Motor insurers in the United Kingdom are not subject to Government control over the rating structures they use or the levels of premiums they charge. The market is highly competitive, and each insurer therefore needs to make the best estimate he can of the premium required for each category of risk, to produce a given level of profit. He also needs to estimate the extent to which departures from such premium levels can be justified, for example in order to have a simple rating structure or to meet competition.

The purpose of this note is to give some examples of the statistical tabulations being produced in one insurance office in the United Kingdom. The statistical system which has been developed covers many aspects of motor insurance management, but in this note we shall direct attention to just two, namely

(i) continuous review of the variations in the claims experience from one risk category to another, as a guide to the relative premiums required for the different categories; and

(ii) continuous review of the changing composition of the portfolio and of the movements in and out, to try to assess the extent to which gains and losses of business can be attributed to pricing differences, marketing strategies, etc.

It is fantastic how the computer has changed our attitude to numerical problems. In the old days when our numerical tools were paper, pencil, desk calculator and logarithm tables we had to stay away from formulas and methods which led to too lengthy calculations. A consequence is that we have a tendency to think of numerical analysis in terms of the classical tools. If we go back to the results of earlier writers it seems, however, very likely that many results and formulas developed by them which had earlier a theoretical interest only could nowadays be applied successfully in numerical analysis.

As an example I take the ruin probability ψ(x). The Laplace transform of ψ(x) is given by the following expression

where c > 1. In fact (c — 1) is equal to the “security loading”. The function p(y) is equal to the Laplace transform of the claim distribution. We assume that the mean claim amount is equal to one, i.e. p′(0) = — 1.

In his book from 1955 [1] Cramer points out that this formula will be more easy to handle if the claim distribution is an exponential polynomial. In this case we have

where

Cramér's results are given on pages 81-83 in his book. We reproduce them here with a slight change of notations only.

In motorcar insurance—at least in Europe—is widely used a merit rating system called bonus (or malus) system, characterized among others by the fact that only the number of claims occurred does modify the premium. A great variety of different bonus systems is in use. Each country—if not each company—seems to prefer a different bonus system and there is a lack of agreement whether one of them is better than the other. Efforts to construct a theoretically optimal bonus system have not been able to change the situation. General methods to evaluate and compare different existing or proposed bonus systems are therefore wanted.

In the present paper the theory of Markow chains is used to develope formulas for some asymptotic properties of bonus systems. Quantities: efficiency of a bonus system, discrimination power of bonus rules and minimum variance bonus scale are introduced. The last of them gives an asymptotical solution for the problem to find locally “best” bonus scales for given bonus rules. Finally the Danish bonus system is used as an illustration.

We call a merit rating system bonus system if the following assumptions are valid:

i. All policies of a given risk group can be devided into a finite number of classes so that the premium of a policy for a given period depends solely on the class for that period.

ii. The actual class is uniquely defined by the class for the previous period, and the number of claims occurred (regulated) during the period.

In many risk theoretical questions a central problem is the numerical evaluation of a convolution integral and much effort has been devoted over the years to mathematical and computational aspects. The paper presented to this colloquium by O. Thorin shows the subject to be topical but the present note stems from a recent paper by H. L. Seal, “Simulation of the ruin potential of non-life insurance companies”, published in the Transactions of the Society of Actuaries, Volume XXI page 563.

In this paper, amongst other topics, Seal has presented some simulations of ruin probabilities over a finite time interval. Some years ago (Journal Institute of Actuaries Students' Society, Volume 15, 1959). I pointed out that the form of ruin probability could be expressed as a successive product of values of a distribution function. To my knowledge no one has attempted to see if this product form was capable of development and the numerical values in Seal's paper prompted me to spend a little time on the problem. In the time I had available it has not been possible to do more than experiment, but the conclusions reached may be of interest to other workers in this field. They showed that calculation is feasible but laborious. However, the knowledge that it can be done may suggest methods of improving the techniques.

