Hitherto not necessarily wholly foreseen constraints on equality

“The judgment ignores the fact that taking a person’s gender into account, where relevant to the risk, enables men and women alike to get a more accurate price for their insurance.”

1. But then as the price gets more accurate, it becomes less like insurance and more like a savings scheme.

2. Just as arguments based on aligning tax burden with expenditure benefit (e.g. recently tuition fees) defeat the counterfactual-insurance (interpersonal redistribution of the luck of birth) function of the welfare state.

3. Pensions are an income stream which needn’t (and can’t, see above) be based on individual pension pots.

4. something to do with Gattaca

Ah, so, absent the annuity market, would anybody benefit?

Women buying private health insurance. That is, they’ll get equality. The ruling might make men more reluctant to get private insurance, thus making the overall risk pool worse.

Isn’t the whole freaking point of the insurance industry to discriminate?

Not necessarily. Pooling risks is usually a lot safer than discriminating against them. In general, when insurers discriminate in pricing, it is not because some actuary somewhere found a great way of doing so; it’s because their customers started discriminating first, by overbuying some kind of underpriced insurance. Nearly all of the data-mining that insurers do is of this ‘defensive’ kind – to identify potential adverse selections rather than to create new risk distinctions.

I’m dimly remembering the Goodin, Le Grand et al thesis that wartime uncertainty undermined the possibility of insuring against various risks and thereby let to the growth of support for more comprehensive risk-pooling via the welfare state. So is there a case for saying that IF political or legal factors undermine insurance markets (or to the extent that they do) there will be a corresponding and countervailing impetus to make state provision? (Which may, of course, be outweighed by other factors.)

Pooling risks is what the insurers can do that the customer (even the perfectly rational and informed one) can’t. Even with an ideal perfectly accurate individual profile (i.e. level of risk is known exactly, based on all relevant individual features), if there is any risk then there is still no guarantee that the payout will match the premiums (minus the vig) over the lifetime of an individual contract (or even of the insuree).

(If it did, that would be pooling risk between timeslices of the insuree – which I expect can be shown to be equivalent to saving/borrowing money as required, if one can make certain wild assumptions about availability of credit, profit rates of insurers,lenders, borrowers, etc. But much insurance is not like that – the payout may vastloy exceed the payees lifetime premium payments).

The way dsquared describes insurer’s behaviour sounds plausible as a kind of market behaviour – as the bookies (should) change theior odds according to the bets they have recieved, ideally so that they can never fail to make a profit – a distributed dutch book manoeuvre. Bu then there is the question of where a price setter kicks that process off.

Dsquared argues that most discrimination by insurers is reactive, rather than proactive. I think he was probably correct in the past, but this is increasingly less true, as insurers are getting smarter. Progressive Insurance, for example, is extremely proactive in price discrimination. They view superior price discrimination as their competitive edge. You can see it in their ads, where they promise that they will try to palm the higher risks off on their competitors. And of course, the reinsurance market has always priced proactively.

In the traditional view of the financial services industry, investment bankers were the smart guys, commercial bankers were kind of dumb, and insurers so dumb that they outsourced all mental functions to their lawyers and actuaries. (AIG was always viewed as the exception to this rule.) Nowadays, commercial bankers are mostly investment bankers, and many insurers are getting smarter.

btw, in the States, the big discrimination issue of the day is credit scoring. Insurers think they have found a correlation between credit scores and risk, or at least propensity to file a claim. Of course, credit scores are correlated with many other things, which raises the usual social justice issues.

This is a much more common feature of industry marketing material than underwriting practice

viz, for example, “Sheila’s Wheels“, which trades massively on the “women have fewer accidents!” factoid, but actually it doesn’t (because how could it) make any particular use of this in pricing – it is simply a (really, really embarrassing) marketing skin over the underlying esure business, and it prices by type of car and NCB. On average, adult women drivers will get cheaper insurance than average male drivers, but this is because the average woman driver has a better average no-claims bonus.

