Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 1 Pricing motor quota share treaties Technical publishing Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 1 Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 3 Contents Foreword 4 1 Introduction 5 2 Expected loss ratio 7 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 Motor third party liability (TPL) Estimating the future claims frequency Proportion of bodily injury claims The average property damage claim The average bodily injury claim Discounting The average premium Fluctuation loading Expense loading 8 8 9 10 12 14 16 18 19 4 4.1 4.2 4.3 4.4 Motor own damage (MOD) Covered loss events Claims frequency, average claim, discounting and average premium Allowing for natural hazards The loadings for fluctuation and expenses 20 20 20 21 22 5 The fixed commission 23 Appendix 1: Mathematical deductions A Minimum number of claims required for a frequency estimate B Minimum number of bodily injury claims for the estimate of the proportion of bodily injury claims C Minimum number of claims for the estimate of the expected claim D Variance loading E Variance loading for natural catastrophe claims 24 24 24 24 25 25 Appendix 2: References 26 Appendix 3: Questions for cedents A Questionnaire for motor third party liability (TPL) B Questionnaire for motor own damage (MOD) 26 26 26 Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 4 Foreword The quota share is an older and seemingly simpler form of reinsurance than the excess of loss treaty. While rating of excess of loss treaties has always been a matter for specialists, scant attention has been paid to the pricing of quota share treaties for some time. In regulated markets with universally binding premium rates, this was not really necessary anyway, because the order of the expected loss ratio was known, as was the level of commissions that would make it possible to operate the reinsurance business at a profit over a sufficiently long period. In deregulated markets, however, and in new markets where the basis for the calculation of primary insurance premiums is still uncertain, there is no reliable empirical foundation for pricing quota share treaties. This publication will deal with how to determine the proper commission for motor quota share treaties in such situations. The procedure described was developed by Hans Schmitter and Pamela Hall in the year 2000 from their analysis of data provided by a Swiss Re cedent. This analysis identifies which of the data available to a primary insurer are important to reinsurance pricing at all; these are compiled in the two questionnaires in Appendix 3. The methods derived in this publication for predicting a loss ratio or an investment income are not limited to reinsurance only, but can also be used in primary insurance. The publication is thus intended for reinsurers who have to price reinsurance covers in practice and for primary insurers who require such prognoses for their planning calculations. All calculations described in this publication and needed for pricing can be performed with the aid of the Excel file motorproppricing.xls developed by Pamela Hall, which is programmed in Visual Basic. Copies are available free of charge from the e-mail address [email protected] Thomas Hiltmann Head of Group Product Management Casualty 4 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 5 1 Introduction Motor quota share treaties normally cover motor third party liability (TPL) and motor own damage (MOD) business. Passenger accident insurance may also be included, but it is insignificant in terms of volume and will not be discussed in this publication. All numerical examples used to illustrate the pricing procedure presented here are based on figures provided by a cedent of Swiss Re. Data from the years 1988 to 1999 were available in the year 2000 for pricing treaties covering the year 2001. These years naturally lose their relevance in the course of time and will eventually become out of date. The pricing procedure is, of course, independent of the time period used, ie from 1988 to 2000, in this example. Nevertheless, to avert any suspicion that the method itself may be obsolescent just because it is explained on the basis of out-of-date figures, the years have been renumbered in the text: the year with the oldest data is referred to as year 1, followed by year 2, etc. Thus, data are available for the twelve years 1, 2, …12; the calculations are performed in year 13 and serve as a prognosis for the next year, ie year 14. Unless otherwise specified, all monetary units are in Swiss francs (CHF). Only in pointed instances, as in Table 3, for example, does CHF 1 000 serve as the monetary unit. In the chapter on motor third party liability insurance, the computations are performed in all detail to make them easy to follow. To avoid repetition, the chapter on motor own damage (MOD) insurance does not go into such detail, except where the calculation procedures differ significantly from those used for TPL. In reinsurance, a distinction is sometimes made between the technical premium and the commercial premium. The technical premium is the premium the reinsurer needs to charge in order to carry on its business. The commercial premium is the premium that is actually charged for a specific treaty in a given instance. Depending on market conditions, this may differ considerably from the technical premium. The publication in hand deals only with the technical premium. Every technical reinsurance premium is the sum of the risk premium, a fluctuation loading and an expense loading. The risk premium is the average claims burden expected over the treaty period. In lines of business in which there may be several years between the time at which the premiums are collected and the time at which the claims payments are made – as in motor TPL – a distinction is made between the discounted and non-discounted risk premium. 5 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 6 The discounted risk premium is that amount which, when invested in interestbearing assets, is on average, together with its investment yield, just enough to pay for the claims. The fluctuation loading is that component of the premium which the reinsurer charges for absorbing the fluctuations in the claims burden. In the long term, it remains with the reinsurer as a profit. The expense loading, finally, is the amount from which the expenses associated with the operation of the reinsurance business are paid, that is to say wages, office rentals, taxes, etc. The reinsurance premium for a quota share treaty is never expressed as a straight technical premium, but always indirectly via a commission. The relationship between the technical premium and the commission is as follows: earned premium – technical premium = fixed commission earned premium This publication will not deal with sliding-scale – as opposed to fixed – commissions. 