Could they? Sure, I guess. Is doing so going to be a more reliable indicator of creditworthiness than existing measures? Not necessarily. Many groups of friends can have wildly varying levels of income and financial responsibility, especially among younger groups of people who all went to high school together then diverged into different careers later.
I've lent people money. Friends, too. Usually that's a tough situation. I live by the maxim that "if you loan money to a friend, be prepared to lose the money or the friend." Still, you want to help your friends out, right?
Thing is, sometimes those friends (or other folks) have friends who enable them. Maybe it's to go out to the bars at night or buy a new game so they can play together. It's hard to predict what their impact will be, at least with the information most of us have to work with.
Now, suppose I did have a considerable amount of information, such as Facebook would. I could make more accurate predictions, but there will always be gaps. I always feel uncomfortable having personal conversations over Facebook because I always feel like someone's looking over my shoulder. Or maybe I don't want the communication remembered. Either way, it's a blank spot that can't be confirmed or denied, so it will always be an imperfect, some might say discriminatory measure.
I'm reminded of the scene in Gattaca when the protagonist is born and instantly a doctor preforms a DNA test to show all the disadvantages he's statistically likely to experience in life: asthma, ADHD, and so on. I'm starting to feel like that's becoming the case today, but with personal history instead of genetics. You can't escape your genome but you can grow up and change yourself from the immature person that made mistakes as an adolescent. Still, the record remains.
This is pretty much the core of privacy that we're losing - the ability to reinvent ones self. At age 18 I left just about everything I knew behind, and I'm not sure it'd be possible to do so now. We have things like bankruptcies that go off of your record in 7 years - I'll bet if given the choice banks would love to see them forever. Full information is not always a good thing - it almost always ignores people's ability to be flexible.
I think that's because our societies have evolved over many centuries with the basic truth that past indiscretions would tend to be forgiven and forgotten in time. For the first time ever, in literally the space of one generation, we have created both mainstream infrastructure that allows unprecedented monitoring of individuals and mainstream infrastructure that never forgets.
It's going to take further generational change to adapt to those new possibilities and decide how -- and even if -- we should use that technology. In the meantime, sadly, varying degrees of bad things are going to happen to a lot of people with varying degrees of innocence through little or no fault of their own.
My greatest concern is that many people in our societies who have the loudest voices in the media and the greatest power through running governments and to some extent big businesses also have very little understanding of the technology and its implications. They have bizarre incentives that promote doing some bad things with it, and little accountability for (or maybe even awareness of) the potential risks they take on our behalf, and worst of all they're convincing ordinary people who don't work with the technology and understand its implications that this is all OK and to be expected.
We could boycott networks & technologies that infringe upon our ability to exist as private individuals and to self-determine: Don't use the wifi toothbrush or the Fitbit, don't use Facebook or Twitter, stay connected with friends using email or (!) the postal service. Use less technology and gain more freedom.
Alternately, market collusion and the unquenchable desire of people to sell their information in exchange for Javascript-enabled friendships will force us into a grim future where the mistakes of the people we love & an assortment of bureaucratically-mandated sensors impede our pursuit of happiness.
Unfortunately I don't expect a boycott will ever work because there will always be statistically significant number of users for Facebook to run their social experiments on, so there won't be any reason for them to stop, even if the populations of entire countries delete their accounts.
I'm still holding out for a privacy-centric social network. Only problem is I won't join it unless enough of my friends do, so I guess we'll see.
>It's going to take further generational change to adapt to those new possibilities and decide how -- and even if -- we should use that technology.
The problem is for most things we're not doing any deciding. We just invent new technology, and then we have to live with its consequences, while never stopping and thinking about it or actually and democratically deciding if we should use it and how we should restrict it.
All that with the sole exception of when there's comercial interests at stake. Then we go out of our way to prevent the technology from messing with them, e.g. adding DRM to help with copyright. But for everything that can mess up the lives of millions of people (be it because of lifestyle changes, environmental damage, invasion of privacy, etc) we either never get to decide and restrict consequences, or we do very little of it.
