), the ultimate liabilities for each year
are:

• 2011: 4, 310 + 0 = 4, 310

• 2012: 3, 420 + 1,060.2

1.040.5 = 4, 459.61

• 2013: 3, 030 + 653.1

1.040.5 + 1,141.6
1.041.5 = 4, 746.79

Thus the total ultimate liability is

4, 310 + 4, 459.61 + 4, 746.79 = 13, 516.40

Therefore the outstanding claims reserve is 13, 516.40 − 5, 720 = 7, 796.40, i.e. $7,796,400.

11. You have been asked to estimate the outstanding claims liability of a general insurer’s workers’ compensation
portfolio as at 31 December 1996 and have been provided with the following initial information:

Total Number of
Claims Reported
Development Year
Accident Year 0 1 2 3
2003 1,500 300 100 50
2004 1,600 300 100
2005 1,600 350
2006 1,500

19
 Projected Total Number
of Claims
Development Year
Accident Year 0 1 2 3
2003
2004 53
2005 105 54
2006 303 97 50

Total (Inflation-adjusted)
Claims Payments ($’000)
Development Year
Accident Year 0 1 2 3
2003 6,000 4,000 2,500 1,500
2004 7,000 4,000 2,000
2005 8,000 3,500
2006 8,000

Perform a payments per claim incurred analysis on total claim payments to estimate the outstanding claim
liability as at the end of the 2006 accident year. Assume a discount rate of 5% p.a., future inflation at 3.5%
p.a., and note that no further payments are made beyond development year 3. You may use the average
PPCIs as the selected “representative” PPCIs.

Solution:
Total incurred claims:

Accident Year Ultimate Claims

2003 1,950
2004 2,053
2005 2,109
2006 1,950

Calculate the PPCIs:

Payments per Claim Incurred

Development Year
Accident Year 0 1 2 3
2003 3,076.92 2,051.28 1,282.05 769.23
2004 3,409.64 1,948.37 974.18
2005 3,793.27 1,659.55
2006 4,102.56

Average 3,595.60 1,886.40 1,128.12 769.23

[Note: values are multiplied by 1000 since original payments table was in $’000]
Fill in the lower triangle:

20
 Projected Payments per Claim Incurred
Development Year
Accident Year 0 1 2 3
2003
2004 769.23
2005 1,128.12 769.23
2006 1,886.40 1,128.12 769.23

Project future claim payments:

Projected Claim Payments ($’000s)

Development Year
Accident Year 0 1 2 3
2003
2004 1,579.23
2005 2,379.20 1,622.31
2006 3,678.48 2,199.83 1,500.00

After inflating and discounting the results, we obtain:

Discounted Projected Claim Payments ($’000s)

Development Year
Accident Year 0 1 2 3
2003
2004 1,595.11
2005 2,403.13 1,615.21
2006 3,715.48 2,190.21 1,472.11

Finally, we get the total outstanding claims liability:

(3, 715.48 + 2, 403.13 + · · · + 1, 472.11) × 1, 000 = $12, 991, 245.

12. Apply the inflation-adjusted Bornhuetter-Ferguson method to the claim payments triangle given in question
4 and thus estimate the insurer’s outstanding claims. You may use the future inflation and discounting
assumptions given in question 4 and assume that the “average” development ratios were selected as the
representative chain ladder development factors. You have also been given the following information:

Year 2004 2005 2006

Earned Premium 14,567,000 13,881,000 12,400,000

Loss Ratio 90% 87% 89%

Solution:
SAME QUESTION AS TUTORIAL 8 Q2. SO PLEASE IGNORE THIS QUESTION.

21
13. The table below shows the amounts paid, in thousands of dollars, for car accident claims by the Bellco Insur-
ance Company for accident years 1994, 1995 and 1996, developed up to the end of 1996.

Claim Payments ($’000)

Development Year
Accident Year 0 1 2
1994 480 236 83
1995 300 155
1996 118

(For example, $83,000 was paid in 1996 for car accidents that occurred in 1994.)

It may be assumed that no claims take longer than two years to finalise.