6 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 7 2 Expected loss ratio The ratio between the claims burden and the earned premium is known as the loss ratio. The average of all possible loss ratios of the year for which pricing is to be performed is referred to as the expected loss ratio. This is the ratio between the expected average claims burden – ie the risk premium – and the earned premium. The expected loss ratio is thus: expected loss ratio = risk premium earned premium In this equation, the risk premium can be replaced by the product of the average number of claims and the average claim size. This gives: expected loss ratio = average number of claims average claim size earned premium If the numerator and the denominator of this fraction are divided by the same number, eg the number of policies, its value remains unchanged. Thus, the expected loss ratio can also be expressed as: expected loss ratio = average number of claims average claim size number of policies earned premium number of policies The ratio between the average number of claims and the number of policies is referred to as the claims frequency or frequency for short; the ratio in the denominator – the earned premium divided by the number of policies – is the average premium per policy. Using these two terms, the expected loss ratio can be calculated as: expected loss ratio = claims frequency average claim size average premium In this form, the expected loss ratio can be calculated independent of the premium volume and premium growth. 7 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 8 3 Motor third party liability (TPL) 3.1 Estimating the future claims frequency The future claims frequency is estimated on a statistical basis. Table 1 shows an example compiled from observations from the six latest years available, ie the years 7 to 12. These enable a prognosis of the year for which pricing is to be performed (the “pricing year”). The year in which a loss occurs is called its occurrence year. This is given in the first column in Table 1. The second column, under the heading “Annual risks”, in principle gives the number of insured vehicles. “In principle” because both vehicles leaving the insured portfolio and new vehicles entering it before the end of the year will not be counted in full but only in proportion to the number of days they were actually insured. Normally, not all claims are reported to the insurer before the end of their occurrence year – claims that occur on 31 December, for example. About 8% of all claims are not reported until the year following the occurrence year, which is the first development year. In the second development year, only a few further claims from the original occurrence year are reported. For example, the number of claims reported from occurrence year 9 increases only from 18 345 to 18 383 in the course of its second development year. As of the fourth development year, the insurer knows practically all the claims. The occurrence year itself is sometimes referred to as development year zero. Table1 Occurrence Annual year risks 7 8 9 10 11 12 252 266 255 894 273 711 275 728 272 217 272 668 Reported claims Development year 0 1 17 566 19 006 17 694 19 327 16 763 18 345 17 206 18 641 16 758 18 058 17 346 Frequency 2 19 048 19 369 18 383 18 680 3 19 055 19 375 18 388 4 19 057 19 378 5 19 057 0.0753 0.0755 0.0670 0.0676 0.0663 Not every claim reported ends up being indemnified by the insurance company. Some claims are settled without payment. These are the so-called nil cost claims and may be the case if the policyholder was not liable at all, for example. Or if the policyholder pays a small claim him/herself, so as not to forfeit part of the no-claim bonus. It would be possible to leave the nil cost claims out of compilations of the kind shown in Table 1, but it is easier to keep them in. This leads to an elevated estimate of the claims frequency, but that does not necessarily entail an error in the pricing of the premiums: because if the nil cost claims are also included in the calculation of the average claim size, they result in a lower average. The overstatement of the claims frequency and the understatement of the average claim size cancel each other out. 8 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 9 Practically all claims have been reported at the end of the first development year so that the claims frequencies can be calculated with sufficient accuracy. Figure 1 shows the claims frequencies for the occurrence years 7 to 11, calculated from Table 1 on the basis of the reported claims at the end of the first development year. 10 Figure 1 9 Claims frequency in % 8 7 6 5 4 3 2 1 0 7 Estimate used in pricing Frequency 6.63% 8 9 10 Occurrence year 11 The changes from occurrence year to occurrence year are surprisingly small. A slight decrease in the claims frequency in the course of time can be observed – other markets may occasionally show an increase. If an explanation is found for such a trend and indicates that, for example, a further decrease is to be expected, the estimate for the future frequency will be lower than the last observed value. Normally, however, the last known value will be the most reliable estimate for the future. In our case, this would be 6.63%, the frequency in year 11. Statistics of the kind shown in Table 1 will provide a reliable basis for predicting claims frequencies only if the number of claims observed is large enough. As a rule of thumb, 1 500 claims are sufficient for a reasonable estimate. Claims frequencies estimated on the basis of only a few hundred claims are unreliable. The mathematical reasons for this are explained in Appendix 1. If experience from the past is used to estimate the future claims frequency, this implicitly assumes that the risk composition of the portfolio is going to remain essentially unchanged over time. However, if, for example, a small insurer takes on a large fleet of rental cars from one year to the next, or if it launches a special campaign to establish a profile as a taxi insurer, this assumption becomes questionable. The effects of such changes in the risk composition of the portfolio must always be assessed separately. 3.2 Proportion of bodily injury claims Just as the change in the claims frequency is only slight from occurrence year to occurrence year, so, too, is the change in the proportion of bodily injury claims over longer periods. A bodily injury claim is defined as any motor TPL claim in which at least one person is killed or injured. Thus, the bodily injury claim consists not only of the payments and reserves for fatalities or injured persons, but also the payments and reserves for property damage due to the same accident, which the motor TPL insurance has to indemnify. 