Car insurance is making its way towards variable pricing on how much and how well you drive. Health insurance will inevitably start to care about what you eat and how much you exercise. Technology enables much more accurate and precise assessment of risk, and businesses with a lot of money riding on good risk assessment (banks, insurance companies) would be stupid not to take full advantage of that. If markets tend towards efficiency, this will happen.
And I'm not sure that's a bad thing. Being young and male, I pay for the sins of my demographic category in car insurance premiums. I'd much rather pay for my actual driving style. Similarly, credit risk can probably be assessed by actual spending behavior rather than history.
If you find it uncomfortable that lending institutions and insurance companies might act in the most rational, efficient manner, consider whether you're really comfortable with capitalism.
> I pay for the sins of my demographic category in car insurance premiums
Does your insurer offer lower premiums if you get a "black box" attached to your car and can demonstrate you drive safely? Asking because I'm curious -- car insurance rates for young men are very high and this was mentioned as a possible mitigation.
Yup. It's not even young men who are screwed over. I'm a 31 year old man with a flawless driving record(zero accidents, zero tickets), having driven over 400k miles over the past 14 years, and yet I still pay more in insurance than my 19 year old female cousin who already has rear ended a person in the 2 years she has been driving.
Allstate and State Farm do too. I partipate in the Allstate one currently. I can cut my bill by 30%. However, it is next to impossible to get the max. You get 10% just for enrolling and I'm on track for another 10%. The interesting part is this is all based on off the app on my phone. I do not connect anything to my OBD.
> Technology enables much more accurate and precise assessment of risk, and businesses with a lot of money riding on good risk assessment (banks, insurance companies) would be stupid not to take full advantage of that. If markets tend towards efficiency, this will happen.
The problem is that you can't eliminate risk[1][2], you can only shift it to someone else. But, we now live in a time in which people with the most powerful computers can shift all of the risk away from them. Consider where that risk ends up getting shifted.
I argue that they can eliminate risk. What happens is the top 10% of the people who cost insurance get the boot.
Now, unless they choose just to not pay out, they can still reduce their risk levels below threshold of the industry. That's what actuarial science is founded on.
Oh, you mean "that branch of science that everyone ignores because it's politically unpalatable?"
"Gee, we promised the moon for our pension obligations and funded them at an absurdly low level, and now we're bankrupt? I don't know how this could have happened!" ~ Municipal governments
For some types of insurance, "top 10% get the boot" may not be possible, as there might be laws requiring the insurer to write a policy for anyone willing to pay the premium.
You can diversify risk which is almost the same thing. The whole point of insurance is that each person has x% probability of dying, so they pay x% * the sum assured, and, as the pool of insured gets large enough, there is enough cheese in the fridge.
Yes, the pool is not always large enough, so you can throw it into a reinsurer from the insurance company. Yes, the reinsurance pool isn't always large enough, so you can retrocede to a few other reinsurers for diversification.
That pool isn't aways large enough, so you can securitise away the tail risks and aggregate loss risks into a cat bond, or a pandemic bond, or a longevity swap, or something fun.
These will then in turn be bought up by pension companies, in order to diversify against their market risks, and the needle returns to the start of the song, and we all sing along like before.
The price is x% of the sum assured discounted at the risk neutral rates, plus admin expenses, plus customer acquisition cost, plus a target percentage of the economic capital (because their is a wee bit of risk), plus adjustments based on your demographic's price sensitivity profile, combined with adjustments based on the market cycle. Of course, you should also factor lapse rates in. Then maybe factor in whether this is a good beginner product for new customers.
How is this relevant to the article at all? As a credit risk modeller, we already use spending behaviour to assess credit risk; and I am pretty sure the insurance industry does too. No retail portfolio can be tailored to the individual level (or the methods are not that sophisticated yet); hence the average behaviour of your demographic will matter whether you like it or not.