Although it does look like something that would be the occasion for a massive bollocking and multiple “You’re Fired” if it ever showed up on The Apprentice, I can confirm that Shelia’s Wheels is a real trading brand of esure plc and thus (via a chain of ownership through Lloyds Banking Group) one of the ways in which the UK taxpayer hopes to recoup some of his bailout cash.

Surely if it’s understood that women live on average longer than men then the sensible solution is to increase the retirement age of women to beyond that of men…

Pretty soon we’ll be seeing annuity-selling companies with websites (and retail outlets) specifically designed to repel women.

A new career awaits me …

Ok … just had a chat with a pensions economist (in the street). From what I recall of the conversation … (filtered through my poor understanding)

1. The ECJ decision has the effect of mandating for one type of financial product what would constitute illegal discrimination for another. I.e. if Tesco offered higher returns to women on for some financial product that would be illegal, but that’s the effect re annuities.

2. The taxation of annuities in the UK is based on a division between capital repayment and interest components which, in turn, depends on an actuarial calculation. That’s now rendered problematic.

3. The ECJ has a history of disastrous judgements re private pensions and often (as in this case) the judges who make the decision all come from countries without private pension industries. Past decisions require that governments guarantee 90% of value of private schemes which introduces massive moral hazard as companies take advantage of insolvency law (pre-pack) to dump their pension liabilities on the state. (Government guaranteeing private debt … reminds me of Ireland.)

4. The real pension scandal at the moment is the Hungarian government grabbing 11 billion euros from private pensions to wipe out (on paper) its deficit. The EU is doing bugger all about this act of state robbery.

Anyone care to correct or elaborate on those points?

Yes but Sam, surely the point is that “you may not consider sex” is being interpreted in a manner which _mandates_ offering higher returns to women wrt to one financial product (an annuity) but _prohibits_ in wrt to another (such as an ISA). And that looks perverse and arbitrary.

@26: But any product whose return is measured by someone’s date of death is going to have highly variable returns for the individual buyer anyway. Does it really matter if the distribution of actual returns doesn’t precisely match a theoretical calculation? Individuals are mainly going to be concerned about the possibility that they may be in a tail, anyway.

In any case, part of the problem may be sending insurance to do social insurance’s job. Despite the similarity of names, the goals are quite different, particularly in regard to how to allocate the costs.

@29: Sure I do. There’s a reason Social Security isn’t a savings plan, was never designed as a savings plan, and can’t be converted to a savings plan without altering it beyond recognition, and it’s not just the desire to start paying benefits immediately for people who were already 64 when it passed. People who die younger subsidize the benefits of people who live longer, but this is accepted because they don’t need the money and can’t take it with them.

“Since that seems to indicate they aren’t actually offering anything meaningfully different to women drivers just for being women, why complain about how the ECJ ruling will cause “premiums to rise artificially in a way that no longer truly reflects women’s risk as drivers or the cost of their claims”?”

The insurer has to protect itself and those insured with it against ruin.

If you know a copper coin comes up heads 1/4 of the time and pays £1 if this happens you can price the risk at £0.25. If a silver coin comes up heads 3/4 and pays £1 if this happens you can price the risk at £0.75. Someone who only offers bets on silver coins wouldn’t offer anything meaningfully different to someone who offers them on silver and copper coins, as they’d both condition on the appropriate risk, just Sheila’s Wheels doesn’t offer anything different to any other company.

Now, if you can’t ‘discriminate’ between silver and copper coins what rate do you offer? You have to protect yourself against being flooded with bad risks. If you price near the copper risk (receive £0.25/go), and get lots of silver coins taking bets (pay £0.75/go) you’ll be bankrupt and you won’t be able to pay out on some contracts. To guard against this you’ve going to have to raise the prices for copper coins away from the true risk – to protect 100% (against getting all silvers) you’ll have to charge £0.75 to everyone.