(a) Assume a discount rate of 7% and suppose that claims inflation rates are as given in the following table:

Years 1994/95 1995/96 1996/97 1997/98

Rate (%) 10 8 6 4

Apply the inflation adjusted chain ladder method so as to obtain an estimate of the total ultimate claims
amount (in current values) arising from accidents that occurred in 1996.

(b) Suppose that for accident year 1995, $250,000 of premiums were earned. Assume a loss ratio of 92%.
Use the inflation-adjusted Bornhuetter-Ferguson method to estimate the outstanding claims amount arising
from accidents in 1995.

Solution:
(a) Adjust all figures to 1996 money:

Claim Payments in 1996 $s

Development Year
Accident Year 0 1 2
1994 570.24 254.88 83.00
1995 324.00 155.00
1996 118.00

Next accumulate:

Cumulative Claim Payments ($’000)

Development Year
Accident Year 0 1 2
1994 570.24 825.12 908.12
1995 324.00 479.00
1996 118.00

Then we calculate the ratios and obtain the development factors:

22
 Development Year
0:1 1:2
Mj∗ 1.4584 1.1006

Next estimate the missing cumulative figures:

Projected Cumulative
Claim Payments ($’000)
Development Year
Accident Year 0 1 2
1994
1995 527.18
1996 172.09 189.40

Then disaccumulate:
Projected Claim
Payments ($’000)
Development Year
Accident Year 0 1 2
1994
1995 48.18
1996 54.09 17.31

Next inflate (for 1996 only), we get:

54.09 × 1.06 = 57.34
17.31 × 1.06 × 1.04 = 19.08

After discounting and summing up the results (for 1996 only), we get:

118 + 57.34 × 1.07−0.5 + 19.08 × 1.07−1.5 × 1, 000 = $190, 671.

(b) Using the chain ladder factors from part (a), the cumulative chain ladder factors and IBNR factors
are:

0:1 1:2
Factor 1.45836 1.10059
Cumulative Factor 1.60506 1.10059
IBNR Factor 0.3770 0.0914

Then we get the Outstanding Claims Estimate:

250, 000 × 92% × 0.0914 = 21, 021.18

After inflating and discounting the results, we obtain:

21, 021.18 × 1.06 × 1.07−0.5 = 21, 541.26.

So the required estimate is $21,541.

23
14. Given the following data, estimate the claims outstanding as at 31/12/2000 using:

(a) the inflation-adjusted chain ladder method;

(b) the payments per claim incurred method;
(c) the inflation-adjusted Bornhuetter-Ferguson method.

Inflation Adjusted Cumulative Paid Claims

Development Year
Accident Year 0 1 2 3 4
1996 29,791 52,054 54,890 55,267 55,345
1997 28,620 31,130 48,885 57,141
1998 27,935 40,060 49,962
1999 37,661 47,683
2000 19,619

Projected
Accident Year
Claim Reported
1996 40
1997 50
1998 19
1999 38
2000 17

Accident Year 1996 1997 1998 1999 2000

Earned Premium 63,907 74,913 71,715 76,802 54,903
Ultimate Loss Ratio 90% 90% 90% 90% 90%

Assume an inflation rate of 10% p.a. and a discount rate of 7% p.a. Furthermore, you may assume that the
“average” chain ladder factors and the average PPCIs are representative, where applicable.

Solution:
(a) Calculate the development factors:

Development Year
0:1 1:2 2:3 3:4
Mj∗ 1.378 1.247 1.083 1.001

Project cumulative payments:

Projected Cumulative Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 57,222
1998 54,118 54,195
1999 59,481 64,429 64,520
2000 27,042 33,733 36,539 36,591

Disaccumulate:

24
 Projected Claim Payments
Development Year
Accident Year 0 1 2 3 4
1996
1997 81
1998 4,156 76
1999 11,798 4,948 91
2000 7,423 6,691 2,806 52

After inflating and discounting the results, we get:

Discounted Projected Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 86
1998 4,420 83
1999 12,546 5,409 102
2000 7,894 7,315 3,154 60

So the outstanding claims estimate is:

7, 894 + 12, 546 + · · · + 60 = $41, 068.

