Memorandum by Legal & General
1. INTRODUCTION
1.1 Legal & General strongly supports
the paper submitted by the Association of British Insurers (ABI)
to the House of Lords Inquiry into the proposed Directive, and
in particular wishes to endorse its conclusions that:
— the present UK legal system, which
permits the use of gender as a factor in insurance where it is
objectively justified, but prohibits unfair discrimination on
grounds of gender, is an appropriate framework for the provision
of goods and services;
— the proposed Directive would increase
the overall cost of insurance to consumers in general;
— the Commission's interpretation of
academic studies into the reasons for mortality differences between
men and women is inadequately supported by the available evidence;
and
— the impact assessment carried out
before publication of the proposed Directive has been inadequate,
in view of the range of unintended consequences which have already
been identified.
1.2 We also share the concerns expressed
by Mr Callum McCarthy, the Chairman of the Financial Services
Authority (FSA), over the increase in capital required by the
industry to cover the additional risk arising from enforced mispricing
of the policies concerned. These policies may include health and
motor policies as well as life insurance and annuities. Firms
will need to make assumptions about the likely gender mix of insurance
customers and to include pricing margins to cover the inevitable
uncertainty over their assumptions. FSA will require them to hold
additional capital to cover the new risk and the cost of servicing
it will be reflected in a higher overall consumer price for the
relevant types of insurance.
1.3 Rather than repeat in detail the concerns
expressed by ABI and FSA, we would like to focus on two specific
issues:
— The failure of the EU Commission
adequately to consider issues of risk pooling, risk pricing and
underwriting.
— The consequences for grouped insurance
and annuity business, where prices will rise more sharply than
for individual business and where it would be quite exceptional
for either gender to benefit.
2. RISK POOLING,
RISK PRICING
AND UNDERWRITING
2.1 The EU Commission's Explanatory Memorandum
introducing the proposed Directive focuses on the argument that
studies which have tried to separate lifestyle, social class and
environmental factors from gender in the analysis of mortality
statistics have shown that the difference in average life expectancy
between men and women lies between zero and two years "with
the conclusion that the growing gap in life expectancy witnessed
in the general population in some Member States cannot be attributed
to biological differences. Sex is at the very best a proxy for
other indicators of life expectancy."
2.2 It is almost certainly true that the
"growing gap" referred to cannot be attributed entirely
to biological differences. But the residual difference after allowing
as far as possible for other factors still exists and is still
significant. The Commission has not established otherwise, despite
the recent reported statement from its spokeswoman that "it
is not a biological fact that women live longer than men: it is
a statistical observation".
2.3 The Commission appears to be drawing
a spurious distinction between "biological fact" and
"statistical observation". It is a biological fact that
men cannot die from cervical cancer, nor women from prostate cancer.
However, the widely-varying incidences of death for men and women
from a variety of groups of causes to which both sexes are susceptible,
as illustrated in the table of statistical observations below,
make it hard to believe that biological differences can be anything
less than a major influence on them.
