Upper bounds for ruin probabilities under model uncertainty

In this thesis, we present the new results found in exploring the asymptotic tail probability of the phase-type scale mixture distributions, and apply a subclass of the phase-type scale mixtures, the class of Erlangized scale mixture distributions, to the approximation of the probability of ruin.We consider the class of phase-type scale mixture distributions, which can be written as MellinnStieltjes convolution P*G of a phase-type distribution G and a nonnegative but otherwise ar-bitrary distribution P. Such a class can also be seen as the class of distributions of the product of two independent random variables: a phase-type random variable Y s G and a nonnegative random variable S s P. We call S the scaling random variable and its corresponding distribution P the scaling distribution. We investigate conditions for such a class of distributions to be either light- or heavy-tailed, we explore subexponentiality and determine their maximum domains of attraction.In this thesis, particular focus is on phase-type scale mixture distributions where the scaling distribution has discrete support m such a class has been recently used in risk applications to approximate heavy-tailed distributions. We extend an existing scheme for numerically calculating the probability of ruin of a classical CramernLundberg reserve process having absolutely continuous but otherwise general claim size distributions. We employ a subclass of the phase-type scale mixtures, which is also a dense class of distributions that we denominate Erlangized scale mixtures (ESM). Such a class corresponds to nonnegative and absolutely continuous distributions which can be written as a MellinnStieltjes convolution P*Gmnof a discrete nonnegative distribution P with an Erlang distribution Gmn. A distinctive feature of such a class is that it contains heavy-tailed distributions when the scaling distributions have unbounded support, and it provides sharp approximations to heavy-tailed distributions. We suggest a simple methodology for constructing a sequence of distributions having the form P*Gmnwith the purpose of approximating the integrated tail distribution of the claim sizes. Then we adapt a recent result which delivers an explicit expression for the probability of ruin in the case that the claim size distribution is modelled as an Erlangized scale mixture. We provide simplified expressions for the approximation of the probability of ruin and construct explicit bounds for the error of approximation. We complement our results with a classical example where the claim sizes are heavy-tailed.

This study examines the implementation of the ruin probability model in risk management in life insurance companies. The main focus of this study is to evaluate how factors such as initial surplus, premium revenue level, and claim frequency affect the ruin probability of insurance companies. Using the collective risk model approach and relevant claim distribution, this study develops two methods to calculate the ruin probability: an analytical approach and a Monte Carlo simulation. The simulation results show that increasing the initial surplus and premium level significantly reduces the ruin risk, while increasing the claim frequency increases the ruin probability. In addition, the gamma claim distribution is more suitable for modeling claims in life insurance than the exponential distribution. Model validation is carried out by comparing the prediction results with historical data of insurance companies, which shows a high level of accuracy. This study provides important insights for insurance companies in designing more effective and optimal risk management strategies

A model for the statistical analysis of the total amount of insurance paid out on a policy is developed and applied. The model simultaneously deals with the number of claims (zero or more) and the amount of each claim. The number of claims is from a Poisson-based discrete distribution. Individual claim sizes are from a continuous right skewed distribution. The resulting distribution of total claim size is a mixed discrete-continuous model, with positive probability of a zero claim. The means and dispersions of the claim frequency and claim size distribution are modeled in terms of risk factors. The model is applied to a car insurance data set.

Vehicle insurance companies in many countries use the Bonus-Malus System to determine the policyholder’s net premium. The determination of net premiums on the Bonus-Malus System is based solely on the frequency of claims and ignores the severity of claims. This is unfair to policyholders who have small claims. To overcome this problem, the net premium determination method in Bonus-Malus System was developed taking into account both the frequency and severity of claims. Frequency and severity be assumed to be independent. In determining the net premium, a posterior distribution of parameters of the frequency and severity distribution is required. In the case of frequency and severity independent, the determination of the posterior distribution for frequency and severity is performed separately. This thesis discusses the determination of net premium based on frequency distribution and severity distribution for frequency and severity independent.

We consider an insurer who has a fixed amount of funds allocated as the initial surplus for a risk portfolio, so that the probability of ultimate ruin for this portfolio is at a known level. We consider the question of whether the insurer can reduce this ultimate ruin probability by allocating part of the initial funds to the purchase of a reinsurance contract. This reinsurance contract would restore the insurer's surplus to a positive level k every time the surplus fell between 0 and k. The insurer's objective is to choose the level k that minimizes the ultimate ruin probability. Using different examples of reinsurance premium calculation and claim size distribution we show that this objective can be achieved, often with a substantial reduction in the ultimate ruin probability from the situation when there is no reinsurance. We also show that by purchasing reinsurance the insurer can release funds for other purposes without altering its ultimate ruin probability.

On a modification of the classical risk process

The ruin probability of an insurance company is a central topic in risk theory. We consider the classical Poisson risk model when the claim size distribution and the Poisson arrival rate are unknown. Given a sample of inter-arrival times and corresponding claims, we propose a semiparametric estimator of the ruin probability. We establish properties of strong consistency and asymptotic normality of the estimator and study bootstrap confidence bands. Further, we present a simulation example in order to investigate the finite sample properties of the proposed estimator.