(b) Disaccumulate:

Inflation Adjusted Paid Claims

Development Year
Accident Year 0 1 2 3 4
1996 29,791 22,263 2,836 377 78
1997 28,620 2,510 17,755 8,256
1998 27,935 12,125 9,902
1999 37,661 10,022
2000 19,619

Calculate PPCIs:

Payments per Claim Incurred

Development Year
Accident Year 0 1 2 3 4
1996 745 557 71 9 2
1997 572 50 355 165
1998 1,470 638 521
1999 991 264
2000 1,154

Average 987 377 316 87 2

Fill in Lower Triangle:

25
 Projected Payments per Claim Incurred
Development Year
Accident Year 0 1 2 3 4
1996
1997 2
1998 87 2
1999 316 87 2
2000 377 316 87 2

Project Claim Payments:

Projected Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 98
1998 1,658 37
1999 11,997 3,316 74
2000 6,412 5,367 1,484 33

After inflating and discounting the results, we get:

Discounted Projected Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 104
1998 1,763 41
1999 12,758 3,626 83
2000 6,818 5,868 1,667 38

So the outstanding claims estimate is:

6, 818 + 12, 758 + · · · + 38 = $32, 766.

(c) Note: the selected development factors are the same as those used in (a).
Calculate Cumulative Chain Ladder Factors and IBNR Factors:
Development Year
0:1 1:2 2:3 3:4
Mj∗ 1.378 1.247 1.083 1.001
Xj 1.865 1.353 1.085 1.001
Fj 0.464 0.261 0.078 0.001

Calculate IBNR Estimates:

Accident Year Osi
1997 95.02
1998 5,040.96
1999 18,037.74
2000 22,918.88

26
 Spread estimates across future development years:

Projected Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 95
1998 4,950 91
1999 12,639 5,301 97
2000 10,024 9,035 3,790 70

After inflating and discounting the results, we get:

Discounted Projected Claim Payments

Development Year
Accident Year 0 1 2 3 4
1996
1997 101
1998 5,264 99
1999 13,441 5,795 109
2000 10,660 9,878 4,259 80

So the outstanding claims estimate is:

10, 660 + 13, 441 + · · · + 80 = $49, 687.

15. The cumulative claims paid each year under a certain cohort of insurance policies are recorded in the table
below, for accident years 2010, 2011, 2012 and 2013.
Development Year
Accident Year 0 1 2 3
2010 2,457 4,196 4,969 5,010
2011 2,648 4,715 5,561
2012 3,084 5,315
2013 3,341

(i) Calculate the development factors under the basic chain ladder technique and state the assumptions un-
derlying the use of this method.

(ii) The rate of claims inflation over these years, measured over the 12 months to the middle of each year, is
given in the table below.
2011 2.1%
2012 10.5%
2013 3.2%

Calculate the development factors under the inflation-adjusted chain ladder technique and state the assump-
tions underlying the use of this method.

(iii) Based on the development factors calculated in parts (i) and (ii), calculate the fitted values under these
two models and comment on how these compare with the actual values.

27
Solution:
(i) BCL development factors and assumptions

The development factors are:

4,196+4,715+5,315
M1 = 2,457+2,648+3,084 = 1.737208

4,969+5,561
M2 = 4,196+4,715 = 1.181686

5,010
M3 = 4,969 = 1.008251

The assumptions underlying the method are:

• There is a consistent delay pattern in the payment of claims between accident years.

• There is enough experience to date in the most recent accident years to allow development of that
experience to estimate outstanding claims.

• The first accident year is fully run off.

• Claims in each development year are a constant proportion (in monetary terms) of total claims in
each accident year.

• Inflation is not allowed for explicitly, rather it is allowed for implicitly as a weighted average of past
inflation.

