The purpose of this article is to illustrate how the pension benefits a pension saver will (expect to) receive will depend on the type of pension scheme chosen. We compare three widely different pension savings products: the “traditional” with-profits scheme involving bonus entitlement (average interest rate product), a market-based Unit Link scheme and, finally, a formula based smoothed investment-linked annuity scheme – TimePension in short – which is on many points a cross between the two prior-mentioned types of savings products.

The three product types mentioned above have been analysed in previous literature, but those comparisons were based almost entirely on the values of pension savings accounts at the expiry of the accumulation period. This article will include the payout phase (decumulation phase) in the analysis, enabling us to analyse the size of paid-out pension benefits themselves as well as the possibilities of adjusting these benefits periodically. Compared to earlier articles, we have also improved the underlying model for the uncertainty of the underlying financial market.

The article demonstrates that expected pension benefits from the three schemes are an increasing function of the allocation to shares in the underlying investment portfolios. TimePension involves the highest allocation to shares and therefore offers, on average, the highest pension benefits, followed by the Unit Link scheme. In the third and last place comes the traditional with-profits scheme, which has a relatively low allocation to shares, but which, in return, also provides relatively safe and stable pension benefits. We also show, however, that the stability of pension benefits from a TimePension scheme is completely level with the stability of benefits from the traditional scheme. Unit Link-based pension benefits, on the other hand, vary far more, and pension savers in this product segment will experience much higher annual adjustments – in a both negative and positive direction – than savers in the other product segments.

In this paper we review the Wilkie asset model for a variety of UK economic indices, including the Retail Prices Index, both without and with an ARCH model, the wages index, share dividend yields, share dividends and share prices, long term bond yields, short term bond yields and index-linked bond yields, in each case by updating the parameters to June 2009. We discuss how the model has performed from 1994 to 2009 and estimate the values of the parameters and their confidence intervals over various sub-periods to study their stability. Our analysis shows that the residuals of many of the series are much fatter-tailed than in a normal distribution. We observe also that besides the stochastic uncertainty built into the model by the random innovations there is also parameter uncertainty arising from the estimated values of the parameters.

We analyze the mortality of couples by fitting a multiple state model to a large insurance data set. We find evidence that mortality rates increase after the death of a partner and, in addition, that this phenomenon diminishes over time. This is popularly known as a “broken-heart” effect and we find that it affects widowers more than widows. Remaining lifetimes of joint lives therefore exhibit short-term dependence. We carry out numerical work involving the pricing and valuation of typical contingent assurance contracts and of a joint life and survivor annuity. If insurers ignore dependence, or mis-specify it as long-term dependence, then significant mis-pricing and inappropriate provisioning can result. Detailed numerical results are presented.

In this paper, we investigate the problem of how to combine operational losses collected from various banks of different sizes and loss reporting thresholds in order to estimate the distribution of operational loss severities for a bank of a given size. We model the severity of operational losses by using the extreme value theory to account for the reporting bias of the external data, and a regression analysis based on the GAMLSS framework to model the scaling properties of operational losses. In contrast to previous studies on the scaling problem, our analysis gives particular emphasis to the scaling properties of the tail of the loss distribution. Contrary to existing knowledge, we find that the size of a bank is an important determinant of the severity of operational losses and that the tail index of the distribution is negatively correlated with the size of the bank. The results indicate that for very large banks, distribution of the operational loss severity can be extremely heavy tailed (i.e. tail index less than 1), a finding which have significant implications for capital calculation as well as for risk management. Furthermore, we also demonstrate that the capital estimates provided by our model is consistent with the industry standards and the model can be used by individual banks to simulate data to complement their internal data.

In the classical risk model, we use probabilistic arguments to write down expressions in terms of the density function of aggregate claims for joint density functions involving the time to ruin, the deficit at ruin and the surplus prior to ruin. We give some applications of these formulae in the cases when the individual claim amount distribution is exponential and Erlang(2).

The prediction of adequate claims reserves is a major subject in actuarial practice and science. Due to their simplicity, the chain ladder (CL) and Bornhuetter–Ferguson (BF) methods are the most commonly used claims reserving methods in practice. However, in contrast to the CL method, no estimator for the conditional mean square error of prediction (MSEP) of the ultimate claim has been derived in the BF method until now, and as such, this paper aims to fill that gap. This will be done in the framework of generalized linear models (GLM) using the (overdispersed) Poisson model motivation for the use of CL factor estimates in the estimation of the claims development pattern.