9 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 10 Table 2 shows the number of bodily injury claims for the occurrence years 7 to 11, as reported from development year to development year. After the first development year, the number of bodily injury claims increases just as insignificantly as the overall number of reported claims (Table 1). Table 2 Occurrence year 7 8 9 10 11 Reported bodily injury claims Development year 0 1 1 713 1 880 1 733 1 899 1 572 1 738 1 607 1 793 1 676 1 856 2 1 889 1 911 1 749 1 805 3 1 892 1 913 1 750 4 1 893 1 915 5 1 893 Figure 2 shows the proportion of the bodily injury claims relative to the total of all reported claims from Table 1 at the end of the first development year. 12 Proportion of bodily injury claims in % Figure 2 10 8 6 4 2 0 7 8 9 10 Occurrence year 11 Figure 2 shows no clear trend. Nor is there any real reason why the proportion of bodily injury claims should systematically increase or decrease from year to year. For this reason, it is expedient to estimate the future figure on the basis of the average of the last few years observed. The average of the years 7 to 11 is 9.8%. Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Like the estimate of the claims frequency, the estimate of the proportion of bodily injury claims is again unreliable, if it is based on too small a number of observations. As a rule of thumb: 1 400 bodily injury claims are sufficient for a reasonable estimate (the mathematical derivation is given in Appendix 1). 3.3 The average property damage claim Table 3 shows the sum of the paid and reserved property damage claims from the occurrence years 7 to 12, the unit being CHF 1 000. The ultimate claims burden in the last column is estimated with the aid of incurred but not reported (IBNR) techniques, in the present example by means of the chain ladder method. Details of the most commonly used IBNR techniques are given in the references provided in Appendix 2, (Boulter and Grubbs). 10 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 Table 3 10:18 Uhr Seite 11 Occurence Claims burden year Development year 0 1 7 49 557 44 229 8 49 315 44 211 9 43 262 41 018 10 42 964 39 955 11 41 178 39 941 12 43 143 2 42 515 40 873 39 493 39 048 3 41 154 40 370 38 283 4 41 216 39 986 5 41 344 Estimated ultimate claims burden 41 344 40 110 38 250 38 039 37 190 37 167 Property damage claims are settled relatively quickly. By the end of the second development year, the difference between the known claims burdens and the ultimate claims burdens is already down to only a few percent, and after the fourth development year practically all the property damage claims have been paid. Representations such as Tables 1, 2 and 3 are called run-off triangles: they show how a quantity (eg the number of reported claims or the claims burden) originating in a given occurrence year changes from development year to development year. Anybody who is familiar with such triangles from pricing excess of loss treaties normally expects the claims burden to increase from development year to development year. In Table 3, by contrast, it declines, which is nothing unusual in primary insurance, however. The average property damage claim is equal to the estimated ultimate claims burden divided by the number of property damage claims. The number of property damage claims is the difference between the total number of reported claims (Table 1) and the number of bodily injury claims (Table 2). Since the number of reported claims only increases slightly after the first development year, it can be calculated with sufficient accuracy from the data of the first development year. Thus, for occurrence year 10, for example, the number of property damage claims obtained is 16 848 (18 641 – 1 793 = 16 848) and, accordingly, the average property damage claim is 2 258 (38 039 000 / 16 848 = 2 258). Figure 3 shows the average property damage claims for the occurrence years 7 to 11. Average property damage claim 2 500 Figure 3 2 000 1 500 1 000 500 0 7 Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 8 9 10 Occurrence year 11 The average property damage claim of the pricing year, year 14, must now be estimated on the basis of the five latest averages observed. The average in occurrence year 7, which is the oldest and thus the least representative for the future estimate, is somewhat higher than the others. The averages in the later occurrence years, by contrast, lie practically on a horizontal line. A statistical analysis likewise shows no significant trend. Under these circumstances, the most recent average (that in year 11), yields the best prognosis for the future. This figure is 2 295. 11 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 12 As for the prediction of a claims frequency or of the proportion of bodily injury claims, there is also a rule of thumb for forecasting the average property damage claim: an average based on at least 8 000 property damage claims is reliable enough (the mathematical reasoning is given in Appendix 1). If possible, the average property damage claim of the individual insurer should be used for pricing, rather than a market average. This is because the amount of the property damage claims and thus their average depends in part on claims settlement practice and therefore differs from one insurer to the next. It must be based on a portfolio of around 150 000 annual risks: the expected number of claims is then 9 000 to 10 500 at a frequency of 6% to 7%. If the property damage claims account for around 90% of all claims, 8 100 to 9 450 property damage claims can be expected. 3.4 The average bodily injury claim By analogy with Table 3, Table 4 shows the bodily injury claims burden from the occurrence years 1 to 12. The monetary unit is again CHF 1 000. In this case, too, the ultimate claims burdens in the last column are estimated with the aid of the chain ladder method. Table 4 Occurrence Claims burden year Development year 0 1 1 30 995 39 325 2 37 713 46 778 3 39 214 47 350 4 40 880 51 485 5 44 025 54 152 6 41 741 51 666 7 40 841 51 836 8 48 770 57 404 9 50 687 58 297 10 54 184 64 170 11 58 829 70 328 12 60 587 2 41 933 47 860 50 974 55 328 57 151 57 737 55 032 62 267 58 190 65 467 3 42 208 49 939 53 669 57 270 61 659 60 880 56 139 64 767 58 595 4 44 498 51 897 55 342 61 742 61 489 61 908 56 084 65 237 5 44 808 52 108 60 270 62 456 61 876 62 921 56 492 6 45 928 54 219 60 551 64 928 63 014 62 647 7 46 271 55 464 59 961 64 913 63 387 8 47 546 56 029 61 258 66 130 9 47 569 56 045 62 298 10 47 866 57 336 Estimated ultimate 11 claims burden 47 827 47 827 57 289 63 201 67 528 65 971 65 507 60 245 70 982 65 421 75 897 86 132 90 158 A comparison with Table 3 shows two fundamental differences between property and bodily injury claims: bodily injury claims take longer to run off than property damage claims. A runoff triangle showing not more than five development years (such as Table 3), would not be sufficient for predicting the average bodily injury claim. This normally requires ten or more development years; the bodily injury claims burden normally increases from development year to development year, whereas the property damage claims burden tends to be overestimated in the early years (though, depending on claims reserving practice, it is also possible for the property damage claims burden to increase as run-off progresses). Dividing the estimated ultimate claims burden by the number of bodily injury claims gives the average bodily injury claim. The number of bodily injury claims is shown in Table 5. Figure 4 shows the average bodily injury claim of occurrence years 1 to 11. 