(ii) Inflation-adjusted chain ladder development factors and assumptions

First we need to disaccumulate the figures to obtain incremental figures. If we do this we obtain the
following results:

Development Year
Accident Year 0 1 2 3
2010 2,457 1,739 773 41
2011 2,648 2,067 846
2012 3,084 2,231
2013 3,341

We now inflate the incremental figures to get real values (in mid-2013 terms):

Development Year
Accident Year 0 1 2 3
2010 2,860.70 1,983.09 797.74 41
2011 3,019.67 2,133.14 846
2012 3,182.69 2,231
2013 3,341

We can now find the cumulative figures in real terms:

28
 Development Year
Accident Year 0 1 2 3
2010 2,860.70 4,843.79 5,641.53 5,682.53
2011 3,019.67 5,152.81 5,998.81
2012 3,182.69 5,413.69
2013 3,341

So the development factors are now:

4,843.79+5,152.81+5,413.69
M1 = 2,860.70+3,019.67+3,182.69 = 1.700340

5,641.53+5,998.81
M2 = 4,843.79+5,152.81 = 1.164429

5,682.53
M3 = 5,641.53 = 1.007268

The assumptions underlying the method are:

• For each accident year, the amount of claims paid, in real terms, in each development year, is a
constant proportion of the total claims, in real terms, from that accident year.

• The first accident year is fully run off.

• Inflation is allowed for explicitly and we assume that both the past and future inflation assumptions
are correct.

(iii) Fitted values

We now use the development factors to find the fitted claim amounts for past years. Using the development
factors for the basic chain ladder method we get:

Development Year
Accident Year 0 1 2 3
2010 2,457 4,268.32 5,043.81 5,085.43
2011 2,648 4,600.13 5,435.90
2012 3,084 5,357.52
2013 3,341

So the fitted incremental payments are:

Development Year
Accident Year 0 1 2 3
2010 2,457 1,811.32 775.79 41.62
2011 2,648 1,952.13 835.78
2012 3,084 2,273.55
2013 3,341

If we compare the fitted and the actual incremental payments, we get the following:

29
 Development Year
Accident Year 0 1 2 3
Fitted 2010 2,457 1,811.32 775.79 41.62
Actual 2010 2,457 1,739 773 41
A-F – -72.32 -2.49 -0.62

Fitted 2011 2,648 1,952.13 835.78

Actual 2011 2,648 2,067 846
A-F – 114.87 10.22

Development Year (Cont...)

Accident Year 0 1 2 3
Fitted 2012 3,084 2,273.55
Actual 2012 3,084 2,231
A-F – -42.55

Fitted 2013 3,341

Actual 2013 3,341
A-F –

We can now repeat the process with the inflation-adjusted chain ladder method. Using the inflation
adjusted development factors on the real data amounts for Development Year 0, we get:

Development Year
Accident Year 0 1 2 3
2010 2,860.70 4,864.17 5,663.98 5,705.15
2011 3,019.67 5,134.47 5,978.73
2012 3,182.69 5,411.65
2013 3,341

If we now disaccumulated and compare with the actual incremental (real) values:

Development Year
Accident Year 0 1 2 3
Fitted 2010 2,860.70 2,033.47 799.81 41.16
Actual 2010 2,860.70 1,983.09 797.74 41
A-F – -20.38 -2.08 -0.16

Fitted 2011 3,019.67 2,114.80 844.26

Actual 2011 3,019.67 2,133.14 846
A-F – 18.34 1.74

30
 Development Year (Cont...)
Accident Year 0 1 2 3
Fitted 2012 3,182.69 2,228.96
Actual 2012 3,182.69 2,231
A-F – 2.04

Fitted 2013 3,341

Actual 2013 3,341
A-F –

The inflation adjusted method appears to fit the original data better, given that the residuals are smaller
in this case. However, there are not many categories of data, and it is not clear whether either method
will provide satisfactory results in the future.

16. Consider the following run-off triangle involving numbers of insurance claims of a certain type that have been
settled in previous years by a certain insurance company. In addition, the discount rate is assumed to be 4%
per annum.

Development Year
Accident Year 0 1 2 3
1998 26 14 7 2
1999 35 20 12
2000 15 10
2001 22

Use the basic chain ladder method (with the “average” chain ladder ratios selected as the development factors)
to estimate the number of claims arising from accidents in 2001 that will be settled in 2003.