The addition of a set of cohort parameters to a mortality model can generate complex identifiability issues due to the collinearity between the dimensions of age, period and cohort. These issues can lead to robustness problems and difficulties making projections of future mortality rates. Since many modern mortality models incorporate cohort parameters, we believe that a comprehensive analysis of the identifiability issues in age/period/cohort mortality models is needed. In this paper, we discuss the origin of identifiability issues in general models before applying these insights to simple but commonly used mortality models. We then discuss how to project mortality models so that our forecasts of the future are independent of any arbitrary choices we make when fitting a model to data in order to identify the historical parameters.

This paper examines income inequality over stages of the later-life course (age 45 and older) and systems which can be used to mitigate this inequality. Two hypotheses are tested: (1) levels of income inequality decline during old age because public benefits are more equally distributed than work income; and (2) because of the progressive nature of government benefits, countries with stronger public income security programmes are better able to reduce income inequalities during old age. The analysis is performed by comparing age groups within seven OECD countries (Canada, Germany, the Netherlands, Norway, Sweden, the United Kingdom, and the United States of America) using Luxembourg Income Study data from around 2000. Both hypotheses are supported. Several conclusions are drawn from the findings.

The purpose of this paper is to identify a workhorse mortality model for the adult age range (i.e., excluding the accident hump and younger ages). It applies the “general procedure” (GP) of Hunt & Blake [(2014), North American Actuarial Journal, 18, 116–138] to identify an age-period model that fits the data well before adding in a cohort effect that captures the residual year-of-birth effects arising in the original age-period model. The resulting model is intended to be suitable for a variety of populations, but economises on the number of period effects in comparison with a full implementation of the GP. We estimate the model using two different iterative maximum likelihood (ML) approaches – one Partial ML and the other Full ML – that avoid the need to specify identifiability constraints.

Conditional Monte Carlo replaces a naive estimate Z of a number z by its conditional expectation given a suitable piece of information. It always reduces variance and its traditional applications are in that vein. We survey here other potential uses such as density estimation and calculations for Value-at-Risk and/or expected shortfall, going in part into the implementation in various copula structures. Also the interplay between these different aspects comes into play.

This is the first of two papers in which we estimate transition probabilities amongst levels of disability as defined in the Australian Survey of Disability, Ageing and Carers. In this paper we describe both the main tools of our estimation and the estimation of the numbers of individuals in different disability categories at annual intervals using survey data that are available at five year intervals. In Paper II we describe our estimation procedure, followed by its implementation, discussion of results and graduation of the estimated transition probabilities.

We consider the Bayesian over-dispersed Poisson (ODP) model for claims reserving in general insurance. We choose two different types of prior distributions for the parameters and then study the different Bayesian predictors. This study leads, on the one hand, to the classical chain ladder predictor and, on the other hand, to Bornhuetter & Ferguson predictors. We highlight (either analytically or numerically) how these predictors are obtained and how their prediction uncertainty can be determined.

In this paper, we carry out an investigation on modelling basis risk and measuring risk reduction in a longevity hedge constructed by index-based longevity swaps. We derive the fitting procedures of the M7-M5 and common age effect+Cohorts models and define the level of longevity risk reduction. Based on a wide range of hedging scenarios of pension plans, we find that the risk reduction levels are often around 50%–80% for a large plan, while the risk reduction estimates are usually smaller than 50% for a small plan. Moreover, index-based hedging looks more effective under a more precise hedging scheme. We also perform a detailed sensitivity analysis on the hedging results. The most important modelling features are the behaviour of simulated future variability, portfolio size, speed of reaching coherence, data size and characteristics, simulation method, and mortality structural changes.

This paper is a continuation of and a supplement to the paper by Jørgensen & Linnemann. Both papers deal with TimePension – a (formula-based) smoothed investment-linked annuity pension scheme. Both papers compare TimePension with other pension savings products using stochastic financial simulation. TimePension as well as the financial model and simulation concept being used in both papers were introduced in the paper op cit.

Jørgensen & Linnemann compare TimePension with a traditional with-profits scheme involving bonus entitlement and an investment-linked Unit Link scheme with a fixed proportion of assets invested in equities and other so-called risky assets. Here, TimePension is compared with two investment-linked life-cycle (target date) products. The focus is especially on the decumulation period, and we show that despite the fact that the life-cycle products reduce investment risk in retirement (compared with TimePension’s fixed proportional allocation of assets), the year-over-year stability in benefits provided by TimePension cannot be matched. Even though the overall investment risk on accumulated benefits is at the same level, TimePension provide year-over-year stability, also in case of sharp price changes in the financial market.

