2011/06/14

Who is the fairest in the land?

The European Court of Justice ruled in March of this year (2011) that discrimination between genders in all forms of insurance is illegal. This rule came as a shock to me, but it has been a long time in the making, since at least 2003. When it comes to fairness, Europe makes even the Rainbow Nation seem like a laggard.

What is fair?

It could certainly (and I would do so) be argued that differentiating between men and women for insurance purposes is exactly the fair thing to do. As a man I am NOT buying the same product as a woman, because my risk is different. It would be unfair to make someone else pay for my risk (or me to pay for someone else’s risk). However, the EU feels differently.

The advocate-general (AG) of the EU released a legal opinion before the final ruling was made, explaining why she thought this ban should be effected. Here the fact that the equal treatment means different situations should not be treated in the same way was mentioned. Curiously, the law in question allowed states to choose whether to allow differentiation (I am going to use the word ‘differentiation’ rather than ‘discrimination’) between men and women or not, which does result in a violation of the principle.

However, the law also remarks there should be no difference in premiums due to maternity and pregnancy. If this does not ALSO violate the principle and, by implication, any insurance that treats effects related to pregnancy with the same costs for both men and women, then I do not know what does. The justification given is that every pregnancy involves a man – because they caused it they should finance it. This seems silly. Many men do not have children and men can have different numbers of children. Why should they pay for the children of other men? This is a really arbitrary justification, in my opinion.

Direct discrimination (that is charging different rates) is allowable if it can be established that there are relevant differences between men and women that necessitate such discrimination. The EU, from my understanding, has decided that the situations of men and women are ALWAYS comparable, which is inherently nonsensical. Differentiation by race and ethnicity is already banned. I would argue that, if race is a direct determinant of risk (it appears not be a significant one), such differentiation should also be allowed.

The AG is of the opinion that the fact that sex and race (unlike, say, age) is not subject to natural change (you have, with some negligible exceptions, no choice over your race or gender) is a justification for discrimination not to be allowed here. This is because, over the course of one’s life, the premiums charged would vary by age for everyone. Over the course of a lifetime all will have been treated equally. This still does not address the fact that we are treating different situations in the same way.

What is, perhaps, justified, is the idea that a statistical link between sex and risk is not enough to establish grounds for differentiation. That is, just because women happen to have fewer car accidents, this may not be because they are women, but because of other factors. If we could measure and account for these other factors we would not need to use sex as a rating factor. However, we need statistics even to measure these other factors. Some framework for statistical justification is therefore necessary. As far as mortality is concerned, the statistics are clear: women live longer. However, it is not clear why. There have been periods in history that women have had both shorter and longer life spans than men. However, it is not the job of insurers to explain why this should be.

The AG is correct in saying that lifestyles, habits, and predilections not determined by sex (rather by choices) should be examined. However, this ruling has now (if I understand correctly) made it impossible, even if it can be proved that, controlling for all these factors, there is a significant difference between the sexes, to differentiate on such grounds. That said, these factors are difficult, if not impossible, to accurately assess statistically.

On the one hand we have that because men and women are inherently different (born different with no choice) we should not discriminate between them. And on the other, because they are not necessarily different (we just have this statistical link) we should discriminate. This is not particularly logical.

The implications

The implications are not as severe as one would think. The uproar by actuaries appears to be more principle-based (because we are taught basically since birth that rating products according to risk is a biblical principle) than because the insurance industry will suddenly fail. Swiss Re, just after the AG’s opinion was announced, said the balance of demand for insurance may alter drastically. I doubt this will prove to be the case. Yes, insurance premiums are likely to go up (not just for previously “low risk” groups, but on average). But will this make a huge difference to anyone?

When discussing the consequences we need to consider both the short term ones (before the ruling gets implemented) and long-term (afterward).

In the long term insurers will no longer be able to charge a different price for males and females (obviously). This means the price they use will be based on an average risk or expected cost for a mix of males and females. Pricing will become just a little (or a lot) more complex (ironically, decreasing the number of rating factors increases complexity). Insurers will need to assume beforehand what mix they will get. As they might get it wrong, this means they will tend to be conservative. For instance, they will assume that more males will take out life assurance as it will be relatively cheaper for them. Prices would also increase because insurers are likely to need higher capital (due to increased uncertainty). They will charge for the cost of this capital.  Average prices are thus likely to increase; insurance will be more expensive overall, although some groups will get cheaper insurance.

Prices of annuities should increase for males and fall for females. This is because women live longer and thus the insurance company will expect to make more payments on the annuity. Until now this has meant women pay more for their annuities than men; the new price will settle somewhere in the middle. Men will subsidise women. Car insurance rates should increase for women (who are, allegedly, safer drivers) and fall for men. Here women will subsidise men. The rates for life insurance will tend to increase for women and fall for men (again because women live longer). Women subsidise men again. There is thus no clear winner here. Both sexes win and lose (and lose overall).