To your specific point, in my country we have the concept of a "no claim bonus" where the insurance premium is drastically lower if there is no claim made in a year - not sure whether there is a similar practice in the US.
My understanding is you have access to total monthly credit usage vs. credit availability, not what I'm spending money on. I would think there's information to be gleaned from whether the distribution of spending across categories looks like a deliberate household budget, whether purchases include luxury items, etc. that would let you spot people likely to bury themselves in credit card debt before they actually do so. Just as people's social networks might tell you about their attitudes towards money when influenced by friends.
I have not worked for the US market the last five years, but as far as I remember there is a law in the US that a bank cannot take adverse action on a existing customer holding the bank's product based on credit history from another bank - so for example, my bank cannot decrease the limit of a card because the customer defaulted in a personal loan held with another bank. Of course, at the time of origination of the credit card, we can use this information.
The whole point being that just because data is available does not mean I can use it even though it may be statistically significant for me to use it.
Again, I am not sure of insurance regulations so there may be a possibility around using more data, if available, in insurance models.
Pretty much all countries have the no claim bonus for driving. (It's in the earliest actuarial exams.)
You can clock up several years of no claims bonuses, and the discount increases with each year.
Interestingly, the no claims bonus can be insured. Now, generally only risk averse people will insure their no claims bonus - and these people also tend to drive more risk aversely and hence have less accidents (or perhaps less likely to make claims - who the hell knows).
So, the very fact that somebody wanted to insure their no claims bonus should in theory give them a discount on their premiums. But you cannot offer a discount for an extra feature! This means you have to dive into your generalized linear model and fiddle with a couple of parameters to make sure there is some sort of nominal charge for the insurance on the no claims bonus.
"Your demographic" is just a feature of your model though. Just because your methods are not very sophisticated, doesn't mean that others can't be upset that you arbitrarily group them by whatever rules you decided to come up with.
>> consider whether you're really comfortable with capitalism
My favourite thing about capitalism is that you actually only have to consider whether or not you're comfortable with this policy. If you don't like it, by no means do you HAVE to take a loan. You DO have to buy insurance in many situations, but at that point it's no longer a true free market anyway.
Insurance is a large-scale perverter of the market, because it muddies the direct relationship between the cost and the value of the service and makes the pricing all screwy. Creating this opening for pencil-pushing middlemen to profit off a service to which they provide no value is a big problem.
This is blatantly apparent in the health industry, where if you're a cash patient that knows the right magic incantations, you can get 90% of the bill waived, and if you don't know those incantations, you're forced into bankruptcy with hundreds of thousands in medical debt. The American health system is despicable and must be rectified.
The specific incantations are different for every provider or hospital. Sometimes you can get it just by saying up front "I'm paying cash today." Sometimes you have to go through a complicated Patient Financial Services system to get the bill reduced, and then you have to say the right words to the right people if you want reduction.
A massive bulk of "medical" labor goes into these cat and mouse games between patients, insurers, providers, and facilities. They're all trying to figure out how to maximize their profit and/or savings and it gets very complicated.
If insurers learn that other people, even cash patients, are getting "a better deal", they'll punish the provider and/or facility by removing them from the network, thus cutting the patient traffic. When providers and facilities bill insurers, they expect to get maybe 10%-20% of the billed amount, so they rack the cost of everything up through the roof so that they'll be OK with the pittance they get in return. And that's the tip of the iceberg. There are a lot of people whose jobs are dedicated to nothing more than these billing shenanigans.
In response to the now dead comment about how government-mandated insurance perverts the market, I agree that's a bigger departure from the idealistic free market model. However cookiecaper is referring to moral hazard: where people will make reckless decisions because insurance is essentially paying someone else to absorb uncertainty. When the risk is assumed by another, you take bigger risks. So there's increased uncertainty. And it has negative externalities, so yeah: another warp in the ideal free market model.
But that's all the more reason why insurance companies want (and IMO should have) better insight into who you are: so the risk is more accurately quantifiable.
The next logical step is for credit agencies to bake this into your credit score.
