A Curious Miscellany of Items Philosophical, Historical, and Literary

Manus haec inimica tyrannis.

Monday, April 25, 2016

Gambling with 23andMe

Up until recently, I was interested in signing up for 23andMe and getting a genetic analysis done. My main reason for wanting to do so is that I was interested in finding out about my heritage, information about how much of me is Celtic or Scandinavian, Italian/Spanish or Arabic or Jewish or Neanderthal or God-knows-what — the more surprising the better. However, I have begun to rethink things. 23andMe would not just provide information on ethnicity. It would also give me quite detailed information on my personal genetic profile of a quite specific and medical nature. This is information I am not so keen on getting.

I am even less keen after reading an insightful and personal newspaper opinion piece. The author narrated her experience with 23andMe. Like me, she was curious. She admitted she knew there were risks (more on that later), but her curiosity got the better of her and she sent away for the kit. She found out various interesting things about herself, such as the fact that she is 2.7% Neanderthal. However, she also found out that she carried two copies of a gene that gave her a 51 to 68% chance of developing Alzheimer’s before age 85. The latter is the kind of knowledge I’d rather do without, for a few reasons.

First, in the absence of a cure for diseases such as Alzheimer’s, knowledge is less likely to be power and more likely to be a source of worry and neurosis. I have to ask myself seriously whether I am prepared for the personal consequences of such knowledge. I suspect not.

Second, those consequences extend beyond myself and may not be entirely personal. As the author of the piece notes, the Alzheimer’s knowledge put her in the dilemma of having to decide whether or not to tell her parents; after all, each of them must possess at least one copy each of the gene.

Obviously, there are many people who do not share my misgivings (or have not reflected on them enough) and are signing up for 23andMe’s service. Judging from the copy on their website, 23andMe’s testing is touted as being able to tell me about:

•    Inherited Conditions
•    Drug Response
•    Genetic Risk Factors
•    Traits (“Explore your genetic traits for everything from lactose intolerance to male pattern baldness”).

Notably absent from this list is precisely the sort of information I was mainly interested in: ethnicity and genetic heritage. It presumably offers some of that too (e.g. Neanderthal genes), but their website and television commercials make clear that they are aiming primarily at end-users concerned with their health and medical profiles.

Despite the troubling information she received about her Alzheimer’s susceptibility, the above-mentioned author says that she is still glad she signed up for 23andMe. With respect, I’m not sure she should be glad. To its credit, 23andMe’s Canadian website is quite forthcoming about a risk of their service that for me is a deal breaker: Under current Canadian law, there are no legal grounds for believing that the privacy of your test results is safe from life insurers. As the website states: “Currently there are no protections for Canadians against discrimination based on their genetics - life insurance companies and employers may request an individual's genetic information or may ask whether an individual has had a genetic test. Currently, there is Canadian legislation under review (Bill S-201) to protect genetic information. 23andMe fully supports Bill S-201 and will continue to advocate for legislation and other actions that can protect Canadians from discrimination in insurance and employment decisions on the basis of genetic information.” Until such protections are in place, the author had better hope that her life insurer has not read her column.

Insurance: Bargain or Bet?

23andMe says it is actively involved in getting the laws here changed so that their customers are not at the mercy of nosy insurers. This brings up a question: Should insurers have the right to see such information? My gut reaction is to say “no”, mainly on the basis of the private nature of the information. But that just sidesteps the issue, by bringing up another question: “Should such information be considered private in this case?”

There is a sense in which the insurers have a point. Consider any other kind of bargain or sale. There is a popular view out there that the old common law’s position on sales was caveat emptor. However, the caveat emptor rule has always been more myth than reality. For instance, vendors are liable for defects in their goods for which they are aware and for which the buyer does not have the opportunity to inspect for herself (if she has had the opportunity to inspect and has simply overlooked the defect, that’s another thing). And with certain important purchases such as a house, I have a positive duty to disclose important known defects even if the purchaser has had the opportunity of inspecting the house.

 By that principle, it seems that in the case of a purchase of life insurance, if I am holding information acquired from 23andMe that says in effect that I am “damaged goods”, and I do not disclose that information to the insurer, the insurance policy ought to be void. If the information is too “private” to disclose, then perhaps I ought not to have purchased insurance in the first place. Ultimately, I am not forced to purchase life insurance. Insofar as I choose to, should I not be subject to the same legal and ethical requirements as any other voluntary bargain?

So far, it’s not looking good for 23andMe and their customers. However, I’m tempted to approach this from another angle. Since the “appeal to privacy” seems to be shaky, perhaps we could instead question whether or not it’s appropriate to characterize the purchase of life insurance as a straightforward bargain or sale. There are elements of the “product” being “sold” here that make it more akin to a bet or gambling contract.

When I take out a life insurance policy, both the insurer and I are placing bets on the timing of a certain event, namely my death. I am betting that this event will occur before a certain time t. What I am staking on the bet are the premiums I pay for the insurance. My prospective payoff would be what the insurers must pay out if I die before t. At first glance, it seems strange that I would place a bet in hopes of an early death, but in reality the bet is a hedge — it offsets any losses suffered by my estate in the event of that unwanted event happening.