Solution:
Accumulate figures:

Cumulative Claim Reported

Development Year
Accident Year 0 1 2 3
1998 26 40 47 49
1999 35 55 67
2000 15 25
2001 22

Then we calculate the ratios and obtain the development factors:

Development Year
0:1 1:2 2:3
Mj∗ 1.579 1.200 1.043

31
 Project cumulative claim payments:

Projected Cumulative Claim Payments

Development Year
Accident Year 0 1 2 3
1998
1999 69.851
2000 30.000 31.277
2001 34.737 41.684 43.458

Hence, the estimated number of claims arising from accidents in 2001 that will be settled in 2003 is
41.684 − 34.737 = 6.947. After 4% p.a. discounting adjustment, the estimate turns out to be 6.550.

17. A colleague of yours has just performed a series of PPCI analyses of general insurance claims data. In the
process of performing these analyses she produced the following tables of inflation-adjusted PPCIs.

Portfolio I
Development Year
Accident Year 0 1 2 3 4 5 6
2000 250 615 490 190 100 65 10
2001 263 630 500 200 100 50
2002 241 617 510 210 85
2003 500 325 290 260
2004 532 302 275
2005 495 335
2006 552

Portfolio II
Development Year
Accident Year 0 1 2 3 4 5 6
2000 260 615 490 190 100 79 50
2001 270 630 500 200 100 81
2002 280 650 510 260 90
2003 490 580 270 200
2004 760 280 220
2005 920 240
2006 1,050

Portfolio III
Development Year
Accident Year 0 1 2 3 4 5 6
2000 260 615 490 190 100 80 50
2001 270 630 500 200 100 75
2002 280 650 510 260 90
2003 290 580 505 200
2004 250 620 500
2005 275 645
2006 295

32
 For each of the portfolios, calculate the all-year average PPCI for each development year and the (most recent)
3-year average PPCI for each development year.

Solution:
The results are shown in the following tables:

Portfolio I
Development Year
0 1 2 3 4 5 6
Average (All) 405 471 413 215 95 58 10
Average (3) 526 321 358 223 95 - -
Average (1) 552 335 275 260 85 50 10

Portfolio II
Development Year
0 1 2 3 4 5 6
Average (All) 576 499 398 213 97 80 50
Average (3) 910 367 333 220 97 - -
Average (1) 1050 240 220 200 90 81 50

Portfolio III
Development Year
0 1 2 3 4 5 6
Average (All) 274 623 501 213 97 78 50
Average (3) 273 615 505 220 97 - -
Average (1) 295 645 500 200 90 75 50

18. [RM2 2020 Final Exam Q3]

The tables below show the total amount of the claim payments made in each development year for a portfolio
of General Insurance policies. The projected amounts are calculated using the inflation adjusted chain ladder
method (the final output after inflating and discounting). Unfortunately, some of the entries (indicated
by letters) are illegible:

Claims Payments ($’000)

Development Year
Accident Year 0 1 2 3
2017 50.2 43.8 x 12.3
2018 58.9 y 32.1
2019 z 63.1
2020 56.3

Projected Claims Payments ($’000)

Development Year
Accident Year 0 1 2 3
2017
2018 a
2019 b c
2020 45.6 23.1 12.2

33
The past inflation rates are:
Year Inflation Rate
2018 5.1%
2019 7.3%
2020 5.4%

The discount rate is 4% pa and the future inflation rate is constant at 5% pa.
(a) Calculate the claim development factors, Mj∗ .
(b) Calculate the values of x, y and z.
(c) Calculate the values of a, b and c.