Practical aspects of stop-loss calculations

of thesis entitled STUDY ON INSURANCE RISK MODELS WITH SUBEXPONENTIAL TAILS AND DEPENDENCE STRUCTURES Submitted by CHEN YIQING for the degree of Doctor of Philosophy at The University of Hong Kong in August 2009 In recent years, insurance risk studies in the presence of dependence structure and heavy-tailed claims are popular topics in actuarial science. In this thesis, we consider the renewal risk model with certain dependence structures, namely negative dependence and quasi-asymptotic independence. We also assume that claim sizes follow a heavy-tailed distribution, in particular, a subexponential distribution. We focus on studying the impact of heavy tails and dependence structures on ruin probabilities and the tail probabilities of aggregate claims. The asymptotic behavior of the ultimate ruin probability in the renewal risk model was studied in the presence of negatively dependent claim sizes and a constant force of interest. A purely probabilistic approach was employed to establish a simple asymptotic formula for the ruin probability. This result coincides with some known results in the recent literature, and discovers a phenomenon that heavy tails can offset the impact of the dependence structure of claim sizes on the ruin probability. A two-dimensional renewal risk model was considered in which the insurance company consists of two dependent classes of business. The dependence between the two classes comes from the fact that their claims arrive according to a common renewal counting process. For this model, several ruin probabilities were defined and simple asymptotic formulae for these ruin probabilities were derived. In particular, the obtained asymptotic formulae hold uniformly for all time horizons. This uniformity greatly enhances the theoretical value of the results. The next problem was to study precise estimates of the large deviation probabilities of aggregate claims with sizes which are negatively dependent. Specially, the problem for the so-called quasi renewal risk model was investigated in which inter-arrival times are also negatively dependent. A new dependence structure was considered and was termed “quasi-asymptotic independence” structure, which is more general and verifiable than some commonlyused dependence structures. The proposed dependence structure covers a wide range of positive and negative dependence. For the case of Pareto-like tails, the tail probability of the sum was proved to be asymptotically equivalent to the sum of tail probabilities. This confirms again that heavy tails offset the impact of dependence between summands. This result was then extended to randomly weighted sums and random sums. All the obtained results have obvious applications to concrete actuarial problems. In the last part of the thesis, each primary random variable representing the net loss during a period was associated with a random weight or a stochastic discount factor. Therefore, the weighted sum represents the stochastic present value of aggregate losses. Suppose that the primary random variables are long tailed, the weights are positive and bounded, and the primary random variables and their weights are independent. The tail probability of the maximum of the weighted sums was shown to be asymptotically equivalent to that of the last weighted sum. STUDY ON INSURANCE RISK MODELS WITH SUBEXPONENTIAL TAILS AND DEPENDENCE STRUCTURES

Considering the classical model with risky investment, we are interested in the ruin probability that is minimized by a suitably chosen investment strategy for a capital market index. For claim sizes with common distribution of extended regular variation, starting from an integro-differential equation for the maximal survival probability, we find that the corresponding ruin probability as a function of the initial surplus is also extended regular variation.

On the probability of ruin in a Markov-modulated risk model

This paper studies a bidimensional renewal risk model with constant force of interest and heavy-tailed claims. We extend the existing models through allowing arbitrary dependences between each pair of inter-arrival times of the two kinds of insurance claims. A precise asymptotic formula for the finite-time ruin probability is obtained when the claims have subexponential tails. Further restricting the claim-size distributions within the class of extended regular variation, we derive the corresponding asymptotic formula for the infinite-time ruin probability.

Claims experience in non-life insurance is contingent on random eventualities of claim frequency and claim severity. By design, a single policy may possibly incur more than one claim such that the total number of claims as well as the total size of claims due on any given portfolio is unpredictable. For insurers to be able to settle claims that may occur from existing portfolios of policies at some future time periods, it is imperative that they adequately model historical and current data on claims experience; this can be used to project the expected future claims experience and setting sufficient reserves. Non-life insurance companies are often faced with two challenges when modeling claims data; selecting appropriate statistical distributions for claims data and establishing how well the selected statistical distributions fit the claims data. Accurate evaluation of claim frequency and claim severity plays a critical role in determining: An adequate premium loading factor, required reserve levels, product profitability and the impact of policy modifications. Whilst the assessment of insurers’ actuarial risks in respect of their solvency status is a complex process, the first step toward the solution is the modeling of individual claims frequency and severity. This paper presents a methodical framework for choosing a suitable probability model that best describes automobile claim frequency and loss severity as well as their application in risk management. Selected statistical distributions are fitted to historical automobile claims data and parameters estimated using the maximum likelihood method. The Chi-square test is used to check the goodness-of-fit for claim frequency distributions whereas the Kolmogorov-Smirnov and Anderson-Darling tests are applied to claim severity distributions. The Akaike information criterion (AIC) is used to choose between competing distributions. Empirical results indicate that claim severity data is better modeled using heavy-tailed and skewed distributions. The lognormal distribution is selected as the best distribution to model the claim size while negative binomial and geometric distributions are selected as the best distributions for fitting the claim frequency data in comparison to other standard distributions.

Ruin probabilities in the presence of regularly varying tails and optimal investment

Asymptotic and numerical analysis of the optimal investment strategy for an insurer