On the insurer’s side, the bet is that I will die after t. His stake is the agreed amount he will have to pay out if I die before t. His payoff is the accumulated (and invested) fund of premiums I have paid up until t. Ideally, t will be defined such that the accumulated premium payments are greater than the ultimate payout the insurer would have to make at that point, the difference representing profit. (Alternatively, it could be an all-or-nothing bet: if after t passes I am still alive, the insurer pays nothing.)

Put this way, insurance sounds very much like a wager. The impression is strengthened by looking at the payoff structure of the transaction. An ideal voluntary bargain displays what is called Pareto improvement, which is a fancy way of saying that the deal makes at least one party better off without making anyone worse off. I pay you $5 and you pour me a beer; I value the beer more than $5 at that moment, and you value the $5 more than the beer. The trade makes each of us better off, since we each acquire something of greater value to us.

Is the insurance wager – or indeed, any wager – Pareto improving? Arguably, no. Most wagers are zero-sum. There must be someone who wins and someone who loses, and if it is a pure zero-sum transaction, one party’s winnings will exactly correspond to the other party’s losses. If I bet you $10 that it will rain tomorrow, and tomorrow it rains, I gain $10 and you lose the same amount. In the case of life insurance, if I die before t, I “win” and the insurer loses. If I die after t, the insurer wins and I lose. Regardless of which outcome happens, we will not both be better off, at least not financially. One of us wins and the other loses.

In reality, insurance contracts seem to be hybrid in nature, having characteristics of both bargains and bets, but more of the latter. They are like bargains in the sense that both parties – at least subjectively – get some value out of the contract: the insurer makes a profit, at least statistically. The insured person, meanwhile, gets peace of mind. So it seems that once one expands the notion of payoffs beyond mere dollars and cents, this is not a purely zero-sum transaction. Of course, this is rather trivial, since it is much the same with any kind of wager: financially, there will be a winner and a loser, but both parties get to experience the pleasant titillation that comes from betting. People who gamble enjoy gambling (for the most part). That must be factored into the payoff structure of the bet. However, depending on the bet, it is questionable whether the pleasure comes at too high a price. In short, the bettor might still be acting quite irrationally. Problem gamblers get a momentary thrill from gambling, but the thrill seems outweighed by the loss of home, family and credit. Except for very small bets, or perhaps in rare examples of Kipling-esque “great-souled” gamblers, I would argue that most wagering is irrational.

Returning to life insurance, it is worth noting that the insured person’s peace of mind is only gained because (i) he lacks information about a future event, and (ii) he is risk-averse, perhaps to an irrational extent. If he had information, and the information was that he was not at risk of premature death, he would not agree to the insurance bargain, because he would stand to gain nothing and his money could be spent elsewhere with more utility. On the other hand, if the information was that his death was going to be premature, he would purchase much more insurance than he otherwise would. As things stand, he is likely either over- or under-purchasing insurance.

As for risk-aversion, if the insured had a realistic attitude towards risk, he would only sign a contract on terms that the insurer would refuse, because insurers base their contract terms on the proposition that, taken in aggregate, they will receive more money than they will eventually have to pay out. In other words, they profit by the fact that people are willing to pay more for insurance than the actual risk says they rationally should. Their main source of profit is essentially the irrationality of those they insure. In a sense, an insurance company constructs a Dutch book. They know what the odds are of a certain kind of event’s happening. They know that people tend to overestimate those odds, and they take advantage of this weakness by offering terms that ensure that they win in the long run, even if they lose in the odd individual case.

(When I refer to “irrationality” here, I don’t mean that it’s outright irrational to purchase life insurance. After all, odds are just odds, and the improbable in this case happens just often enough that one may find it worthwhile to hedge against it, at least in terms of the concomitant peace of mind one gets from it. My risk-aversion may be irrational, but it is also a basic psychological fact that I must take into account. It’s a fact that makes the wager worthwhile to me. But from a strictly financial point of view, my money would be better spent or invested elsewhere.)

If I have succeeded in characterizing life insurance as a gambling wager rather than a straightforward bargain of sale, then the conversation around the duty of disclosure can be seen in a different light.

First, it’s worthwhile noting that, with a few exceptions, the common law will not enforce gambling contracts (though it so happens that insurance is one of those exceptions).

Second, as we saw, in a bargain the seller has a legal and moral duty to disclose any defects in his goods that he is aware of and that the purchaser has not the opportunity of inspecting for herself. This duty does not typically apply to gambling wagers. I may take advantage of knowledge I exclusively hold and place a bet without disclosing that I have such knowledge. This is considered fair dealing in wagering, but not in contracts. Gambling is ipso facto taking a risk; a contract is not. In gambling we are supposed to hold our cards close; in contracts, we are expected to lay them out on the table if they affect the other party.

If life insurance is essentially a wager, then I should be entitled to keep whatever I find out from 23andMe to myself, not just on grounds of a supposed right to privacy, but because the insurer simply has no claim on it.  An insurer constructs a wager based on the statistical probability of an event’s happening, and precisely because it’s based on statistical probability, they can offer that bet to all comers. In each wager they accept a degree of risk, but taking all their policies together, their venture is relatively low-risk, assuming they’ve constructed their wagers properly. In demanding full disclosure of information, it seems insurers want to make their particular brand of gambling risk-free. The law might say they have a right to full disclosure of what they deem pertinent information, but morally they are being rather poor sports.