Home
> Uncategorized > Insurance and Big Data Are Incompatible
Insurance and Big Data Are Incompatible
February 23, 2017
My newest Bloomberg View piece about how that FitBit could be bad for your health:
That Free Health Tracker Could Cost You
Categories: Uncategorized
mathbabe 
Well, Duh …
LikeLike
I kind of want to agree want to agree with you, but I don’t think most people understand this. It’s quite alarming, actually, how little people understand about things they interact with/use regularly. It’s the type of thinking that allows people to be happy with their ACA-subsidized health insurance while decrying Obamacare.
LikeLike
and now a word from our sponsor …
LikeLike
(OT): I noticed that “Weapons of Math Destruction” is listed as a supplemental reading to this course: http://callingbullshit.org/syllabus.html
LikeLike
I remember Zip Code and Address were (probably still are) used as factors in determining auto insurance rates. The thinking was that an address pin pointed a physical location in a community, and that house has a value, as do the others around it. The implication was that home owners who own property worth over a certain threshold could help determine the best rate. In an area where the housing prices are low? Might ding you. Own or rent? Might ding you if you rent. Own a house over a certain value? That’s a good thing, because the thinking is if you can afford the mortgage of the house you’re in, you’re probably pretty stable and safe, which would lower your rate. Is this Zip Code in an area where the median salary is higher than other Zip Codes? And on and on and on.
God only knows what’s in there now. Point being, whether you’re willing provide additional data or not, insurance companies will ALWAYS build a model that is to their advantage.
LikeLiked by 1 person
The issue isn’t the data, it’s that the insurer is currently legally allowed to use the data to price discriminate. Legally, they charge a higher base price and then credit some of it back to healthy people, but that’s functionally identical to charging people for being sick.
If you collected health data, and used it to direct people toward ways that they can take care of their health better (which generally will save both them and their insurer money), I don’t see any problem with that. But the cash “incentives” are a loophole in “community rating” rules that are supposed to ensure that insurance works as insurance — as a way for all of us to pool our risks together, such that the lucky (healthy) subsidize the unlucky. Just like folks with homeowner’s insurance whose house doesn’t burn down, subsidize those who suffer fires.
LikeLiked by 2 people
One could argue, quite convincingly, that health insurance should not guided by the profit motive. Unfortunately, for this time, we have lost that argument. Rating risk is at the heart of profitable insurance schemes and fine tuning that risk to require those with greater need to share more of that risk IS the insurance model. Still your point is well taken. As we become more and more subject to big data we will see more and more of the cost of risk shifted to those perceived as the riskiest.
LikeLiked by 1 person
Insurance wrong business model for a social good like health care.
LikeLiked by 1 person
Even if insurers knew everything that can be known about about our current and past health statuses, there’s still a gamble to be made by betting on future *changes* in health statuses. If an insured person’s overall likelihood of getting sick increases, then he or she gets a payout from that kind of insurance. So the insurance business then becomes a matter of betting on a portfolio of people whose overall statuses decline at different rates, and recruiting new people into the portfolio.
Randomness or stochasticity doesn’t go away that easily, just because we know a little more!
As it happens, this is the basis of the only free market idea for health insurance that seems to make even a little bit of sense to me. I *can* tell you at least one big problem with this proposal: In the deregulated insurance market this envisions, insurance companies would have to be free to trade in human futures in the same way as corn futures or pork bellies. This is risky, and yucky sounding.
But I lack the mathematical competence to give the analysis this idea needs. MathBabe, can you help? (There’s also a basic economics question, which I’m also incompetent about, which is whether the information asymmetry problems in healthcare are unsolvable or not. Information asymmetry in healthcare seems like a huge obstacle to me.)
The idea is “time consistent health insurance,” by University of Chicago free market economist John Cochrane. https://faculty.chicagobooth.edu/john.cochrane/research/index.htm
Here’s a popularization: http://faculty.chicagobooth.edu/john.cochrane/research/papers/cochrane_cato_final.pdf
I’m not trying to sell Cochrane’s idea. But I think that it is behind one aspect of Republicans’ Obamacare replacement proposals: a requirement for continuous enrollment as a replacement for the individual mandate. This needs thorough analysis. Since there’s no experience with it, it needs mathematical analysis.
LikeLike
Yeah I’m thinking universal health care.
LikeLiked by 1 person
I’d vote in favor of socialized medicine. I have a very strong ideological preference in favor of socialized medicine, not just insurance, but with doctors working for the state, as in the UK, Italy, or Sweden.
I’m asking you, as an expert in the mathematics of financial markets, whether Cochrane’s idea makes sense mathematically, not whether it’s a good idea. I don’t think it is a good idea. But I do think that this idea underlies some Republican proposals, so it’s important.
But on the topic of the politics of healthcare, Colorado voters rejected single payer in 2016 by over 75%, just as California voters did in 1994, and Swiss voters in 1996. The average voter lacks my ideological preferences but already has health insurance, which he or she worries about being replaced by something worse.
LikeLiked by 1 person
I may be missing something, but this Reuters article fails to acknowledge that health insurers can only vary price by five variables:
– age
– geographic location
– tobacco use
– individual vs. family plan
– type of plan selected
They can’t vary rates by pre-existing conditions or any variables that hint at one’s health, like obesity or joining Weight Watchers.
The practice being described would be blatantly illegal, I think. Is there any evidence that health insurers are doing this?
http://obamacarefacts.com/factors-affect-health-insurance-costs/
LikeLike
Interesting point in an FT article today: If Big Data is seen by companies and investors as an asset, regulations on its use, like the one in Europe kicking in next year, will have a big bottom-line impact on its value. They do a read-across from oil – if the government were to regulate it more dramatically, what would happen, etc. Not entirely airtight argument, but a conversation starter perhaps. https://ftalphaville.ft.com/2017/03/09/2185767/when-capitalism-wants-to-data-mine-you/
LikeLike