Seal's first simulation example is the calculation of ruin probabilities when the distribution of the interval of time between the claims is negative exponential and the individual claim distribution is also negative exponential. Instead of following Seal's method we can determine the “gain per interval” and find that if λ is the security loading the frequency function for the gain z is

i.e. a Laplace distribution.

In the wake of the technical development of our era we are increasingly faced with claims of extremely large amounts.

Whereas even in the past century large claims were mostly due to elemental or natural forces such as earthquakes, hurricanes and floods, there are today other causes, conditioned by human factors, that have considerably increased in importance. Another fact to be observed is that the risks that are passed on to the insurance market assume a more serious character year by year. Large building complexes, giant tankers, containers, atomic power stations, dams and jumbo jets pose for the insurers problems which are in no way easy to solve, since the risk covers required touch the limits of market capacity and their rating is subject to great uncertainties. At the same time, the readiness to underwrite such covers is a matter of utmost importance for the private insurance industry.

For the coverage of such large risks, various insurance forms have been developed in the past years within the range of non-proportional methods. Besides the excess of loss and stop loss covers in their usual form, the cumulative loss cover and the largest claim cover have attracted particular attention. The cumulative risk cover has, for example, been dealt with in [11]) and has also been frequently offered in practice. The coverage of the largest claim or, more generally, of the sum of the n largest claims does not appear to have gained a proper foothold in practice in spite of various valuable contributions in this field—considering particularly [2], [3], [5], [8] and [14]. The fact that it was during 1963/64 that various authors analysed these problems is no coincidence, for it was then that in two ASTIN Colloquia these topics were discussed.

The following numerical values of ruin probabilities, Ψ(u, T) for finite times T, have been calculated by the method proposed in “Analytical steps towards a numerical calculation of the ruin probability for a finite period when the risk process is of the Poisson type or of the more general type studied by Sparre Andersen”, presented to this colloquium by Olof Thorin. The notations used in the sequel follow those of Thorin.

Two distributions of the individual claims are considered, viz.,

The latter distribution is a rather crude attempt to interprete the extremely skew distribution (Swedish non-industry fire insurance 1948-1951) considered by Cramér in his treatise “Collective Risk Theory”, Jubilee volume of Försäkringsaktiebolaget Skandia (1955) pp. 43-45.

Likewise two distributions of the interoccurence times are considered, viz.,

The d.f.B was considered by Sparre Andersen (TICA 1957 vol. II pp. 225-227).

Note that the first moment equals one in all the d.f. mentioned.

Though the analytical machinery also seems to work for o ≤ c ≤ 1 the Ψ values are indicated only for some values of c > 1. As known from Thorin's paper c stands for 1 + λ, where λ is the premium

loading, which means that for c = o Ψ(u, T) corresponds to the tail of the d.f. for the total amount of claims during the period (o, T).

E. Franckx [1] has established the distribution function of the largest individual claim of a portfolio. By assuming the number of claims to be Poisson distributed, H. Ammeter was able to develop the distribution function of the total loss excluding the largest individual claim [2] as well as the distribution function of the nth largest claim [3].

Of course, the nth largest claim is dependent on the largest claim, second largest claim and so on, down to the (nth — 1) largest claim. If we assume the number of claims to be Poisson distributed and the amount of the individual claim to be Pareto distributed, the correlation between the mth largest and the nth largest claim can be expressed by an analytical formula which is susceptible to numerical computation.

With this knowledge we shall be able to compute the variance of the sum of the n largest claims and moreover the correlation between the sum of the n largest claims and the total loss amount. Although an excess of loss reinsurance treaty and a treaty reinsuring the n largest claims are very different in their construction, this paper will show that from a practical point of view there exists a similarity between the two treaties. The correlation coefficient between the sum of the n largest claims and the sum of all claims exceeding a certain limit enables us to assess the degree of similarity.

The correlation coefficient and thus the degree of similarity will prove to be high even in case of the reinsurance of only a small number of largest claims.

Finally, the knowledge of the two first moments of the sum of the n largest claims allows us to compute the premium and the security or variance loading for the reinsurance of the n largest claims.