12 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc Table 5 25.02.2003 10:18 Uhr Seite 13 Occurrence year Bodily injury claims Development year 0 1 953 1 959 1 939 1 854 1 769 1 658 1 713 1 733 1 572 1 607 1 676 1 2 3 4 5 6 7 8 9 10 11 1 2 120 2 146 2 153 2 051 1 912 1 797 1 880 1 899 1 738 1 793 1 856 50 000 Figure 4 Average bodily injury claim 45 000 40 000 35 000 30 000 25 000 20 000 15 000 10 000 5 000 0 1 3 5 7 Occurrence year 9 11 Unlike the average property damage claim, the average bodily injury claim shows a distinct upward trend. On the basis of Figure 4, the average bodily injury claim of the pricing year, year 14, would be forecast at around 50 000. The trend line in Figure 4 is calculated by simple linear regression of the average bodily injury claims on the occurrence year, as explained in any statistics textbook. The trend value for the pricing year, year 14, would give a good prognosis for the future, if the number of claims on which the computation of the trend is based is large enough. As the rule of thumb elaborated in Appendix 1 shows, the computation base should be around 21 000 bodily injury claims per occurrence year. In our example, with only some 2 000 observations per occurrence year, the trend value obtained for the pricing year is not a reliable prognosis. In this case, it would be better to calculate the average bodily injury claim per occurrence year from the data of several companies together. In Switzerland, for example, where roughly 22 000 bodily injury claims are reported every year, the average of the entire market would have to be used. For further calculations, the trend value of the market as a whole for occurrence year 14 is assumed to be 50 000. 13 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 Average bodily injury claim 50 454 10:18 Uhr Seite 14 The question remains as to how representative the average bodily injury claims of years 1 to 11 are for the future. Most markets have not experienced any catastrophic claims on the scale of the Gotthard, Mont Blanc or Tauern tunnel losses in any of the old occurrence years. Nevertheless, such catastrophes can occur and pricing must allow for them. If we assume, for example, that a loss event triggering claims totalling CHF 100 million occurs once every ten years in Switzerland, the estimate of the average bodily injury claim increases from 50 000 to 50 454 as the following calculation shows: 50 000 22 000 10 + 100 000 000 = 50 454 22 000 10 + 1 This figure is used as the basis for the following calculations. 3.5 Discounting To calculate the average motor TPL claim from the average property damage claim and the average bodily injury claim, the two types of claim have to be weighted and added. In line with Section 3.2, the weights in our example are 90.2% for the property damage claims and 9.8% for the bodily injury claims. The estimate of the average claim of the pricing year is thus: 0.902 2 295 + 0.098 50 454 = 7 015 However, the amount of 7 015 per claim need not be available in the pricing year; it is sufficient to have a smaller amount which later, in the course of run off, is incremented by the return on its investment to attain 7 015 at the time of payment. This smaller amount is called the discounted average claim. In the following, we use the example of the average bodily injury claim to show how the discounted value is calculated. Table 6 shows the paid bodily injury claims burdens for occurrence years 1 to 12. As in Table 4, the unit is CHF 1 000. Table 6 Occurrence Paid bodily injury claims year Development year 0 1 1 5 839 12 289 2 6 721 15 461 3 7 067 15 449 4 7 673 17 099 5 7 006 15 019 6 7 002 16 253 7 7 135 14 837 8 6 985 15 076 9 6 625 14 370 10 6 635 15 242 11 7 506 15 673 12 7 421 2 16 343 20 071 20 300 22 673 20 674 21 886 19 176 20 734 18 812 20 263 3 19 625 24 408 23 864 27 484 25 019 26 197 23 712 24 855 22 504 14 4 22 616 28 027 27 674 31 377 29 424 30 425 27 571 29 371 5 24 891 31 321 30 677 35 654 33 857 35 691 31 858 6 27 482 34 920 37 419 38 565 37 984 40 063 Swiss Re: Pricing motor quota share treaties 7 30 138 38 515 41 497 42 784 42 950 8 33 775 41 202 44 058 45 861 9 34 902 43 373 47 227 10 36 986 45 781 11 38 457 Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 15 Table 7 Occurrence year 1 2 3 4 5 6 7 8 9 10 11 12 Average: Part of paid claims in ultimate claims burden Development year 0 1 2 3 12.2% 25.7% 34.2% 41.0% 11.7% 27.0% 35.0% 42.6% 11.2% 24.4% 32.1% 37.8% 11.4% 25.3% 33.6% 40.7% 10.6% 22.8% 31.3% 37.9% 10.7% 24.8% 33.4% 40.0% 11.8% 24.6% 31.8% 39.4% 9.8% 21.2% 29.2% 35.0% 10.1% 22.0% 28.8% 34.4% 8.7% 20.1% 26.7% 8.7% 18.2% 8.2% 10.4% 23.3% 31.6% 38.8% 4 47.3% 48.9% 43.8% 46.5% 44.6% 46.4% 45.8% 41.4% 5 52.0% 54.7% 48.5% 52.8% 51.3% 54.5% 52.9% 6 57.5% 61.0% 59.2% 57.1% 57.6% 61.2% 7 63.0% 67.2% 65.7% 63.4% 65.1% 8 70.6% 71.9% 69.7% 67.9% 9 73.0% 75.7% 74.7% 10 77.3% 79.9% 11 80.4% 45.6% 52.4% 58.9% 64.9% 70.0% 74.5% 78.6% 80.4% Table 7 is derived by dividing the paid claims by the associated estimated ultimate claims burdens (in the last column of Table 4). On the assumption that all payments are made mid-year and the reserve of the 11th development year is paid in the middle of the 12th development year, the average run-off period in years prior to payment is obtained as follows: 10.4% 0.5 + (23.3% – 10.4%) 1.5 + … + (100% – 80.4%) 12.5 = 6.2 The run-off period thus calculated could be used to discount the average bodily injury claim, if the reinsurer received the premiums on 1 January of the pricing year, which, of course, is never the case. For this reason, the average investment period is as follows: average investment period = average run-off period – average delay in receipt of premiums. If premiums are received with one year’s delay, the average investment period in our example is 5.2 years. A 4% interest rate, for example, yields a discounting factor of: 1 / 1.045.2 = 0.816 oder 81.6% It is advisable to discount with the same interest rate as is used in pricing excess of loss treaties, with the five-year government bond rate, for example. The discounting factors for the average property damage claim are calculated in exactly the same way. However, the average run-off period of the property damage claims is much shorter than that of the bodily injury claims. As a result, the average investment period will also be shorter, and the discounting factor higher. The following calculations are based on the assumption that the average run-off period for the property damage claims is 0.9 years. In this case, the investment period is 0.9 – 1 = –0.1 years. This means that the reinsurer is advancing money to the primary insurer for the duration of 0.1 years to settle the property damage claims. The discounting factor is then 1.040.1 or 100.4%. 15 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 Average bodily injury claim 50 454 Average discounted claim 6 113 10:18 Uhr Seite 16 Applying the two discounting factors identified above, the average discounted motor TPL claim is obtained as follows: 0.902 2 295 1.004 + 0.098 50 454 0.816 = 6 113 Only the average discounted claim is of importance to pricing; however, the nondiscounted average (7 015 in our example), is also of interest for planning purposes. 