PERCENTAGES OF DEATHS BY CAUSE
|
| | 20-29
| 30-39 | 40-49
| 50-59 | 60-69
| 70-79 | 80+
|
|
| Infectious diseases | male
| 2% | 3%
| 2% | 1%
| 1% | 1%
| 1% |
| female |
4% | 3%
| 1% | 1%
| 1% | 1%
| 1% |
| Cancers | male
| 8% | 14%
| 24% | 37%
| 39% | 32%
| 22% |
| female |
20% | 38%
| 47% | 54%
| 46% | 32%
| 14% |
| Circulatory | male
| 6% | 14%
| 30% | 35%
| 40% | 43%
| 43% |
| female |
8% | 13%
| 17% | 19%
| 29% | 40%
| 45% |
| Respiratory | male
| 2% | 4%
| 4% | 6%
| 8% | 12%
| 17% |
| female |
5% | 3%
| 5% | 6%
| 9% | 12%
| 15% |
| Other health related | male
| 23% | 24%
| 22% | 14%
| 9% | 10%
| 15% |
| female |
30% | 24%
| 22% | 14%
| 11% | 14%
| 23% |
| Violence and accidents | male
| 58% | 41%
| 17% | 6%
| 2% | 1%
| 2% |
| female |
32% | 18%
| 9% | 4%
| 2% | 1%
| 2% |
|
MORTALITY RATES FOR FEMALES AS A PERCENTAGE OF RATES FOR
MALES
|
| 20-29
| 30-39 | 40-49
| 50-59 | 60-69
| 70-79 | 80+
|
|
| Infectious diseases | 87%
| 57% | 42%
| 79% | 87%
| 82% | 82%
|
| Cancers | 94%
| 137% | 126%
| 94% | 72%
| 63% | 56%
|
| Circulatory | 52%
| 48% | 37%
| 35% | 45%
| 60% | 87%
|
| Respiratory | 79%
| 44% | 70%
| 69% | 70%
| 64% | 74%
|
| Other health related | 50%
| 53% | 61%
| 69% | 77%
| 84% | 124%
|
| Violence and accidents | 22%
| 24% | 33%
| 41% | 48%
| 64% | 95%
|
| All causes | 39%
| 52% | 65%
| 65% | 61%
| 64% | 84%
|
|
Source: ONS data for England and Wales 2001, quoted in
"Longevity in the 21st Century" by R C Willets and others,
Institute and Faculty of Actuaries, 2004.
2.4 The modest difference in overall longevity between
the sexes disguises a very substantial percentage difference in
mortality rates at most ages, which is a more relevant statistic
for life insurance provision. It also hides some very large differences
within the major causes of death at some ages, notably in relation
to cancers, circulatory disorders and violent/accidental deaths.
It may be instructive to focus on questions arising in just two
of these areas:
— The female death rate from cancers in the age
group 30-49 exceeds that for males. A more detailed analysis shows
that the excess over the more usual pattern, in which female mortality
is around 50 per cent to 70 per cent of the male rate for most
ages and causes, is wholly accounted for by breast and cervical
cancers. Should the excess cost of cancer-related claims be shared
by men in this instance?
— Men under age 50 are three to five times more
likely to die through violence or accidents than females of the
same age. Is this a "biological fact" (eg the consequence,
through the process of Natural Selection, of an inherited disposition
to behaviour that would achieve "alpha male" status
in a social group), or is it the statistical consequence of a
set of lifestyle choices driven by social factors? Does it matter?
If insurers are to offer life cover to this age group, and have
no practicable way to distinguish most of the risk-takers, should
the excess mortality cost arising from the small proportion of
risk-takers—mainly male—be borne by women as well
as by men?
2.5 More generally, there is certainly evidence that
factors such as occupation, income/wealth levels, tobacco smoking,
etc have a profound effect on sickness and mortality levels. In
many cases, these exceed the differences attributable to gender.
It is also the case that none of these factors uniquely determine
the longevity either of a single group subdivided according to
such criteria, or of individuals within it. Unsuspected—occasionally
inexplicable—diseases can strike down even the healthiest
and fittest, while others in apparently high-risk groups live
to a great age.
2.6 It could therefore be argued that any underwriting
process which seeks to assign individuals to such groups and charge
accordingly for insurance or annuities is inherently inequitable,
since greater subdivision, if feasible, might more accurately
identify the level of risk. On the other hand, an exhaustively
detailed investigation would be likely to increase the number
of individuals deemed uninsurable: a consequence which may be
seen as socially unacceptable.
2.7 In practice, an underwriting process can work only
where:
— allocation of individuals to risk groups is made
on objective and verifiable criteria such as age, sex, occupation,
family history of early death or (with some limitations) smoking
habits: criteria such as diet, exercise or tolerance of stress
may reflect significant risk factors but are hard to measure objectively
and are open to mis-statement with little risk of detection; and
— sufficient data exists to provide a reasonable
level of confidence in setting premiums for each such group: given
the extent of available data, there are practical limits to the
level of credible subdivision.
