1985 General Insurance Convention

UNEXPIRED RISK RESERVE (URR) Now called "Additional Amount for Unexpired Risks"

The only formula I haue euer seen for calculating the URR (and the one applied by the DTI) basically requires that the total premium reserves at the end of a year should be:- (Claims ratio for the year) x (Unearned premiums at end of the year). If the result exceeds the Unearned Premium Reserve (UPR) then the difference must be set up as URR. This seems unobjectionable until one analyses the components of each term. Taking the claims ratio as (Claims arising during the year) : (Earned premiums for the year) it is clear that part of the earned premiums (EPs), with corresponding claims, comes from the UPR brought forward from the previous year. If that previous year's premiums were adequate to support the claims (so that the UPR then needed no supplementing by URR) but the premiums written in the current year (which usually provide roughly half the EPs for the year) are insufficient to carry the resulting claims then the claims ratio for the year will be about half-way between the low ratio appropriate to the previous year's carry-forward and the high ratio appropriate to the earned portion of the current year's premiums. This is an inadequate multiplier for the current year's unearned premiums (UPs) to which the undiluted high ratio ought to be applied. Reverse the situation so that the previous year's premiums were too low, the current year's rates being satisfactory, and the formula can produce a URR when it is not needed. The implicit assumption that the claims ratio appropriate to the EPs (even after eliminating the effects of the carry-forward from the previous year) is a suitable multiplier for the UPs can also be a mistake. If inadequate premiums are charged for only part of a year their contribution to the year's EPs will vary according to the months in which they were in use whilst their effects on the UPs will vary in the opposite direction (e.g. if the period is early in the year they will make a substantial contribution to the EPs but little to the UPs, and vice versa if the period is late) so the formula is liable to produce results bearing little resemblance to the real need. In a business such as Motor, where rates need frequent revision, it would not be difficult to find a rating series to be inadequate for the whole period it was in force; even more likely in an inflationary era would be to find rates inadequate for business effected towards the end of the life of a rating series. As a check on these diagnoses, and to get some measure of their effects, I did a few trial calculations with highly idealised examples and these, with my assumptions, are shown in the Appendix. The trials confirm expectations and the presence of both overstatements and understatements of the true reserve in the calculated figures should provide food for thought to both offices and DTI. It will be seen from lines 20/21 of the calculations that an unnecessary URR can be eliminated by adding the previous year's URR to the EPs for the year (an obvious adjustment once it has been pointed out) but there remains a problem in determining the proper value of this URR. Once the concept of inadequate rates applying for only part of a year has been introduced there appear to be considerable difficulties in the practical calculation of an accurate amount even for the offices, and much more so for the supervisory authorities. Cont/..

It seemed that there ought to be some improvement if the UPR brought forward and the corresponding claims were taken out of the formula, but lines 22/24 of my calculations show that this idea produces distortions that could be worse than before . This approach is another way of eliminating the unwanted URR derived from the previous year's inadequate rates so would be attractive if its errors could be avoided. I have established that the calculation of the URR poses problems, but I have yet to find a satisfactory practical solution. Offices no doubt will have some idea of when their rates were inadequate and could, therefore, estimate their contribution to EPs and UPs; they might, however, find it difficult to determine the claims arising from these EPs. Knowing how much of the UPR is derived from inadequate rates suggests that the URR could be calculated by applying a percentage deficiency factor, but the calculation of this factor also requires that the appropriate claims be known and there would then remain a problem to determine how much profit there is in the remainder of the UPR for offset against the deficiency. The problems are far worse for the supervisory authorities with the limited data available to them (particularly the claims for a year being one individed figure) so any solution they can produce is unlikely to yield more than a very rough approximation; we should, however, hope that they will not expect a URR when it is unnecessary. The rather complicated formula that the DTI applies in practice (recognising that real business covers more than the straightforward cases I have considered in making the above analysis) contains features that are liable to produce distortions even if the basic principle was correct. Four points that merit comment are:(a) The premiums employed in the formula are a mixture of items, some taken direct from the returns and hence as calculated by the office, some taken from the Department's calculations of unearned premiums (using the 24ths method) as a check on the accuracy of the office's work. Where there is a biassed difference between the office's and the Department's calculations (there is plenty of scope for this in the 24ths method) the selection of the terms going into the formula seems designed to ensure that the distortions in the answer are maximised. Both the premium and claims figures in the formula are net of reinsurance. Where the reinsurance is arranged as a proportion of the original contract (e.g. Surplus or Quota Share) I presume offices will analyse the premiums by months in the same way as the gross premiums so that calculations of UPs (whatever the method) will be properly comparable between the gross and reinsurance premiums. An Excess of Loss treaty, however, is divorced from any original policy and is arranged by many offices to cover the same period as their financial year. If the premium for the treaty is due at the beginning of the year, and shown in Month 1 of the premium analysis in the statutory returns, there will be no UP for this contract at the end of the year, the whole premium being offset against the gross EPs. (The 24ths method will Hence as far as this treaty is produce l/24th as the UP). concerned the UPs (in the numerator in the formula) will be