While this particular application mentions loans, credit scores are used for a lot more things than loans. It would be interesting to live in a world where you can't get a roof over your head because your deadbeat brother has a terrible credit score.
>Health insurance will inevitably start to care about what you eat and how much you exercise.
At a conference in 2007, I learned that health care companies were already exploring data they could get from those free get-a-discount grocery store cards. Unfortunately that's about as much as I know about it. I wish I would have asked the lecturer for more information at the time.
> Health insurance will inevitably start to care about what you eat and how much you exercise.
This is already happening, I met the founder of http://wesavvy.com/ recently which uses your phone to track exercise and reduce your insurance premium.
My understanding of the idea, which may be rubbish, runs as follows.
1. Find a way to identify good risks - in this case regular teeth brushers, or occupants of a crack-house that share a toothbrush.
2. Hence, offer lower premiums to low-risk customers than those companies that charge an average premium to average-risk customers.
3. Get a few customers together and make sure the shareholders of the polyopoly dental insurance mammoths get wind that you are thieving their customers. Panic sets in among the incumbents. Innovation!! Drat!!
4. Get bought out for a Unicorn bill before you even pay a dental claim or cover your costs.
I think there's a big difference between these two things.
When you use a phone to track exercise and reduce insurance premiums, that's a good thing, it encourages (or at least rewards) people who exercise and become healthier and might need less medical care.
When you use social networks to track and monitor who your friends are in order to approve you for credit, that causes a different set of rewards and incentives. It rewards you for choosing who you are willing to be friends with based on their creditworthiness. It punishes people who have associations with people with poor credit.
It essentially keeps good people down too. If you are a role model in your group of peers. If you're the guy out of your group of friends who got an education and got a steady job. If you pay your bills and don't carry debt, and you want to buy a house and have the money and responsibility to do so, you should be allowed to. When you do that, it shows those friends what you can do, and it allows you to move forward.
On the other hand, if you get an education, get a job, pay your bills, and don't carry debt, but when you go to buy a house your mortgage application is denied simply because you have some friends who are struggling on facebook, what is the impact of that? It just causes you to stop moving up. Maybe it discourages you. Maybe it makes you realize that the world doesn't really want you to be better than what your friends have become.
What the problem is has less to do with being judged by your actions, but rather being judged by the actions of those you associate with.
If you find it uncomfortable that lending institutions and insurance companies might act in the most rational, efficient manner, consider whether you're really comfortable with capitalism.
The trouble is that the situation you describe is idealised and probably unachievable. In the real world, it seems more likely that a different, more complicated set of strict automated rules will be created, and it will simply create different loopholes through which some innocent people fall and get treated unfairly as a result. There will just be a lot more loopholes with fewer people falling through each one, making the system as a whole less accountable.
I learned a useful lesson just after I graduated from university. I had a decent job offer, I wanted to consolidate my modest amount of student debt into a sensible place as the student banking perks were expiring, and I wanted a small loan to buy a car to commute to that new job (which was far cheaper than either moving or taking public transport). I went to the bank I'd been with for years as a student, with all the paperwork including the signed contract for the new job, and explained what I was looking for. The guy I was speaking to looked a little sheepish, said that he thought what I was asking to do seemed perfectly reasonable... and then off the record suggested I didn't formally apply for it, because he already knew the computer would say no. My new job didn't start until the following week, and apparently that would mean the computer would have considered me to have no income at that time and so all the application would have done was get an automatic refusal and a mark as such on my credit record.
I think the moral of the story is that more data is only as useful as the analysis you make of it, and trying to codify and automate matters as complex as a typical adult's financial affairs is unlikely to lead to a reliably good analysis. In fact, I think it's even the law in my country these days that any such automated decision has to be reviewed by a real human on request, which might or might not be worth much in practice but does give some idea of the scale of the problem.
I went through the same thing! After 18 months on my first real job, I had zero debt (ever), no dependents, and income of three times the median for the area, and yet I was still classified as subprime.