3.6 The average premium The denominator in the equation for the expected loss ratio in Section 2, the average premium, must also be predicted for the pricing year, as it changes from year to year for three reasons: 1 the composition of the portfolio changes, and since not all risks pay the same tariff premium, the average tariff premium also changes; 2 increases and reductions in the tariff affect the average premium; 3 every risk changes its bonus/malus (no-claim bonus) class every year and thus pays a different bonus/malus level, that is to say a different percentage of the tariff premium or 100% premium every year. The only exceptions are those risks in the category with the lowest premium rate that have experienced another claim-free year and those in the category with the highest premium rate that have caused yet another accident. As a result, the average bonus/malus level of the overall insured portfolio changes from occurrence year to occurrence year: all markets that operate bonus/malus systems exhibit a gradual decline in the average bonus/malus level, as more and more risks move towards the lower levels. Table 8 shows the premium development in years 1 to 12. The average premium is the ratio between the earned premiums and the number of annual risks. If the average premium is then divided by the average bonus/malus level, the average 100% premium is obtained. Figure 5 shows the development of the average bonus/malus level. Up to year 8, the year-on-year decline is in the order of half a percent. The blip in year 8 can be traced to the liberalisation of the market: in the data on which this publication is based, year 8 represents the calendar year 1995, the last year prior to liberalisation in Switzerland. To attract the vast numbers of accident-free drivers, new premium discounts were introduced, throwing the system slightly off balance. Since then, the average bonus/malus level has been dropping more rapidly, which makes it necessary – sooner or later – to revise the tariffs upward. Table 8 Occurrence year Annual risks Earned premium Average premium 1 2 3 4 5 6 7 8 9 10 11 12 225 551 233 099 240 598 246 004 247 890 248 637 252 266 255 894 273 711 275 728 272 217 272 668 102 852 692 118 085 646 129 129 440 132 631 564 140 247 736 142 626 124 144 283 632 146 001 541 144 213 834 138 609 053 132 005 901 129 622 873 456 507 537 539 566 574 572 571 527 503 485 475 16 Swiss Re: Pricing motor quota share treaties Average bonus/malus level in % 60.7 60.2 60.1 59.6 59.0 58.3 57.7 57.3 52.7 50.6 49.2 47.7 Average 100% premium 751 842 893 905 959 984 991 996 1 000 993 986 997 25.02.2003 Figure 5 10:18 Uhr Average bonus/malus level in % Motorfahrz_Tarifierung_en_dc Seite 17 70 60 50 40 30 20 10 0 2 4 6 Year 8 10 12 Figure 6 shows the average premiums and the average 100% premiums for years 1 to 12. 1 000 Figure 6 Average 100%-premium Premium 800 600 Average premium 400 200 0 2 4 6 Year 8 10 12 To calculate the premium estimate for the pricing year, year 14, we need: the last known value, an estimation of the effects of changes in the tariff (if planned) and an estimation of the average bonus/malus level in the pricing year. The last known value is 475, the average premium for year 12. We assume that changes in the tariff will lead to an average increase of 8% in premiums between the years 12 and 14. If the average bonus/malus level changes on the same scale as from year 11 to year 12, it will decline by 1.5% per year, reaching 44.7% in the year 14. Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 Average bodily injury claim 50 454 Average discounted claim 6 113 Average premium 481 Discounted loss ratio 84.26% Hence, we obtain: 475 1.08 44.7 = 481 47.7 The expected discounted loss ratio is then: 6.63% 6 113 = 84.26% 481 The expected non-discounted loss ratio would be: 6.63% 7 015 = 96.69% 481 17 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 18 3.7 Fluctuation loading The various approaches commonly used for calculating the fluctuation loading in pricing excess of loss treaties can also be used in proportional pricing. The loading which Swiss Re has been successfully applying in non-proportional motor business for many years is a variance loading. For further information on the practical application of the variance loading, in particular on how to determine the loading factor, refer to Appendix 2 (Schmitter and Bütikofer), in which the product of the loading factor and 1 000 000 is defined as the fluctuation factor. For proportional business, the loading takes the following form: Fluctuation loading in % of premium = loading factor 1 000 000 discounted expected loss ratio average discounted claim in CHF million quota share in % 65 Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 Average bodily injury claim 50 454 Average discounted claim 6 113 Average premium 481 Discounted loss ratio 84.26% Fluctuation loading 1.26% For a loading factor of, for example, 0.15 10–6 and a quota share of 25%, the fluctuation loading will be 1.26% in our case: 0.15 0.8426 0.006113 25% 65 = 1.26% The derivation is explained in Appendix 1. It is important to use the same loading factor for both excess of loss and quota share treaties. This makes the two types of treaty directly comparable: an excess of loss with a fluctuation loading of likewise 1.26% raises the reinsurer’s fluctuation by the same margin as the quota share in the present numerical example. The reinsurer then has no reason to give one of the treaties preference over the other. However, the discounted risk premium of the XL treaty will be about ten times lower than the 84.26% of the quota share treaty, since excess of loss treaties typically absorb high fluctuations even at low risk premiums – in return for high fluctuation loadings charged. 18 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 19 3.8 Expense loading Expense loading represents the expenditure (wages, office rentals, taxes, etc) incurred by the reinsurer in connection with the individual treaty. Its amount is independent of the size of the treaty, even if in practice it is often (rather simplistically) expressed as a percentage of the risk premium. It is calculated as follows: expense loading = amount of expenses expected premiums quote share Assuming that the expenses amount to CHF 100 000, the quota share is 25% and the cedent’s expected premium income for the pricing year is 125 000 000, the expense loading is obtained as follows: 100 000 = 0.32% 125 000 000 0.25 Estimates used in pricing Frequency 6.63% Proportion of bodily injury claims 9.8% Average property damage claim 2 295 Average bodily injury claim 50 454 Average discounted claim 6 113 Average premium 481 Discounted loss ratio 84.26% Fluctuation loading 1.26% Cedent’s earned premiums CHF 125 million Expense loading 0.32% The maximal possible commission is the difference between 100% and the technical premium. The technical premium as the sum of the discounted risk premium, fluctuation and expense loadings in our example is: 84.26% + 1.26% + 0.32% = 85.84% Thus, the maximal possible fixed commission on motor TPL alone is: 100% – 85.84% = 14.16% 19 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 20 4 Motor own damage (MOD) Pricing for a motor own damage cover is essentially the same as pricing for the liability cover: The expected loss ratio is calculated from the claims frequency, average claim and average premium; then fluctuation loading and expense loading are added. In the following, we highlight only the major differences from liability pricing, without going into all the details of the calculations and estimations. 