The standard theory in such cases says that demand for insurance for groups for whom insurance becomes relatively more expensive will fall and will increase for those for whom it becomes cheaper. For instance, car insurance will now become more attractive for men, whereas women may choose to forgo getting cover. The extent to which this happens means that people who would have otherwise have enjoyed cover (at fair rates) will now no longer be covered; it will also push up premiums further as the mix of business will be skewed toward the higher risk groups. Gen Re does not expect the extent of this effect to be very great, at least for life insurance. I suspect for car insurance the effect might be greater. 

The increase in costs may be significant for some classes (perhaps 25% for motor insurance for females). However, for insurance such as term assurances, where the premium is very small compared to the sum assured or in comparison to the insured’s income, even quite large increases may not be keenly felt. There are also classes of insurance (such as those that are primarily savings products) where the difference in riskiness between genders makes little difference to the premium in any case.

Insurers are likely to try to find other ways of classifying risk, in lieu of being able to use gender. That is, they will try to evaluate things such as habits and behaviour. Insurers will, however, need to be careful of indirect discrimination, which is using things such as height (women are shorter) or length of certain fingers (for men the index finger tends to be shorter than the ring finger) to indirectly identify women. Already car insurance is planned that will make use of a device that measures the driver’s riskiness by monitoring the car’s movement. This is a novel design that should really have been implemented even without the ruling – a rare case, perhaps, of regulation causing innovation.

I personally feel that a kind of equilibrium will be reached in which each insurer has a similar proportion of men and women for a particular type of insurance (except where there is no tangible difference in risk between the genders). If a company is overweight on whichever gender is less risky, its rates are likely to be low and it will attract riskier clients (even if it only attracts more clients, its business mix should move toward the market average). Another possibility is that insurers will compete on the mix of business they attract. Those that attract the least risky population (through marketing) can charge the lowest premiums. I have my doubts as to whether such a strategy can work in the long term.

Some insurance companies such as Sheila’s wheels in the UK sold only to women. This will no longer be allowed. The company has indicated it intends to continue to market to women (though obviously it cannot exclude men). It remains to be seen if this will succeed. Apparently they have never turned away men in any case (though I dare say they never needed to and would have charged them appropriately).
In the short term one will expect the subsidising gender to buy more insurance and the subsidised gender to wait for the prices of products to move in their favour. That is, for instance, men will buy annuities now, while they can still get it cheaply, and women will wait until they can get annuities subsidised by men. Prices may be volatile as this demand fluctuates.

There is also a lapse effect, in that, for example, male life insurance policyholders may lapse just prior to or after the enactment of the new law as they could get cheaper rates then. This is only temporary, but is likely to cause losses for insurers.

Where does the last domino fall?

Even as I write this, the debate rages on. Now that gender has been thrown in the river, the sluice gates stand ready. Why do we discriminate by age and health? Perhaps these should go too. And this would have an effect orders of magnitude bigger than current ruling as age and health are much more significant determinants of risk. The EU is currently in the processing of debating this issue. Let us hope the its sees sense this time. There is some hope that age, at least will be allowed. However, using the AG’s own reasoning, we should charge people with cancer any more for insurance (they had no choice in the matter). I cannot see insurance still being viable if this happens. I am not sure how to avoid the last, ridiculous domino from falling.

The ridiculous

As I understand the principle correctly, as a man, I can now buy a policy that covers ovarian cancer and it would be illegal to charge me a different premium. It would also be illegal to refuse to let me buy the policy. It is not clear what happens with policies that cover gender-specific diseases. Would it even be legal to cover, say, ovarian cancer, but not prostate cancer (which would automatically attract a certain gender)?

It is not clear how to handle gender-specific underwriting either, e.g. family history of breast cancer. Again, it would be rather silly to charge me extra because my mother had breast cancer. Gen Re seems to think the principle of not treating different situations in the same manner will save them here. However, as it has been so blatantly ignored already, I am not so certain.

It is common to differentiate by occupation for income protection covers. This rating factor could be used for other kinds of insurance as well. This clearly differentiates risk between people. However, as there are some occupations that are closely linked to gender, I am not certain whether this would not constitute indirect discrimination.

If the judgement also prohibits insurers from asking for the gender of applicants, this would create a serious problem. Insurers need to monitor their business mix or they will be unable to charge the correct premiums. Internally, insurers will also analyse the claims experience of men and women separately. It would be very hard to charge reasonable prices without doing this. Fortunately, there is no indication that the law implies such a requirement.

In group insurance it is common to charge a single rate based on the composition of the group, e.g. by age and gender. An extreme would be if an insurer were to insure to groups of equal size, one all male and one all female. Should it charge both groups the same rate? Such a direct comparison will usually be impossible to make, but how are insurers now supposed to charge group business? A company with all female or all male individual business could quite legitimately charge an appropriate premium for those policies.

