Differentiation (and homogenisation) of the insured

One of the most important methods through which the insurer succeeds in making sure that the calculated premium of the insurance is sufficient to cover benefits is by setting markedly different premiums for different segments of the risk community; in other words the insured are differentiated according to their risk. In the case of life insurances – not including annuity insurance – this is traditionally done by determining two premium tables¹⁴ for each product, one for men and one for women. The different premium is set according to the age of entry, and accordingly the default is differentiation by age and gender.¹⁵ Beyond this, risk assessment (or underwriting) is generally (but not always) carried out, where efforts are made to quantify risks resulting from individual health status and behaviours such as drinking, smoking, consumption of drugs, etc. In addition, occupation (e.g. dangerous vocation – miner, policeman, etc.), sports and hobbies (e.g. extreme sports) are taken into account and it is decided whether it is necessary to increase the premium based on age and gender calcu-lations. In other words, the insured are further differentiated by risk.

The essence of differentiation may also be expressed in a way that it enables the splitting of the risk community, which is inhomogeneous from the point of risk, into homogeneous parts (or at least more homogeneous segments

14In the EU we have to use past tense in this respect, because of the Gender Di-rective. The rationale of this directive is questionable and we only hope that the coming Age Directive will not prohibit differentiation according to age, which could lead to the end of life insurance as we know it. Later I will discuss the topic of prohibiting gender-differentiation (which is – mistakenly – considered as discrimination in the Directive).

15Naturally the premium depends (typically in a linear manner) on the sum insured, and the term of the insurance (if there is any difference in this respect at all, e.g. in the case of life-long annuities we may assume that they have only one type of term), but this difference is not due to the risk of the insured.

pared to the entire risk community). Consequently, differentiation also means homogenisation at the level of segments of the risk community.

Nevertheless, differentiation cannot always be carried out for various rea-sons, or it is not always possible to create an adequately high number of risk

“sub-communities” with a sufficient number of members. In such cases, ho-mogenisation of the risk community may be an independent solution, which is the opposite of differentiation in a certain sense, (though from another point of view, these assume or complement each other). In the course of homogenisa-tion, risks deemed to be unmanageable, or individuals deemed to be unman-ageable risks, are (temporarily or permanently) excluded from the insurance and from the risk community (waiting-time, exclusion, exemption, refusal).

Eventually lower and upper limits are defined for the possible sum assured, where attempts are made to remove extreme risks. (I will also detail this type of method in relation to annuities).

Although differentiation and homogenisation are generally applied in paral-lel, they can replace each other to a certain extent. If there is less differentia-tion (e.g. due to business or cost consideradifferentia-tions, to simplify underwriting an insurance policy), the importance of homogenisation increases. By applying a greater degree of differentiation, risks that would otherwise be excluded, or insured persons who constitute an extreme risk, can also be insured by the insurer.

2.1.1. DIFFERENTIATION IN THE CASE OF VOLUNTARY ANNUITY INSURANCE

The significant difference between the above cases and the case of voluntary annuity insurance is that traditionally no risk assessment is applied (neither exclusion, nor refusal). A different premium is applied based on age and gen-der (and possibly a lower and/or upper limit is set for the possible insurance amount – see below), but state of health, occupation, drinking, smoking etc.

are not assessed. Instead, the basic supposition is that annuity insurance is by definition purchased by those who have the best life expectancy and whose expected remaining life is longer than the average. Therefore, mortality is not calculated from the public mortality table, but rather from the annuitant select mortality table that indicates significantly lower mortality.

The reason for the above practice is that the risk of annuity insurance is pre-cisely contrary to that of all other, traditional life insurances (term, endow-ment, fixed term, whole life); that is the risk to be managed is not that the insured individual dies much sooner than the average, but on the contrary, that the insured party dies at an older age than average. Therefore, traditional risk assessment methods and their whole approach cannot be applied in this case,

since in the case of annuity insurances the best client is the client who poses the greatest danger in other forms of insurance. (Therefore no exclusion or refusal of the client is applied).