2.8 In a competitive market without regulatory or legislative
constraint, insurers will weigh the cost, complexity, reliability
and customer acceptability of potential risk criteria against
the additional pricing precision and hence competitive advantage
they might bring. Given the correlations between risk factors,
moderately significant but objectively verifiable ones will be
selected rather than highly significant ones that are harder to
measure or verify. Compromises are inevitable, including the use
of (say) gender or occupation as a proxy for other risk factors
such as lifestyle-related ones. Variations between the practices
of different insurers sharpen competition and penalise poor-quality
practices in assessing risks.
2.9 It is of course the prerogative of Government to
decide that specific practices in pricing risk are contrary to
the public interest and should be proscribed. (The controversy
over the use of genetic test results is an example of concerns
which might lead to such a decision in the absence of responsible
conduct by insurers.) Any decision to limit the use of risk pricing
criteria is likely to raise the overall cost of insurance, for
one or more of the following reasons:
— insurers have to build in margins against an adverse
mix of higher and lower risk cases which they can no longer distinguish,
as noted by FSA; or
— higher-risk individuals are enabled to enter the
market to buy cover at a subsidised cost; or
— given the intangible nature of insurance, customers
who would be overcharged may instead buy cover from insurers in
other jurisdictions that are not subject to the same constraints.
The social benefit of such a proscription must be weighed
against the additional cost of insurance that could be expected
to follow.
2.10 It is our contention that the use of gender as an
objective criterion for the pricing of life assurance and annuities
is justifiable because it reflects real biological differences
and because it acts as an effective proxy for other risk factors
which could not be adequately captured through other selection
criteria. Without it, the overall cost of these products would
rise.
3. GROUPED INSURANCE
AND ANNUITY
BUSINESS
3.1 The proposed Directive might simply have required
that men and women buying insurance or annuity products should
have access to the same price and terms, assuming no other relevant
differences between them. Instead, it specifies that:
— the requirement for equality extends to all purchasers,
not just individuals (Article 1); and
— the use of sex as a factor in the calculation
of premiums and benefits for insurance and related financial services
is prohibited (Article 4).
3.2 The issues arising for grouped business, in which
Legal & General has a substantial presence, differ from those
for sales to individuals. The two largest markets involved are
group life assurance and related benefits sold to companies or
to trustees of company pension schemes for the benefit of employees
(2003 premiums £1.3 billion) and bulk purchase annuities
to buy out the liabilities of pension scheme trustees for present
and past employees (2003 premiums £1.2 billion).
3.3 In both cases, the benefits to individuals are nearly
always formula-driven and existing legislation prohibits the use
of differing formulae to reflect the different costs of providing
them to male and female employees. So the requirement for equal
prices will not benefit one gender at the expense of the other.
3.4 Also, the proposed Directive will cause insurance
or annuity providers in the individual market to assess the relative
costs of providing benefits to men and women and make a cautious
assumption about the proportion of each gender who will buy the
product (the element of caution bringing with it an overall increase
in cost to customers). Experience will lead them to vary that
assumption as necessary. For grouped business, although insurers
generally know the sex of each individual, they will not be allowed
to use that information in calculating the single price that will
apply for the whole group. Some employers have populations of
current or past employees which are for a variety of reasons predominantly
male or predominantly female. Given the much larger case sizes
involved, the margin of caution needed in setting an assumed gender
mix in pricing must therefore be higher than for individual business.
3.5 In the case of bulk purchase annuities, there are
two other particular factors:
— An employer with a large and ongoing pension scheme
is under no compulsion to buy annuities at all, but might, if
the terms were sufficiently attractive, buy annuities only for
female members of the scheme. This would put further upward pressure
on bulk purchase annuity prices.
— It is commonly the case that a limited pool of
funds is available to trustees to purchase the highest possible
proportion of pension entitlement they can in the absence of an
employer who is able to inject additional funds to meet the full
entitlement. An increase in the cost of provision would further
reduce the proportion of full entitlement that could be afforded.
3.6 In conclusion, extending the proposed Directive to
grouped business would bring benefits to consumers only in quite
unusual circumstances, and would be likely to increase the cost
to employers or to reduce the benefits for employees, in ways
that we suspect the EU Commission has not considered.
April 2004
|