You’ll make a nice profit should you insure lots of coppers, this inefficency is good for you and bad for people wanting insurance. But, if you don’t do this, you can’t be sure of meeting your obligations – also bad for people wanting insurance. You can see why people are pissed off, the ECJ has pointlessly caused a lot of damage.

In practice, the inefficiency will crowd out people at the margin: some women won’t drive who otherwise would. Because Sheila’s Wheels have a very gender biased pool of insured, and because they’ll all be facing substantial premium rises, they’re in a tricky situation. There’s a lot of uncertainty: if they price wrong, their pool of insured shifts, or volume falls off, they’ll be in a lot of trouble, at the other end of the spectrum someone who insures only men doesn’t have to worry too much.

2. The taxation of annuities in the UK is based on a division between capital repayment and interest components which, in turn, depends on an actuarial calculation. That’s now rendered problematic.

Not working as a life company actuary I’m not sure exactly how much of a problem this really is but I doubt it is much of one. On a practical level SamChevre points out you just use a set of unisex mortality rates in the calculation.

Another reason I doubt its a huge problem is that I suspect most annuities purchased from insurers don’t even require this kind of calculation for tax. They will have been bought using money from people’s occupational or personal pension plans (which had tax relief on the contributions paid into them and tax relief on most of the investment returns) and therefore income tax is payable on all the payments from the annuity.

The kind of calculation you talk about is, IIRC, applicable only where you are buying an annuity from savings that are not in a tax exempt pensions savings vehicle.

@ Chris:

@26: But any product whose return is measured by someone’s date of death is going to have highly variable returns for the individual buyer anyway. Does it really matter if the distribution of actual returns doesn’t precisely match a theoretical calculation? Individuals are mainly going to be concerned about the possibility that they may be in a tail, anyway.

Well if I extend your argument you could say that age shouldn’t factor into the calculations either then. Because charging men and women the same rates is something like charging a 65 year old the same annuity rate as a 67 year old. Now sure the 67 year old may live another 40 years and the 65 year old only 5 more years but you wouldn’t bet on it and it would still be perverse.

I can almost see someone lining up to try and push through an age discrimination claim now.

Off topic but one thing to note is that due to the way life offices use postcode/size of your pension pot as risk factors, the poorer you are the better the annuity rate you are likely to get. This is due to the massive gap in life expectancy between the rich and poor. One hopes there isnt a judgement round the corner banning this practice.

Finally I wouldn’t be a real actuary unless I added some disclaimers: the views expressed here are my own and not those of my employer or the actuarial profession etc.

“Not if the only choices are to buy at those rates or not at all. Not buying at all leaves you at risk of running out your assets and plunging into abject poverty at 85, if you live that long.”

It doesn’t. You could just maintain the capital and live off the interest. An annuity’s an improvement on that, but not much of one. You don’t have to misprice much before people will take out bonds rather than annuitise and leave their kids the cash.

You’ll make a nice profit should you insure lots of coppers

Are they really safer drivers, though, or do they just tend not to end up in court?

You don’t have to misprice much before people will take out bonds rather than annuitise and leave their kids the cash.

Citation needed, I think.

A quick look at the decision suggests that this was the key moment:

30 It is not disputed that the purpose of Directive 2004/113 in the insurance services sector is, as is reflected in Article 5(1) of that directive, the application of unisex rules on premiums and benefits. Recital 18 to Directive 2004/113 expressly states that, in order to guarantee equal treatment between men and women, the use of sex as an actuarial factor must not result in differences in premiums and benefits for insured individuals. Recital 19 to that directive describes the option granted to Member States not to apply the rule of unisex premiums and benefits as an option to permit ‘exemptions’. Accordingly, Directive 2004/113 is based on the premiss that, for the purposes of applying the principle of equal treatment for men and women, enshrined in Articles 21 and 23 of the Charter, the respective situations of men and women with regard to insurance premiums and benefits contracted by them are comparable.