1. At an earlier ASTIN Colloquium participants were invited to present notes on problems which they considered as important but unsolved. There was little response to this invitation, presumably because a problem, once it is well formulated, is almost solved.

In this Note I do not present any new problems. In stead I try to outline a framework which may be useful for analysing different risk problems and seeing them in their proper perspective. In my view, a framework of this kind is urgently needed to place today's actuarial work on a sound foundation.

2. In general an insurance contract will define two stochastic processes. We lose little by assuming that the processes are discrete, and describing them in the following manner:

(i) The payment process: x0, x1 … xt …, where xt is the amount which the company pays to settle claims in period t, or at time t.

(ii) The premium process: p0, p1 … pt …, where pt is the premium which the company receives in period t, or at time t.

If the contract is concluded at time t = o, the Principle of Equivalence requires that

For the typical short-term contract with premium payable in advance (i) will reduce to

3. For a long-term insurance contract one usually requires that the inequality

shall hold for all τ. This means that the company must never be a net creditor of its customer.

1. M. Karl Borch a présenté, dans plusieurs de ses communications à l'ASTIN), un modèle de comportement de l'assureur, qu'il qualifie lui-même de simplifié, mais qui met néanmoins en évidence une intéressante propriété d'optimalité. L'assureur maximiserait la valeur actuelle des dividendes futurs en laissant les bénéfices éventuels s'ajouter aux fonds propres jusqu'au niveau Z; les montants qui porteraient alors les richesses libres au dessus du niveau Z seraient intégralement distribués. L'espérance mathématique optimale V(S, Z) de la valeur actuelle des dividendes futurs dépend, dans ces conditions, du capital S initial. D'autre part, il reste entendu que si S > Z, alors (S — Z) est immédiatement „remboursé”, et, si S = o, la société cesse, en principe, ses opérations, toutefois, M. Karl Borch a envisagé la possibilité d'un renflouement).

2. L'utilisation pratique de ce modèle pose deux problémes particulièrement importants:

— la société a-t-elle le droit de fonctionner avec le capital S?

— les capitalistes accepteraient-ils d'investir S?

2.1. Le premier de ces problèmes a été envisagé par nous-même dans une étude précédente). Il apparaît que la contrainte imposée par l'autorité de tutelle peut-être sommairement résumée par l'obligation d'un capital minimum T. La valeur actuelle optimale V(S, T, ZT) des dividendes futurs est alors diminuée par rapport à la valeur précédemment considérée (correspondant à T = o), soit V(S, o, Z0), tandis que le niveau ZT de la barrière reflexive est placé plus haut que Z0.

2.2. Il reste à considérer le second problème, c'est-à-dire, finalement, celui de l'opportunité du placement des capitaux dans une société d'assurances dommages. Cela nécessite que soit défini un critère de choix des placements.

In order to make this report clear to those without experience in insurance matters, we first present some basic facts about the insurance business. In so doing we intentionally omit certain facts irrelevant to the present study. The most important omission of this kind is our assumption that the insurance business operates without administrative expenses and without sales costs. We also assume that the insurance business is run in such a way that no profit is made. It will be evident to readers already directly connected with insurance problems what further omissions we have made and why we have made them.

Insurance is the establishment of a contract between the insurance company and the insured person This contract is by tradition called the “insurance policy” and the insured person is called the “policyholder”. By agreeing to the insurance policy, the policyholder commits himself to paying certain premiums to the insurance company, and the insurance company commits itself to paying certain amounts to the policyholders. The conditions under which such payments are to be made are of many different kinds: the policy-holder dies, the policyholder becomes ill, a homeowner's house is burnt down, or the policyholder has a collision in his car. The circumstances under which a payment is to be made to the policy-holder are described in detail in the insurance policy. The amount to be paid is either fixed or variable: in life insurance the amount to be paid is always fixed and stated in the insurance policy, but in most other forms of insurance the intention is that the insurance compensate the policyholder for the losses he might incur as a consequence of the events covered by his policy.