The consequence of the above practice is that buying annuity insurance is mainly worthwhile for those who have reasonable cause to expect a long life.

Consequently, on the one hand this practice strictly constrains the potential range of clients, and on the other hand it justifies the insurer’s preliminary assumption of the client’s long life. Naturally this does not apply in the case of mandatory annuity insurances, since people have no choice with regard to concluding an insurance contract.

Recently, some insurers became disturbed by the consequences of tradition-al practice since it constrained their client base, so the so-ctradition-alled impaired annu-ity, with premium differentiated according to state of health, emerged on the British and American markets. The logic is that clients suffering from certain sicknesses and damaged health are offered a preferential tariff, i.e. their state of health is also taken into consideration in the premium (in addition to age and gender). Naturally this already presupposes a certain risk assessment, the logic of which is precisely the opposite of the traditional method: those who

“simulate” illness/disability must be weeded out and not those who claim to be healthy. The phenomenon may also be interpreted so that, similarly to other sectors of voluntary insurance, competition in the annuities market has led to differentiation of the insured.¹⁶

We must realise that while traditional risk-assessment penalizes attitudes and behaviour that is generally regarded as negative (unhealthy lifestyle, ad-dictions), in the case of impaired annuities these proven negative habits may provide the basis for a preferential premium.¹⁷

16Pre-empting later sections of this book to a certain extent, the question may be raised: why, in the case of voluntary annuity insurances, is differentiation made accord-ing to state of health and not accordaccord-ing to level of education, which is precisely contra-ry to what is raised with respect to mandatocontra-ry annuities in the latter part of this work.

The probable reason for differentiation according to educational attainment not occur-ring in the case of free market annuities is that annuity insurance is voluntarily conclud-ed almost exclusively by people who have completconclud-ed some form of higher conclud-education, while in the mandatory system this is also obligatory for those with a lower level of education. Naturally, the possibility of differentiation according to educational attain-ment could gradually decrease in future if college certificates become universal.

17Though no impaired annuity is necessarily needed for this, as indicated by a piece of news that was published in the Hungarian tabloids and caused quite a stir. According to these reports: “The Dutch Paerel Leven voluntary supplementary pension insurance fund has proposed charging higher premiums to heavy smokers based on the fact that their life expectancy is shorter.” See: http://www.stop.hu/articles/article.php?

id=284988, 1 March, 2008 Source: MTI

In summary, differentiation in the case of annuities may be envisaged accord-ing to those important factors which have a high impact on remainaccord-ing life expectancy – and therefore on the premium of the annuity. These factors in-clude:

• Place of residence, housing conditions

• Workplace and nature of work (e.g. physical, intellectual, etc.)

• Level of education

• State of health (e.g. disability, sickness)

• Lifestyle, drinking, smoking, etc., sports

• Marital status

• Income (and via this, the magnitude of the expected annuity)

• Adverse selection – provided the insured party has a choice

It is important to note that these factors are not independent from each other;

for example, state of health is related to lifestyle, and probably also to housing conditions and level of education, and vice versa, etc. This must be taken into consideration at the final determination of differentiation factors. Below, I examine and rank the possible use of these factors (with the exception of ad-verse selection, which I detail in the coming chapter).

2.1.2. OPTIONS FOR THE DIFFERENTIATION OF MANDATORY ANNUITIES

According to the above, the pure actuarial approach would require a deeper differentiation in groups of diverse risk, since the calculation becomes increas-ingly stable by further deepening differentiation. In addition, the stochastic equivalence between the service received and the premium paid becomes valid for increasingly narrow groups, in other words; the deeper the differentiation, the more equitable the annuity. On the other hand, reference made to the stability of the calculation also naturally means that the less differentiated the annuity, the less stable the calculation, and especially in the case of several, competing ser-vice providers, and some mechanism is required to manage this instability.

At this time no-one doubts that annuity benefits provided for identical assets must be differentiated according to age. This is particularly fortunate from a regulatory point of view, because the lack of this differentiation would either make a mandatory annuity system unmanageable, or would require drastic measures (basically necessitating a uniform age of retirement) that both the theory and practice of the Hungarian pension system has already left behind.