4.1 Covered loss events Unlike compulsory third party liability insurance, to which statutory regulations apply, motor own damage insurance is less uniform. It covers damage to the policyholder’s vehicle for which no one else is liable: losses and damage due to collision, theft, hail, storms, fire, etc. In Switzerland, Germany and Austria, there are two standard products which will be called Inclusive of collision (IC) and Exclusive of collision (EC) for the purposes of this publication. The IC product covers the types of damage mentioned inclusive of collision, whereas the EC product covers the same damages exclusive of collision. In other markets, other combinations are common, collision, glass and theft, for example. Before pricing these products, it is important to check back with the primary insurer about what the various products actually cover. The claims frequency, average claim and average premium must then be determined separately for each product. 4.2 Claims frequency, average claim, discounting and average premium Apart from natural hazards such as hail, storm or flooding, which must be treated separately, the claims frequencies for the various products are determined in the same way as for liability: at the end of the first development year, the number of claims is known and is divided by the number of annual risks. As in TPL, the last known value is normally the most reliable estimate for the future. As a rule of thumb, and as in TPL, 1 500 claims are sufficient for a reliable prognosis. The average claims of the various products depend on the covered perils, the insured vehicles and the amount of any policyholder deductibles. Run-off of claims is similar to the run-off of property damage claims in TPL: by the end of the first development year, practically all claims have been paid. Again, trend computations are recommended for predicting the average claim size. The minimum number of claims per occurrence year needed for a reliable prognosis depends on the product. In EC in Switzerland, the reliability threshold is taken to be roughly 6 000, in IC 9 500 (the reasoning is given in Appendix 1). Because of the rapid claims settlement, as in the case of property damage claims under TPL, discounting is of no practical importance. The average premium of the pricing year is calculated in exactly the same way as for liability. In particular, in the case of products with a bonus/malus system, the development of the average bonus/malus level must also be taken into account. 20 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 21 4.3 Allowing for natural hazards Natural hazards such as hail, storm or flooding are responsible for the greatest fluctuations in the motor own damage claims burden. The assumptions compiled in Table 9 concerning the MOD portfolio illustrate the further computations with concrete numerical examples. Table 9 Estimate for the pricing year IC EC Frequency without natural hazards Average discounted claim Average premium Earned premiums Number of annual risks 27.4% 2 780 1 015 66 000 000 65 025 12.5% 1 645 316 35 000 000 110 759 The average annual claims burden due to natural hazards depends on the return period of the covered natural events, the average number of risks affected and the average claim per affected risk. The primary insurer’s risk portfolio influences each of these three variables: an insurer operating only locally will be less frequently affected by a hailstorm than one whose portfolio is spread over a whole country; a natural event, a hailstorm, for example, affects a larger proportion of a small local insurer’s risks than that of a large insurer whose risks are spread over the whole country; the average claim per risk depends on any policyholder deductibles that may be applicable and on the individual insurer’s claims settlement practice. The following assumptions apply to the further calculations: the return period of natural events is 4 years; the proportion of risks affected per natural event is 5%; the average claim per risk is 1 645 (the average EC or similar claim provides a reasonable estimate, if more precise data are not available). This yields the average annual claims burden due to natural hazards: proportion of risks affected number of annual risks average claim size return period or per risk: proportion of risks affected average claim size return period In our numerical example we obtain: 0.05 1 645 = 21 4 Thus, all the variables required for predicting the expected discounted loss ratio are known, and insertion of the values from Table 9 yields: 27.4% 2 780 + 21 = 77.12% for IC 1 015 12.5% 1 645 + 21 = 71.72% for EC 316 21 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 22 4.4 The loadings for fluctuation and expenses From the various possibilities available for calculating fluctuation loading, we again choose the variance loading, as in the case of TPL. No loading is necessary for the fluctuations in the claims burden without natural events, because these fluctuations are negligible. However, the fluctuations as a result of natural events are significant. The loading in % of the premiums is calculated as follows (the derivation is given in Appendix 1): Fluctuation loading in % of premium = loading factor 1 000 000 average catastrophe loss in CHF million maximum catastrophe loss in CHF million / return period / earned premiums in CHF million quota share in %. On the basis of the above assumptions, the average catastrophe loss is: 0.05 (65 025 + 110 759) 1 645 = 14 458 234 or, in CHF million, 14. 458 234 Estimates used in pricing Natural hazards: Return period 4 years Proportion of risks affected per natural event 5% Average claim per risk 1 645 IC: Frequency without natural hazards 27.4% Average discounted claim 2 780 Average premium 1 015 Number of annual risks 65 025 Discounted loss ratio 77.12% EC: Frequency without natural hazards 12.5% Average discounted claim 1 645 Average premium 316 Number of annual risks 110 759 Discounted loss ratio 71.72% Fluctuation loading 3.36% Expense loading 0.24% It should be noted that the loadings must be calculated collectively for all products affected by the same natural catastrophe, and not for each product separately. In our example, IC and EC are calculated together. The maximum catastrophe loss is likewise the overall loss of all products. If the maximum loss is taken as 25 000 000, our numerical example yields the following result: Fluctuation loading in % of premium = 0.15 14. 