Solution:
(a) With the historical claim payments you need to inflate the 2020 accident year payment (no inflation
required). The projected payments for the 2020 accident year should be deflated and discounted.
From these, we can calculate cumulative projected payments. The resulting projected payments are
1 2 3
2020 100.6 122.8 134.4
Given the cumulative projected payments, the development factors are the ratio of the consecutive
years’ cumulative claims.
0:1 1:2 2:3
Mj∗ 1.7867 1.2209 1.0947

(b) The factor relating development years 2 and 3 is:

121.5 + x × 1.054
= 1.0947.
109.2 + x × 1.054
Solving this we get: x = 19.6802.
The factor relating development years 1 and 2 is:
207.9 + x × 1.054 + y × 1.054
= 1.2209
175.8 + y × 1.054
Substituting x and solving for y we get: y = 60.1314.
The factor relating development years 0 and 1 is:
238.9 + y × 1.054 + z × 1.054
= 1.7867
126.3 + z × 1.054
Substituting y and solving for z we get: z = 92.4750.
(c) Noting the inflation and discounting step for projected payments.
Equation for a:
(98.7 + y × 1.054) × (1.0947 − 1)
a= = 15.8
(1.040.5 /1.051 )
Equation for b:
(63.1 + z × 1.054) × (1.2209 − 1)
b= = 36.5
(1.040.5 /1.051 )
Equation for c:
(63.1 + z × 1.054 + b × 1.040.5 /1.051 ) × (1.0947 − 1)
c= = 19.3
(1.041.5 /1.052 )

34
19. [RM2 2021 Final Exam Q5]
The tables below show the cumulative claim payments made in each development year for a portfolio of
General Insurance policies.

Claims Payments
Development Year
Accident Year 0 1 2 3
2018 101 40 4 28
2019 100 61 4
2020 120 10
2021 136

You can assume that claims are fully run-off after 3 years. Assume the discount rate is 5%. The premiums
received for each year from 2018 to 2021 are 195, 200, 190 and 196 respectively.
(a) Calculate the reserve required at the end of 2021 using the basic chain ladder method.
(b) Calculate the reserve required at the end of 2021 using the Bornhuetter-Ferguson method.
(c) An actuary is considering modelling the future claims assuming that individual development factors are
lognormally distributed with the following parameters:
Development Year
Parameter 0:1 1:2 2:3
µ 0.192352 0.045259 0.001427
σ 0.032148 0.055606 0.056853
If the individual development factors are lognormally distributed, show that the cumulative development
factor (to calculate the ultimate claims) is also lognormally distributed.
(d) Calculate a 99% upper confidence limit for the outstanding claims relating to the 2021 accident year.
(You don’t need to spread it across the different development years or discount them.)

Solution:
(a) First, accumulate and calculate ratios:

0:1 1:2 2:3

Mj∗ 1.345794393 1.026490066 1.193103448

Then project, disaccumulate and discount.

Claims Payments
Development Year
Accident Year 0 1 2 3
2018
2019 31.09
2020 3.36 23.95
2021 45.89 4.51 32.11

Total outstanding reserve: 140.9194.

(b) Estimated loss ratio = 173
195 = 0.8872.
Then calculate Fj :

35
 0:1 1:2 2:3
Mj∗ 1.345794 1.02649 1.193103
Xj 1.648206 1.224709 1.193103
Fj 0.39328 0.183479 0.16185

Calculate OSi

Accident Year Pi Ri OSi

2019 200 0.887179 28.71795
2020 190 0.887179 30.92804
2021 196 0.887179 68.38632

Spread and discount:

Claims Payments
Development Year
Accident Year 0 1 2 3
2018
2019 28.03
2020 3.56 25.36
2021 35.60 3.50 24.91

Total outstanding reserve: 120.9507.

(c) Let Rj,j+1 denote the development factor between development years j and j + 1.
Given Rj,j+1 ∼ LN (µi , σi2 ), for i = 0, 1, 2.
Then we have log (Rj,j+1 ) ∼ N (µi , σi2 ).
This implies:

log (Rj,j+1 , Rj+1,j+2 ) = log (Rj,j+1 ) + log (Rj+1,j+2 ) ∼ N µi + µi+1 , σi2 + σi+1
2

That is the product of lognormally distributed variables is also distributed lognormally (through the
additive property of normally distributed random variables). From this property, the cumulative de-
velopment factor, which is a product of individual development factors should also have a lognormal
distribution.
(d) The cumulative development factor (to ultimate) for AY2023 is lognormally distributed with param-
eters: µ = 0.239038, σ = 0.085778.
Hence, a 99% upper confidence limit for the cumulative development factor is

eµ+σ×2.3263 = 1.5505.

