The providers of financial products offer cover against risk events. Individuals or companies buying these products all have different features - no two people in the world are alike in every respect, not even identical twins. A product provider could assess each individual or company and determine the premium to charge and the cover to provide for each risk it considers.
This approach works when the risks are rare and large and it is very difficult to group them. Marine hull and cargo covers are a good example: not only are ships generally different from each other but the cargos they carry and the routes they travel accentuate the differences. It is appropriate and practical to assess each risk individually.
Other risks are smaller and individual assessment would be prohibitively expensive. For these risks the provider usually has access to a large amount of data concerning how the population behaves. If the population can be divided into relatively homogeneous groups, a price can be determined that applies to all risks in that group.
If a product provider can pool independent homogeneous risks, then as a result of the Central Limit Theorem the profit per policy will be a random variable that follows the normal distribution with a known mean and standard deviation. The company can use this result to set premium rates which ensure that the probability of a loss on a portfolio of policies is at an acceptable level.
Irrespective of how a provider constructs its homogeneous risk pools, there will be a range of risks in the pool. In life assurance, mortality and morbidity risk increases rapidly at later ages. If the provider sets a rate for male lives aged 82 (presumably based on the expected experience of a l ife aged 82.5), then a person aged 82.9 will be getting better terms than appropriate given the risk that person poses. If everyone aged 82.9 realised this and took out policies, the pricing assumption of an average age of 82.5 would be wrong, and the company would incur a loss.
Selection is taking advantage of inefficiencies in a provider’s pricing basis to secure better terms than might otherwise be justified, normally at the expense of the product provider.
Selection is not a fraudulent, immoral, or illegal activity.
Careful underwriting is the mechanism by which the company ensures that its risk groups are homogeneous.
The risk groups are defined using rating factors, eg age, gender, medical history, height/ weight, lifestyle.
In theory, a provider should continue to add rating factors to its underwriting system until the differences in risk between the different categories of the next rating factor are indistinguishable from the random variation between risks that remains after using the current list of rating factors.
Both the ability of prospective policyholders to provide accurate responses to questions and the cost of collecting information limit the extent to which rating factors can be used. In general, a proposal form should not ask for information which requires specialist knowledge. For example, the cost of undertaking extensive blood tests has to be weighed against the expected cost of mortality or morbidity claims that will be ‘saved’ as a result of having this information.
From a marketing point of view, prospective policyholders will want the process of underwriting to be straightforward and quick.
In practice, rating factors will be included if they avoid any possibility of selection against the company, and satisfy the time and cost constraints of marketing. This decision is often driven by competitive pressures. If several companies introduce a new rating factor, which is a good descriptor of the underlying risk, then other companies will need to follow this lead or risk adverse selection against them.
When a life table is constructed it is assumed to reflect the mortality experience of a homogeneous group of lives, ie all the lives to whom the table applies follow the same stochastic model of mortality represented by the rates in the table. This means that the table can be used to model the mortality experience of a homogeneous group of lives which is suspected to have a similar experience.
If a life table is constructed for a heterogeneous group then the mortality experience will depend on the exact mixture of lives with different experiences that has been used to construct the table. Such a table could only be used to model mortality in a group with the same mixture. It would have very restricted uses.
In such cases the tables are separate (different) during the select period, but combined after the end of the select period.
In addition to variation by age and sex, mortality and morbidity rates are observed to vary:
* between geographical areas, eg countries, regions of a country, urban and rural areas
* by social class, eg manual and non-manual workers
* over time, eg mortality rates usually decrease over time.
Such factors are:
* occupation
* nutrition
* housing
* climate / geography
* education
* genetics
It is rare that observed differences in mortality can all be ascribed to a single factor. It is difficult to disentangle the effects of different factors because of the relationships between them.
For example, mortality rates of those living in sub-standard housing are (usually) higher than those of people living in good quality housing. However, those living in sub-standard housing usually have less well-paid occupations and lower educational attainment than those living in good quality housing. Part or all of the observed difference may be due to these differences and not to housing differences.