Our results show that not only investment profiles define the stability of annuity benefits over time. In addition, more fundamental elements of the product design are important. The perspective on product design and development is Danish, but two of the compared products are generic life-cycle products that exist in equivalent forms in many countries. Similarly, the smoothed income annuities could also become an alternative product design in an international perspective.

We compare results for 12 multi-population mortality models fitted to 10 distinct socio-economic groups in England, subdivided using the Index of Multiple Deprivation. Using the Bayes Information Criterion to compare models, we find that a special case of the common age effect (CAE) model fits best in a variety of situations, achieving the best balance between goodness of fit and parsimony. We provide a detailed discussion of key models to highlight which features are important. Group-specific period effects are found to be more important than group-specific age effects, and non-parametric age effects deliver significantly better results than parametric (e.g. linear) age effects. We also find that the addition of cohort effects is beneficial in some cases but not all. The preferred CAE model has the additional benefit of being coherent in the sense of Hyndman et al. ((2013) Demography50(1), 261–283); some of the other models considered are not.

One of the pricing strategies for Bonus–Malus (BM) systems relies on the decomposition of the claims’ randomness into one part accounting for claims’ frequency and the other part for claims’ severity. By mixing an exponential with a Lévy distribution, we focus on modelling the claim severity component as a Weibull distribution. For a Negative Binomial number of claims, we employ the Bayesian approach to derive the BM premiums for Weibull severities. We then conclude by comparing our explicit formulas and numerical results with those for Pareto severities that were introduced by Frangos & Vrontos.

Actuarial ratemaking is usually performed at product and guarantee level, meaning that each product and guarantee is considered in isolation. Moreover, independence between policyholders is generally assumed. In this paper, we propose a multivariate Poisson mixture, with random effects correlated using a hierarchical structure, to accommodate for the dependence that may exist between unobserved risk factors across Home and Motor insurance and between policyholders from the same household. The hierarchical structure accounts for the fact that Home insurance covers the whole household, whereas Motor insurance policies are subscribed by specific policyholders within the household. The model allows to periodically correct the a priori expected claim frequencies using the reported number of claims in any of the considered products. Applications show that the impact of the number of claims reported in Motor insurance on the number of claims expected in Home insurance is larger than the other way around. Moreover, an out-of-sample analysis validates an improved predictive power. Also, the model allows to identify more rapidly the riskiest households.

In this paper, we consider a number of practical and theoretical aspects of the Wilkie asset model, many of which apply to any similar model used for simulation over time. We discuss the experience of the Wilkie model since 2009. We then discuss the variables that can form the working set, the input set and the output set, all of which may be different. There are different ways of simulating, either in a linear parallel structure or in a branching tree structure. We then discuss the initial conditions required, which may be market conditions at some date, or may be “neutral” initial conditions, which may be defined in different ways. One method of generating initial conditions would be to simulate them randomly, from their own long-term distribution, and we show how to calculate the means, variances and covariances of these. What we call “neutralising parameters” may have a role, and we discuss how these may be found. Finally, we suggest using additional information in the first periods of the simulation to adjust the formulae or parameters for a limited “select period”.

This paper studies estimation of the conditional mean squared error of prediction, conditional on what is known at the time of prediction. The particular problem considered is the assessment of actuarial reserving methods given data in the form of run-off triangles (trapezoids), where the use of prediction assessment based on out-of-sample performance is not an option. The prediction assessment principle advocated here can be viewed as a generalisation of Akaike’s final prediction error. A direct application of this simple principle in the setting of a data-generating process given in terms of a sequence of general linear models yields an estimator of the conditional mean squared error of prediction that can be computed explicitly for a wide range of models within this model class. Mack’s distribution-free chain ladder model and the corresponding estimator of the prediction error for the ultimate claim amount are shown to be a special case. It is demonstrated that the prediction assessment principle easily applies to quite different data-generating processes and results in estimators that have been studied in the literature.

This paper investigates the use of censored regression quantiles in the analysis of claim termination rates for income protection (IP) insurance. The paper demonstrates the importance of modeling quantiles given the growing interest of regulators and others in stochastic approaches to valuation of insurance liabilities and risk margins.