458 234 25 / 4 / (66 + 35) 25% = 3.36% The expense loading is calculated in exactly the same way as for TPL. Assuming that the expenses amount to CHF 60 000, the loading for IC and EC together is 0.24%: 60 000 = 0.24% (66 000 000 + 35 000 000) 0.25 22 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 23 5 The fixed commission Table 10 compiles all the data necessary for calculating the maximal possible fixed commission for a motor quota share treaty covering liability, IC and EC. The discounted expected loss ratio, the fluctuation loading and the expense loading in the bottom line are weighted averages from the corresponding figures for liability, IC and EC insurance, the estimated premiums serving as the weights. Table 10 Class Estimated premiums in millions Discounted expected loss ratio in % Fluctuation loading in % Expense loading in % TPL IC EC Total 125 66 35 226 84.26 77.12 71.72 80.23 1.26 3.36 3.36 2.20 0.32 0.24 0.24 0.28 The maximal possible commission which the reinsurer can pay on this motor treaty is: 100% – 80.23% – 2.20% – 0.28% = 17.29% 23 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 24 Appendix Appendix 1: Mathematical deductions A Minimum number of claims required for a frequency estimate Let the number N of claims from a portfolio of s risks follow a Poisson distribution. Let the frequency f = E[N] / s be estimated by N / s. N has to be sufficiently high so that N / s will not deviate more than c = 5% from f with a probability of 95%. For the Poisson distribution, Var[N] = E[N], and the distribution of (N – E[N]) / √E[N] is asymptotically normal with mean 0 and standard deviation 1. From | N / s – f | ≤ c · f follows | N – f · s | / √(f · s) ≤ c · √(f · s). For c = 0.05 we obtain from the normal distribution 0.05 · √(f · s) = 1.96, or, resolved for f · s, f · s = 1 537. B Minimum number of bodily injury claims for the estimate of the proportion of bodily injury claims Let the probability of a claim being a bodily injury claim be r. Let the number K of bodily injury claims out of n claims follow a binomial distribution with mean n · r and variance n · r · (1 – r). K has to be sufficiently high so that K / n will not deviate more than c = 5% from r with a probability of 95%. (K – n · r) / √[n · r · (1 – r)] is asymptotically normal with mean 0 and standard deviation 1. From | K / n – r | ≤ c · r follows | K – n · r | /√[n · r · (1 – r)] ≤ c · n · r/√[n · r · (1 – r)]. For c = 0.05 we obtain from the normal distribution 0.05 · √(n · r) = 1.96 · √(1 – r), or, resolved for n · r, n · r = 1 537 · (1 – r). For r ≈ 0.1, n · r ≈ 1 383. C Minimum number of claims for the estimate of the expected total claim Let the n claims X1, ..., Xn be independent and identically distributed with expected value E and coefficient of variation v. n has to be sufficiently high so that ∑ Xi / n will not deviate more than c = 5% from E with a probability of 95%. According to the central limit theorem, ∑ Xi / n follows an asymptotic normal distribution with mean E and variance E2 · v2 / n. From | ∑ Xi / n – E| ≤ c · E follows | ∑ Xi / n – E| · √n / (E · v) ≤ c · √n / v. For c = 0.05 we obtain from the normal distribution 0.05 · √n / v = 1.96, or, resolved for n, n = (1.96 · v / 0.05)2. In Western Europe, the coefficients of variation in motor TPL are about 2.3 for property damage claims and 3.7 for bodily injury claims. This yields n = 8 129 for property damage claims and n = 21 037 for bodily injury claims. In Switzerland, the coefficient of variation for EC claims is about 2, for IC claims, about 2.5. This yields n = 6 147 and n = 9 604, respectively. 24 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 25 D Variance loading The loading recommended in this publication is proportional to the variance assumed by the reinsurer. Define as: f the loading factor, if the Swiss franc is used as the monetary unit E the discounted expected individual claim V the variance of the discounted individual claim v the coefficient of variation the expected number of Poisson-distributed claims r the expected discounted loss ratio P the earned premium q the quota share in %. Then the fluctuation loading in % of the premiums is loading = f · · (E2 + V) · q2 P·q = f · (1 + v2) · r · E · q In Western Europe, the coefficient of variation in motor TPL is around v = 8. Thus, 1 + v2 = 65. The loading factor f depends on the reinsurer’s capital cost and profit target and can change from year to year. It lies in the order of 10–7 CHF–1. In Appendix 2 (Schmitter and Bütikofer), the product f · 106 is referred to as the fluctuation factor. 1.5 Variance loading for natural catastrophe claims The fluctuation loading for natural catastrophe claims is based on the following simple estimate of E2 + V: let the catastrophe loss X be limited by the maximum claim size M, X ≤ M. Then, E2 + V = E[X2] ≤ E[X · M] = E · M. If this upper bound is used as the estimate for E2 + V, the fluctuation loading in percent of the premiums is obtained as follows: loading = 25 f··E·M·q P Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc Appendix 2: References 25.02.2003 10:18 Uhr Seite 26 Boulter, Anthony and Grubbs, Dawson (2000): Late claims reserves in reinsurance. Swiss Re. Schmitter, Hans and Bütikofer, Peter (1997): Estimating property excess of loss risk premiums by means of the Pareto model. Swiss Re. Appendix 3: Questions for cedents A Questionnaire for motor third party liability (TPL) Claims number of annual risks for the last 3 years number of claims (including nil cost claims) for the last 3 years prediction of claims frequency. If this differs from the most recent observed frequency, please state reasons. number of bodily injury claims for the last 3 years run-off triangles for the property damage claims burden over the last 7 years, separately for paid and reserved amounts run-off triangles for the bodily injury claims burden over the last 12 years, separately for paid and reserved amounts Premiums last year’s earned premiums are any changes to the tariff planned? How many policies will be affected by these changes? How will they affect the premium level? spread of risks over the bonus/malus classes average bonus/malus level prediction of the average bonus/malus level for the pricing year B Questionnaire for motor own damage (MOD) Claims Perils covered: for each combination of covered perils: number of annual risks for the last 3 years number of claims for the last 3 years prediction of claims frequency. If this differs from the most recent observed frequency, please state reasons. claims burdens (splitting into paid and reserved amounts not necessary) for the last 3 years has there been a natural catastrophe in the last 3 years? Return period of natural catastrophes? Percentage of policies affected? Premiums Same questions as for motor TPL 26 Swiss Re: Pricing motor quota share treaties Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 27 Other publications in the “Technical publishing” category include: Late claims reserves in reinsurance “Late claims reserves in reinsurance” is a second and thoroughly revised