As regards any other differentiation, the various opinions are totally divergent.

Looking at this issue from the point of view of differentiation possibilities as regards annuities, the major obstacles to differentiation (apart from age) are

not independent from each other. There is some overlap between them and therefore these might theoretically be listed separately. Primarily due to the overlapping, I do not highlight the general problem of differentiation by gen-der as a special obstacle, although the list below covers this issue.

1. Connection to the annuity of Pillar I 2. EU regulations

3. Equity problems – relation to other social welfare systems 4. Incentives for undesired trends

5. Available data, assessability of the given parameter Let us analyse these in the above order.

Connection to the annuity of Pillar I. Private pension funds in Hungary were established with the aim of enabling the annuity purchased from accumulated assets to complement the annuity received from Pillar I, where the latter re-mains the larger component within the total pension. Although it was not spec-ified that the annuity of Pillar II must operate according to the same principle as the Pillar I annuity (and in fact several factors such as enabling a choice between different kinds of annuity suggested precisely the opposite), it is (would have been) a logical requirement that the two parts of an individual’s pension match each other, i.e. that major parameters like differentiation must be in harmony within the two pillars.

Looking at the annuity of Pillar I from the perspective of differentiation, we may say that it includes a differentiation by age, although not in a perfectly correct way actuarially, but it exists nevertheless. However, as differentiation by gender is becoming increasingly less common¹⁸, we may practically assume that is does not exist. There exists differentiation according to the level of accumulated entitlements (income), i.e. the digressive scale, but this is also going out of fashion. There is no differentiation according to place of resi-dence, level of education, state of health, marital status, lifestyle, drinking, smoking, etc., but there exists differentiation according to job and workplace to the extent that a preferential retirement age is determined for certain occupa-tions (miners, ballet dancers, members of the armed forces).¹⁹ Although this preferential age of retirement is not calculated based on lifetime being different from the average, but to a certain extent on an arbitrary basis and while

18Becoming less prevalent because the lower retirement age of women may be con-sidered as a form of differentiation by gender, even though this makes the annuity even less correct from an actuarial perspective.

19In 2011 the majority of these were cancelled for new entrants. One of the excep-tions are ballet dancers.

sidering other factors (e.g. in the case of ballet dancers the opportunity to carry on the given vocation²⁰).

So it may be assumed that (using wording that takes into account the char-acteristics of Pillar II) in Pillar I there is no (current or later) differentiation by gender, or according to the magnitude of accumulated assets, or any other kind of differentiation with the exception of differentiation by workplace and the nature of the work performed, but differentiation does not happen in an actuar-ially correct manner in this case either.

So, in contrast to Pillar II annuities, Pillar I represents (would have repre-sented) the requirement that it should include no form of differentiation except by age and certain occupation groups. This issue is especially sharply raised when it comes to differentiation by gender, as annuities were differentiated according to this criterion by Hungarian insurers until 21 December 2012. This (i.e. to immediately ensure that this general practice of private insurance com-panies should not be applied to mandatory annuities) is probably the reason why the Private Pension Funds Act has always stipulated that the mandatory annuity must be calculated using the unisex mortality table.

Though the requirement posed by Pillar I is clear, it is still not necessary to restrict options for differentiation in Pillar II. However, the use of the unisex mortality table is such a strong requirement that no divergence is possible from this perspective.

The EU regulations characteristically do not concern themselves with what options member states allow or prohibit for differentiation in the case of insur-ance products and annuities, with one important exception: differentiation by gender. The so-called Gender Directive adopted in 2004 (“COUNCIL DI-RECTIVE 2004/113/EC of 13 December 2004 implementing the principle of equal treatment between men and women in the access to and supply of goods and services”) prohibits the application of different premium calculations ac-cording to gender, although not with immediate effect and while taking into account the specific regulatory frameworks of member states. The provision is justified according to items (18) and (19) of the Preamble as follows:

“(18) The use of actuarial factors related to gender is widespread in the pro-vision of insurance and other related financial services. In order to ensure equal treatment between men and women, the use of gender as an actuarial factor should not result in differences in individuals' premiums and benefits.