The ruling has created a quagmire of ambiguities. Ambiguities I believe are unnecessary and do not contribute to the good of the world in any way.

Utility

The question that must be asked, is “was it worth it?” I am a bit of a cynic, but if does not benefit society in some “real” way, then it was probably useless. Everyone can have the warm and fuzzy feeling (everyone except actuaries, that is) that discrimination has been rooted out, that we live in a fair society. But what else? Women have not really benefited as far I can see. They may even lose more than the men. Overall, society has probably not gained anything else, but somewhat more expensive (and ostensibly fair) insurance.
The “right to underwrite”?

Insurers love to quote this principle. This is what allows insurers to charge different prices for the “same” product. It is the defining characteristic of insurance, which makes it different from every other product. Insurance is not a cup of coffee. There is no difference between the coffee I drink and you drink. But the fundamental essence of insurance is that it changes with the risk insured. It should not be treated in the same way as a cup of coffee.

South Africa

Some actuaries in South Africa are of the opinion that similar laws will be enacted here as well. As South Africa is my home country this a particular concern of mine. South African law tends to mimic European Law with a lag. However, I hope the legal technocrats will take a more reasoned approach to the meaning of discrimination and the principles of fair treatment. South Africa has taken its own route in insurance in other areas (it introduced critical illness cover to the world, and uses income and education as rating factors, which may not be politically acceptable in Europe). I do not, however, expect this to happen again.

Final word

This judgement appears to have been born from a combination of poorly designed legislation (allowing arbitrary exceptions) and ideology. Unfortunately, this was an ideology that, instead of embracing and recognising differences between humans, ignores them. I would say the basic principle of equality that substantively similar situations should be treated in the same way and not others is a good one. It is sad that the court would espouse such a principle and then ignore it in the same ruling.


Some references

General articles for consumers

  • BBC, 2011a. Insurance and pension costs hit by ECJ gender ruling. BBC News Business. Available at: http://www.bbc.co.uk/news/business-12606610.
  • BBC, 2011b. Insurance gender ruling and you. BBC News Business. Available at: http://www.bbc.co.uk/news/business-12608777.
  • Burrows, B., 2011. Billy Burrows’ annuity update: March 2011. thisismoney. Available at: http://www.thisismoney.co.uk/pensions/article.html?in_article_id=524073&in_page_id=6.
  • Evans, T. & Hyde, D., 2011. How EU gender rule affects insurance and pensions. thisismoney. Available at: http://www.thisismoney.co.uk/insurance/article.html?in_article_id=524068&in_page_id=4.
  • Neligan, M., 2011. EU insurance gender ruling bad for consumers -UK min. Reuters. Available at: http://www.reuters.com/article/2011/03/17/insurance-gender-idUSLDE72G0Y820110317.
  • Sinner, M. & Neligan, M., 2011. EU court bans insurers from pricing on gender. Reuters. Available at: http://www.reuters.com/article/2011/03/01/us-europe-insurance-idUSTRE7201MU20110301.

Gen Re’s response

  • Webersinke, A., 2011. “An important moment for gender equality in the European Union” 1 or the starting point for many more differentiating risk factors to come? Risk Matters, 5(March 2011).

Statements by Swiss Re

  • Swiss Re, 2011a. Swiss Re disappointed with EU judgment on gender pricing. Available at: http://www.swissre.com/rethinking/reinsurance_regulations/Swiss_Re_disappointed_with_EU_judgment_on_gender_pricing.html.
  • Swiss Re, 2011b. Swiss Re voices concern over EU legal opinion on gender. Available at: http://www.swissre.com/rethinking/reinsurance_regulations/Swiss_Re_voices_concern_over_EU_legal_opinion_on_gender.html.

The judgement by the court and the AG’s opion (if you know how these should really be cited, please let me know):

  • European Court of Justice, 2011. C-236/09 Judgement. Available at: http://curia.europa.eu/jurisp/cgi-bin/form.pl?lang=EN&Submit=rechercher&numaff=C-236/09.
  • Kokott, J., 2010. C-236/09 Opinion. Available at: http://curia.europa.eu/jurisp/cgi-bin/form.pl?lang=EN&Submit=rechercher&numaff=C-236/09.

Sheila’s wheels’ response

  • Sheila’s Wheels, 2011. Statement from Sheilas’ Wheels on the European Court of Justice ruling. SheilaŹ¼s Wheels Media Centre. Available at: http://www.sheilaswheels.com/media/news/EUROPEAN_COURT_OF_JUSTICE_RULING.html.

Finger lengths:

  • Wikipedia, 2011. Digit ratio. Wikipedia. Available at: http://en.wikipedia.org/wiki/Digit_ratio.


No comments:

Post a Comment