A 99% upper confidence limit for total claims is

1.5505 × 136 = 210.8699.

So, an upper 99% confidence limit for outstanding claims is 210.8699 − 136 = 74.8699.

20. (2020 RM2 Assignment2 Q3)

36
 You are an analyst for a general insurance company assessing an insurance portfolio. You are required to
assess the reserves required for the portfolio. You are provided with the claim payments triangle and past
inflation experience in the attached spreadsheet – Q20 (Practice_Set4).xlsx. You can assume that future
claim inflation is 6% pa. The discount rate is 4.5% pa.
(a) Calculate the total reserves that need to be held for this portfolio under the Basic Chain Ladder method.
State any assumptions you make.
(b) Calculate the total reserves that need to be held for this portfolio under the Inflation Adjusted Chain
Ladder method. State any assumptions you make.
(c) Discuss, with appropriate numerical illustration, which of the two methods would you recommend be
used for setting reserves for this portfolio.

Solution: See attached spreadsheet Q20-Sol (Practice_Set4).xlsx.

(a) See worksheet ‘part a’. The total outstanding reserves is $2,809,652.
BCL method assumptions:
• There is a consistent delay pattern in the payment of claims between accident years.
• There is enough experience to date in the most recent accident years to allow development of
that experience to estimate outstanding claims.
• The first accident year is fully run off.
• Claims in each development year are a constant proportion (in monetary terms) of total claims
in each accident year.
• Inflation is not allowed for explicitly, rather it is allowed for implicitly as a weighted average of
past inflation.

(b) See worksheet ‘part b’. The total outstanding reserves is $2,839,645.
IACL method assumptions:
• For each accident year, the amount of claims paid, in real terms, in each development year,
is a constant proportion of the total claims, in real terms, from that accident year.
• The first accident year is fully run off.
• Inflation is allowed for explicitly and we assume that both the past and future inflation as-
sumptions are correct.
(c) See worksheet ‘part c’. Based on the actual-fitted method, both the methods fit the data equally
well (or poorly) with a few of the cells giving ’large’ errors (>30).
Based on the mean square error method (square root of the sum of the square differences), the two
methods give very similar MSE (BCL = 126.53; IACL - 126.69). Hence, either of the two methods
could be recommended for use.

21. (2020 RM2 Assignment2 Q4)

The attached spreadsheet – Q21 (Practice_Set4).xlsx provides the claim payments triangle, the cumulative
claims reported and the inflation rate for a particular insurance portfolio.
You have also been given the following additional information:
• Future inflation from mid-2006 is estimated to be 5% p.a..
• The risk-free rate of return is estimated to be 4% p.a.
• $30,000,000 of premiums were earned in 2019.

37
 • At the end of 2019, case estimates for outstanding payments arising from accidents occurring in 2019
totaled $2,000,000.
• The ultimate loss ratio has been estimated to be 90% for all accident years
(a) Calculate the outstanding claims reserve at the end of 2019 using the inflation-adjusted PPCI method.
State any assumptions you make.
(b) Discuss the appropriateness of the average PPCIs calculated in (a) above as representative of future claims
behaviour for this portfolio. If not, suggest more appropriate representative PPCI for each development
year, and the new reserve required. Justify your answer.
(c) Using the Bornhuetter-Ferguson method, estimate the inflation-adjusted amount of claims that will be
paid in 2022 relating to accidents that have already occurred.

Solution: See attached spreadsheet Q21-Sol (Practice_Set4).xlsx.