To avoid a sudden readjustment of the market, the implementation of this rule should apply only to new contracts concluded after the date of transposition of this Directive.

20Combined with the arbitrary assumption that the individual will not be able to conduct any other occupation to earn a living and consequently must retire.

(19) Certain categories of risk may vary between the genders. In some cases, gender is one but not necessarily the only determining factor in the assessment of risks insured. For contracts ensuring those types of risks, Member States may decide to permit exemptions from the rule of unisex premiums and benefits, as long as they can ensure that underlying actuarial and statistical data on which the calculations are based, are reliable, regularly up-dated and available to the pub-lic. Exemptions are allowed only where national legislation has not already ap-plied the unisex rule. Five years after transposition of this Directive, Member States should re-examine the justification for these exemptions, taking into ac-count the most recent actuarial and statistical data and a report by the Commis-sion three years after the date of transposition of this Directive.“

In Article 5 of the Directive (“Actuarial factors”) it specifically states:

(1) Member States shall ensure that in all new contracts concluded after 21 December 2007 at the latest, the use of gender as a factor in the calculation of premiums and benefits for the purposes of insurance and related financial services shall not result in differences in individuals' premiums and benefits.

(2) Notwithstanding Paragraph 1, Member States (including Hungary) may decide before 21 December 2007 to permit proportionate differences in indi-viduals' premiums and benefits where the use of gender is a determining factor in the assessment of risk based on relevant and accurate actuarial and statistical data. The Member States concerned shall inform the Commission and ensure that accurate data relevant to the use of gender as a determining actuarial factor are compiled, published and regularly updated. These Member States shall review their decision five years after 21 December 2007, taking into account the Commission report referred to in Article 16, and shall forward the results of this review to the Commission”.

So in summary: whilst member states must prohibit differentiation by gen-der in premium calculation, they are free to temporarily decide not to imple-ment it in the case of certain types of insurance. This temporary exemption may theoretically be endlessly extended, but should they fail to do so this ex-emption will terminate and the unisex tables must be applied.

This was the situation until 1 March 2011, when the Court of Justice of the European Union made this Article invalid with effect from 21 December 2012. This means that in future, differentiation according to gender is not an option in the case of insurance products (Court of Justice of the European Un-ion [2011]). However, in theory this rule is not valid for mandatory annuities because they fall under the sphere of authority of a different Directive to the one governing employer pension funds, but it is valid for mandatory annuities pro-vided by insurance companies. Although it seems that independently from the Gender Directive, and especially because of the reasons explained in the previ-ous chapter, a unisex rate must also be applied in the case of mandatory annuity.

Equity problems – relation to other social welfare systems. Although some of the above-listed aspects of differentiation may be justified and correct, mainly taking into consideration life style, drinking, smoking, sports etc., and accordingly with regard to state of health, they may nevertheless lead to ine-qualities under the existing social insurance systems. In the case of annuity, people with no unhealthy habits and who lead a healthy lifestyle and partici-pate in sports would be expected to have a longer lifetime, meaning they will have to pay a higher premium for the same annuity. This is fair if this same factor is also taken into consideration (with an opposite effect) in other sys-tems, because a healthy person probably requires less healthcare, so it would

Equity problems – relation to other social welfare systems. Although some of the above-listed aspects of differentiation may be justified and correct, mainly taking into consideration life style, drinking, smoking, sports etc., and accordingly with regard to state of health, they may nevertheless lead to ine-qualities under the existing social insurance systems. In the case of annuity, people with no unhealthy habits and who lead a healthy lifestyle and partici-pate in sports would be expected to have a longer lifetime, meaning they will have to pay a higher premium for the same annuity. This is fair if this same factor is also taken into consideration (with an opposite effect) in other sys-tems, because a healthy person probably requires less healthcare, so it would