Occupation can have several direct and indirect effects on mortality and morbidity.
Occupation determines a person’s environment for often 40 or more hours each week. The environment may be rural or urban, the occupation may involve exposure to harmful substances such as chemicals, or to potentially dangerous situations such as working at heights. Some occupational effects may be moderated by health and safety at work regulations.
Some occupations are naturally healthier, whereas some work environments give exposure to a less healthy lifestyle.
Some occupations by their very nature attract more healthy or unhealthy workers. This may be accentuated by health checks made on appointment or by the need to pass regular health checks, eg airline pilots. However, external factors can distort a presumed state of health; for example former miners who have left the mining industry as a result of ill health and then chosen to sell newspapers will inflate the morbidity rates of newspaper sellers.
A person’s occupation largely determines their income, and this permits them to access a particular lifestyle, content and pattern of diet, quality of housing and access to healthcare. The effect on mortality and morbidity can be positive or negative.
Nutrition has an important influence on morbidity and in the longer term on mortality.
Poor nutrition can increase the risk of contracting many diseases and hinder recovery from sickness. In the longer term, the burden of increased sickness can influence mortality.
Excessive or inappropriate eating can lead to obesity and an increased risk of associated diseases (heart disease, hypertension) leading to increased morbidity and mortality.
Inappropriate nutrition may be the result of economic factors - lack of income to buy appropriate foods or the result of a lack of health and personal education resulting in poor nutritional choices. There are also social and cultural factors which encourage or discourage the consumption of certain foods and drinks, such as alcohol.
The standard of housing encompasses not only all aspects of the physical quality of housing (state of repair, type of construction, heating, sanitation) but also the way in which the housing is used, such as overcrowding and shared cooking.
These factors have an important influence on morbidity, particularly that related to infectious diseases (from tuberculosis and cholera to colds and coughs) and thus on mortality in the longer term.
The effect of poor housing is often mixed up with the general effects of poverty.
Climate and geographical location are closely linked. Levels and patterns of rainfall and temperature lead to an environment which is amicable to certain kinds of diseases, such as those associated with tropical regions.
Effects can also be observed within these broad categories - differences between rural and urban areas in a geographical region. Some effects may be accentuated or mitigated depending upon the development of an area, with industrial development leading to better roads and communications.
Natural disasters (such as tidal waves and famines) will also affect mortality and morbidity rates, and may be correlated with particular climates and geographical locations.
Education influences the awareness of the components of a healthy lifestyle, which reduces morbidity and lowers mortality rates. It encompasses both formal education and more general awareness resulting from public health and associated campaigns.
This effect can be apparent in aspects such as:
* increased income
* choice of a better diet
* the taking of exercise
* personal health care
* moderation in alcohol consumption and smoking,
* awareness of the dangers of drug abuse
* awareness of a safe sexual lifestyle
Some of these are direct causes of increased morbidity such as smoking and excessive alcohol consumption, which lead to diseases such as lung and other forms of cancer, and strokes. A healthy lifestyle with improved fitness can lead to an enhanced ability to resist diseases.
Genetics may give information about the likelihood of a person contracting certain diseases, and therefore may provide improved information about the chances of sickness or death. Such information may be used in isolation for the individual in question or, more usefully, by combining it with the life histories of the current and past generations of the family.
Genetics is a rapidly developing new area of study for the medical profession. There are increasing numbers of specific diseases being identified where genetic information provides firm predictive evidence of the chances of sickness or death relative to a person of average health.
Each group is defined by a specified event (the select event) happening to all the members of the group at a particular age, eg buying a life assurance policy or retiring on ill-health grounds.
The mortality or morbidity is estimated for each group and for the population that is not exposed to the specific event. The mortality / morbidity patterns in each group are observed to differ only for the first s years after the select event. The length of select period is s years.
The differences are temporary, producing the phenomenon called temporary initial selection.