(a) See worksheet ‘part a’. The total outstanding reserves is $1,801,693.
PPCI method assumptions:
• For each accident year, both the number and average amount of claims relating to each devel-
opment year are constant proportions of the totals from that accident year.
• The first accident year is fully run off.
• Inflation is allowed for explicitly and we assume that both the past and future inflation as-
sumptions are correct.
(b) There is an increasing trend in the PPCIs for development year 0, and the later PPCIs for devel-
opment year 1 are considerably greater than the earliest PPCI for this development year. Based
on these observations, the average PPCIs seem too low to be representative for these development
years.
There is not enough information about development years 2 or 3 to determine whether there are
any trends in the PPCIs for these years, so the average PPCIs can be seen to be best estimates for
these years.
See worksheet ‘part b’ for the PPCIs.
The revised representative PPCI for development year 0 is the PPCI for 2019/dev year 0 plus a
trend component (a further adjustment for the increasing trend). The revised representative PPCI
for development year 1 is the average of the latter two PPCIs for development year 1.
The updated reserve is $2,058,534.
(c) See worksheet ‘part c’. The total amount payable in 2022 is $1,087,944.

22. Mack suggests a test of Calendar Year Effects based on classifying the development factors in each development
year as high or low, relative to the other values in the same DY. A value of fi,j that is higher than the median
value for DY j is labelled L (for large); a value below the median is labelled S (small). Values on the median
are disregarded. We then consider the frequency of L and S development factors across each anti-diagonal.
If there is no calendar year effect, roughly half of the development factors in each calendar year should be
large, and half small, and the numbers should be binomially distributed, with probability parameter 0.5. Let
Sk denote the number of ‘S’ labels in calendar year k, and let Lk denote the number of ‘L’ labels in calendar
year k. We exclude calendar year 0 which has only one value. We test the null hypothesis, that calendar
years are independent, by considering the distribution of the smaller of Sk and Lk each year. This is not
binomially distributed under the null hypothesis, but its mean and variance can be derived from the binomial
distribution. The test procedure is as follows:
1. For k = 1, 2, . . . , I − 1, find Sk , Lk , nk = Sk + Lk , and mk = ⌊(nk − 1) /2⌋. Under the null hypothesis of
no calendar year effect, both Lk and Sk have a bin (nk , 0.5) distribution.

38
2. Let Zk = min (Sk , Lk ) for k = 1, 2, . . . , I − 1. Using the binomial distribution, it can be shown that
 
nk nk − 1 nk
E [Zk ] = − and
2 mk 2nk
 
nk (nk − 1) nk − 1 nk (nk − 1) 2
Var [Zk ] = − + E [Zk ] − E [Zk ]
4 mk 2nk
PI−1
3. The test statistic is Z = k=1 Zk . We assume that, approximately,

I−1
X I−1
X
Z ∼ N (E[Z], Var[Z]) where E[Z] = E [Zk ] and Var[Z] = Var [Zk ]
k=1 k=1

(using the null hypothesis that calendar years are independent).

Then the p - value for the 2 -sided test is
!!
|Z − E[Z]|
p=2 1−Φ p .
Var[Z]
Apply this method to test the second data set in the Excel file “Lecture Examples.xlxs”.

Solution: The L-S triangle is shown in the table below.

Development Year j
AY i 0 1 2 3 4 5 6 7 8
0 L L S L S L ∗ L ∗
1 L L S S S S L S
2 L L ∗ L ∗ L S
3 S L L L L S
4 S S L S L
5 S S S S
6 L S L
7 ∗ S
8 S

We review each anti-diagonal, starting from the second, to assess the S-L distribution
 by calendar
 year.
A
|14−12.6875|
summary of the results is given in the table below. The p-value for the test is p = 2 1 − Φ √
3.6621
=
0.4928, indicating that there is no significant evidence of a calendar year effect in this data.

CY, k Sk Lk Zk nk mk E [Zk ] Var [Zk ]

1 0 2 0 2 0 0.5000 0.2500
2 1 2 1 3 1 0.7500 0.1875
3 2 2 2 4 1 1.2500 0.4375
4 3 1 1 4 1 1.2500 0.4375
5 3 3 3 6 2 2.0625 0.6211
6 2 3 2 5 2 1.5625 0.3711
7 3 4 3 7 3 2.4063 0.5537
8 6 2 2 8 3 2.9063 0.8037
Total 20 19 14 39 12.6875 3.6621

39