(b)

Cont/..

gross and the EPs (in the denominator) net. There is obvious scope for similar, if reduced, distortion if the X/L premiums are paid quarterly in advance. No doubt it could be arranged for such premiums to be spread over the months before applying the formula but an easier answer would be to employ only gross premiums leaving the nettingdown for reinsurance to be dealt with in the claims figures alone, the answer should be about the same. (c) The claims in the formula are the office's estimate of their cost plus the costs of settling them, also as estimated by the office. Hence the weaker the office's standards for claims reserves the smaller the URR emerging, which seems the opposite of what the supervisory authorities should be aiming for. The formula adds to the result of these calculations a term (of an amount equal to half the year's management expenses) said to be the DTI's estimate of the costs of winding-up the office. Whilst it seems perfectly reasonable for the Department to check that the reserves are adequate to look after such a contingency I can see no legal or logical justification for making it part of the URR.

(d)

W. ROWLANDSON 8.7.85

APPENDIX

Tests of the Principles Underlying the Calculation of the Unexpired Risk Reserve as applied by the Department of Trade and Industry

These tests are intended purely for the purpose of investigating the mathematics of the method so all extraneous influences have been eliminated. Thus it is assumed that:(a) (b) There is no commission and expenses The premium income is distributed evenly throughout each year and the 24ths method of calculating unearned premiums is accurate. The claims occur evenly throughout any period, are reported immediately and the amounts are known accurately at once. The policies are all standard 12-month contracts, with no movements during the year. There is no reinsurance and no inflation.

(c)

(d)

(e)

For the examples an office is assumed to have been writing for a long time a constant amount of business for 2,400 of annual premium each month, on which the claims for the year also amount to a steady 2,400. During year Y it reduced the rates of premium by 20%, then restored them to their former level 4 months later. During the period of the reduction the number of policies did not change and the claims continued at their normal level of 2,400 per annum for each month's business. The period of the reduction was given varying commencing dates, avoiding any that would cause the period to spread into year Y + 1. Each month's normal business produces earned premiums accruing at 200 per month against which arise 200 of claims. Whilst this situation continues there is no need for any URR. The reduced premiums produce a monthly accrual of 160 so there is a monthly deficiency of 40 against the claims, providing a direct measure for calculating the URR required. At the end of year Y + 1 all business then in force is back to normal terms so again no URR is needed. The data set out overleaf can now be deduced.

TRIAL Commencing True URR Date needed for of at Period 31/12/Y

CALCULATIONS 1 January 1 1 Nay 960 1 July 1,280 1 Sept. 1,600

1. 2.

March
640

320

Experience

Year Y 14,400 26,880

3. 4. 5. 6. 7. 8. 9. 10.

UPR (all

brought forward at l/l/Y Normal)

Written Premiums (WPs)

Year Y

carried forward at 31/12/Y:Normal Premiums Reduced Premiums Total 12,800

1,280
14,080

11,200 2,560 13,760 13,120 27,520 28,800

9,600 3,840 13,440 13,440 27,840 28,800

8,000 5,120 13,120 13,760 28,160 28,800

6,400 6,400 12,800 14,080 28,480 28,800

EPs

from EPs in

WPs in

in

Year Y (= 4-7)
Y (= 3+8)

12,800 27,200 28,800

Total Claims

Year

Year Y
of URR at 31/12/Y

Calculation

11. 12.

Claims Deduct

Ratio UPR

x UPR (= 7)

(=109x7

14,908

14,400 640

13,903
463

13,418 298

12,944 144

828

Experience for Year Y+1 13. 14. 15. 16. 17. UPR brought forward at l/l/Y + 1
WPs UPR in

(= 7)

14,080
28,800 14,400 28,480 28,800

13,760 28,800 14,400 28,160 28,800

13,440 28,800 14,400 27,840 28,800

13,120 28,800 14,400 27,520 28,800

12,800
28,800 14,400 27,200 28,800

Year Y+l (all Normal)

carried forward at 31/12/Y + 1 (all Normal) (= 13+14-15)

Total EPs in Year Y+l

Calculation

31/12/Y

1 14,562 162 14,727 327 14,897 497 15,070 670 15,247 847

18. 19.

Claims Deduct Revision

Ratio UPR of

x UPR (= 15) URR

(= 1716x15)

calculation

at

31/12/Y

+ 1 14,400 0 14,400 0 14,400 0 14,400 0 14,400 0

20. 21.

Claims + (EPs + True URR

(= 17(16+2) x 15)

31/12/Y)

calculation

31/12/Y

excluding in Year in 7 Y

carry-forward from Year Y-l

22. 23. 24.

Claims in Year (half of 10) Claims Deduct in UPR

Y from WPs
in

22 EPs (= 7)

8 x UPS

15,840 1,760

13,091 291