31 Accordingly, there is a risk that EU law may permit the derogation from the equal treatment of men and women, provided for in Article 5(2) of Directive 2004/113, to persist indefinitely.

32 Such a provision, which enables the Member States in question to maintain without temporal limitation an exemption from the rule of unisex premiums and benefits, works against the achievement of the objective of equal treatment between men and women, which is the purpose of Directive 2004/113, and is incompatible with Articles 21 and 23 of the Charter.

33 That provision must therefore be considered to be invalid upon the expiry of an appropriate transitional period.

That makes it look as though the judges would have been more interventionist had they agreed to an indefinite exemption than in fact they were – i.e. that the legislation basically reasonably clearly mandates this result.

Which at least renders redundant the preceding two paras of what looks like gobbledigook that is uspposed to address the issue whether the benefits supplied by an annuity are the same (‘copmparable situation’) despite differing life expectancies.

Thank you for the link Tim. The arguments in the judgment suggest that the case was decided on technical grounds, but the advocate general’s submission contains more substantive reasoning (although one does not know how it was received). Here is the crucial bit:

60. That restraint on the part of the Court must be connected with the prominence which the prohibition of discrimination on grounds of sex has in European Union law. Direct discrimination on grounds of sex is… only permissible if it can be established with certainty that there are relevant differences between men and women which necessitate such discrimination.

This is basically the same as the U.S. Supreme Court’s reasoning in Manhart (“Even though it is true that women as a class outlive men, that generalization cannot justify disqualifying an individual to whom it does not apply.”) There seem to me compelling reasons to prohibit discrimination on the basis of mere statistical association in the standard sort of civil-rights cases (e.g. Bowe v. Colgate-Palmolive), but not in the case of insurance. The requirement that there be certainty (e.g. of date of death in the case of life annuities) is completely incompatible with the logic of insurance contracts. If you knew when you were going to die, you wouldn’t need to buy an annuity, you would just save.

To the extent that there are good arguments for prohibiting risk-classification, it seems to me that they are to be made in a second-best framework. For example, this paper by Walter Bossert and Marc Fleurbaey (“Equitable Insurance Premium Schemes”) is awesome and important: http://www.springerlink.com/index/nmtlm7l5w81r71xv.pdf (gated, sorry). These issues haven’t been sufficiently explored, however, in part because of the insistence that risk-classification constitutes a form of straightforward discrimination (and thus there is no need to appeal to second-best considerations).

Yeah the decision reads to me as though the point is very much limited to using sex per se as an input to the sums determining rates. It doesn’t mandate equality: if some factors more specific than mere sex can be found that can explain (and thus displace the need to advert to) the correlations between e.g. sex and life expectancy, there would be no problem I reckon. But the stuff about specific mechanisms is pretty alien to the whole actuarial project, which is at bottom just about correlations.

(Myles, as indicated, the apparent gobblidgook is the first two paras of this bit:

27 The Council expresses its doubts as to whether, in the context of certain branches of private insurance, the respective situations of men and women policyholders may be regarded as comparable, given that, from the point of view of the modus operandi of insurers, in accordance with which risks are placed in categories on the basis of statistics, the levels of insured risk may be different for men and for women. The Council argues that the option provided for in Article 5(2) of Directive 2004/113 is intended merely to make it possible not to treat different situations in the same way.

28 The Court has consistently held that the principle of equal treatment requires that comparable situations must not be treated differently, and different situations must not be treated in the same way, unless such treatment is objectively justified (see Case C‑127/07 Arcelor Atlantique et Lorraine and Others [2008] ECR I‑9895, paragraph 23).

29 In that regard, it should be pointed out that the comparability of situations must be assessed in the light of the subject-matter and purpose of the EU measure which makes the distinction in question (see, to that effect, Arcelor Atlantique et Lorraine and Others, paragraph 26). In the present case, that distinction is made by Article 5(2) of Directive 2004/113. )