edition of the popular 1989 Swiss Re publication. Targeted at general management, claims managers, underwriters and accountants, the publication gives insight into why reinsurers need to hold additional claims reserves over and above the claims estimates that the ceding insurer supplies. Order no.: 207_8955_en/de Estimating property excess of loss risk premiums by means of the Pareto model This is the revised edition of the highly popular title “Estimating property excess of loss premiums by means of the Pareto model” (1978). The Pareto model is often used to estimate risk premiums in property excess of loss with great loss priority, as in such treaties, loss experience is often inconsistant and can be confusing. Order no.: 207_9798_en/de Setting optimal reinsurance retentions The primary goal of reinsurance is to maintain, at an acceptable level, the random fluctuations in the results of primary insurers. The Swiss Re publication “Setting optimal reinsurance retentions” reveals how this can be effectively achieved. The work is aimed equally at primary insurers, reinsurers and reinsurance brokers who advise their clients on arranging reinsurance programmes. The concepts described are applicable to every class of insurance and are not limited to the casualty or property damage classes. Order no.: 207_01288_en/de The economics of insurance How insurers create value for shareholders The greatest shareholder value is generated by those insurers who identify and capitalise on the best business opportunities and have optimum operating efficiency. Although this publication is founded on a detailed economic study, it caters specifically to practitioners. Examples that clearly demonstrate how to apply the structure in practice are provided, making it invaluable for insurance management seeking to improve long-term profitability. Order no.: 207_01310_en/de An introduction to reinsurance This teaching aid explains the reinsurance system to prespective underwriters and introduces working methods involved using examples and graphics. The publication does not pretend to be self-explanatory but is intended as a support for both trainer and trainees. Order no.: 207_9682_en/de/es/pt The insurance cycle as an entreprenuerial challenge The fact that insurance buyers, cedents, reinsurers and industrial lines insurers have to live with fluctuating market prices does not mean that the individual company has to sit out the cycle compliantly. There are supply-side strategies which ensure better results over the cycle than by passively standing by and allowing the cycle to take its course. Order no.: 1492633_02_en/de The brochure entitled Swiss Re Publications includes a complete overview of all available Swiss Re publications. Order no.: 1492220_02_en Non-proportional reinsurance accounting This publication offers a general and easy-tounderstand overview of the essentials and processes of non-proportional treaty business. The first part of the two-part publication presents the business process using a fictitious portfolio and the reinsurance programme developed for it. The teaching material is supplemented by a workbook containing case studies which can also be used for reviewing and consolidating the information. Order no.: 207_00218_en/de Publications in the “Technical publishing/Casualty” series: Introduction to rating casualty business Excess-of-loss rating in longtail branches such as motor or general liability is no easy matter since estimates must be made of claims incurred which will not in fact be determined by court ruling until 10 or 20 years later. Every round of renewals presents the new challenge of calculating estimates of future claims burdens based on previous data or current risk profiles. This publication shows the considerations on which this procedure is based and serves as an introduction to the essential steps involved in this demanding field of work. Order no.: 207_99211_en/fr/de Liability and liability insurance: Yesterday – today – tomorrow Increasingly, the momentous changes brought about by technological progress are contributing to a marked rise in compensation claims, notably when technological progress is seen to have failed. To illustrate the changes that have taken place in liability insurance, this publication covers the net of liable parties and the types of liability loss for which claims are made. The judicial interpretation of policy wordings, the types of triggers for coverage and the definition of insured liability loss are each examined. Order no.: 205_01292_en/de How to order To order a copy of a publication, please send an email to [email protected]. You can also place your order via our portal at www.swissre.com. Please include the title of the publication, the order number and the language abbreviation. Language versions are indicated at the end of the order number as follows: English _en German _de French _fr Italian _it Portuguese _pt Spanish _es Swiss Re’s P&C publications are edited and produced by Technical Communications, Chief Underwriting Office. Motorfahrz_Tarifierung_en_dc 25.02.2003 10:18 Uhr Seite 28 Hans Schmitter Chief Underwriting Office Dr Hans Schmitter is a specialist in casualty business at Swiss Re’s Chief Underwriting Office. He studied mathematics and lingusitics at the University of Bern. Following two years as a Fortran programmer at the Swiss Federal Office of Statistics, Hans Schmitter joined Swiss Re in 1973 to work in the area of pricing concepts and tool development. In 1991, he moved to the Europe Department of Winterthur insurance company, where he was active in primary insurance pricing. Returning to Swiss Re, he has worked since 1999 in the Product Management Casualty unit. In addition to the publication in hand, Hans Schmitter has also authored the following: “Estimating property excess of loss risk premiums by means of the Pareto model”, and “Setting optimal reinsurance retentions”. © 2003 Swiss Reinsurance Company Title: Pricing motor quota share treaties Author: Hans Schmitter Chief Underwriting Office Translation by: Swiss Re Group Language Services Editing and realisation: Technical Communications Chief Underwriting Office Graphic design: Galizinski Gestaltung, Zurich Photograph: Alan Schein, Blue Planet/The Stock Market The material and conclusions contained in this publication are for information purposes only, and the author(s) offers no guarantee for the accuracy and completeness of its contents. All liability for the integrity, confidentiality or timeliness of this publication or for any damages resulting from the use of information herein is expressly excluded. Under no circumstances shall Swiss Re Group or its entities be liable for any financial or consequential loss relating to this product. Order number: 1493126_02_en Property & Casualty, 03/03, 3000 en Motorfahrz_Tarifierung_en_dc Swiss Reinsurance Company Mythenquai 50/60 P.O. Box CH-8022 Zurich Switzerland Telephone +41 43 285 2121 Fax +41 43 285 5493 [email protected] Swiss Re publications can also be downloaded from www.swissre.com 25.02.2003 10:18 Uhr Seite 30
© Copyright 2026 Paperzz