Why? Because I'd never gotten a credit card before. Had I simply moved existing purchases onto it, and taken the same (zero) revolving debt, I would have appeared as much less risky. [1]
What's worse is that when I went to the "what's my FICO"/credit report sites, they would consistently say that my credit score should be 740, even though this never translated into any observable consequences in terms of ease of getting loans.
And what made that even worse was that there were actual critics of the subprime practices then that refused to believe this was happening, that anyone classified as subprime must have deserved it in some way, even after I offered to forward this information over!
[1] Yes, I know, I should have known to pointlessly get a credit card and switch which buckets expenses go in, which is somehow informative about my credit risk.
Anti-discrimination laws don't cover everything. For instance, it is generally OK to put out a casting call for a blonde male actor, and a company can refuse to hire someone in a wheelchair for a warehouse job.
Those are both examples of employment. It's true that there are some exceptions to employment anti-discrimination laws, but they don't apply at all to how much you charge someone for health insurance.
(And not to belabor your example, but the person in the wheelchair has to actually be unable to do the job even with "reasonable accommodations" from the employer. It's not enough to say "it seems like it might be dangerous" or "we don't have ramps")
I'm pretty sure men and women are now required to get the same rates for health insurance under ACA. How are they allowed to charge people differently based on age? Well, because there's no law against it. Age discrimination is illegal in employment and housing, but not in what you charge for insurance.
Prices for health plans on an ACA exchange can only vary based on age, family size, smoker status, and a geographical "community rating" that's designed to account for local cost of care. No other factors may affect pricing, and anyone who is willing to pay the price must be enrolled.
"If you find it uncomfortable that lending institutions and insurance companies might act in the most rational, efficient manner, consider whether you're really comfortable with capitalism."
Basing credit scores off of other people's credit scores is about as rational and efficient as basing my temperature and blood pressure readings on the average of my friends' readings.
You surely can find other cases in which you end up paying more than the average. In the end, what matters is information asymmetry in any negotiation. In those terms, the consumer seems to be in an increasingly weaker position with the recent technological changes, social media, etc.
As a consumer I don't think that allowing market micro-segmentation is a sensible strategy. It's more something that I don't know very well how to stop, and it worries me.
> 'However, it seems pretty unlikely to me that a scoring system rooted in the community you’re a member of is going to be helpful.'
It seems incredibly likely to me that one's social set are going to influence one's actions, e.g. the type of food one eats, the fashions one wears—or one's likelihood of defaulting on a loan.
What would be surprising would be were it otherwise.
Would it not end up penalizing the exceptions? I have a friend who grew up in a ghetto area. Most of the kids in his neighborhood didn't do much AFAIK. If it was assumed he would behave like the average person of his neighborhood he might not have made it out.
Yup, but so too do credit scores in general. If Harry was unlucky, lost his job and missed a few credit-card payments, then his credit card will suffer regardless of whether he is now actually likely to default; the trouble is that he is the type of person who will default.
It's unfortunate, but that's what happens whenever one tries to categorize people. And it's unavoidable, even with extensive work in-person, as anyone who's ever been deeply disappointed or pleasantly surprised by a family member or close friend can tell. If Pete couldn't tell that his brother would steal his car, can he really be certain that a stranger won't default on his loan? All any of us can do is play the probabilities.
>> Yup, but so too do credit scores in general. If Harry was unlucky, lost his job and missed a few credit-card payments, then his credit card will suffer regardless of whether he is now actually likely to default
While it is true but that's completely under Harry's control. Consumer unsecured credit is a volume game so it is not going to be possible to underwrite based on personal history. So you need to look at the historical behavior of similar people and the data will show he has a higher risk of default.
That said, Harry can definitely improve this situation by getting a secured credit card from a credit union and build his credit back up. However, if he gets dinged for his friend, he has no way out of it without unfriending everyone from his prior life on Facebook.
Lending can be more personal but the economics of unsecured lending makes this challenging. Credit unions get closest to this but the smaller ones who can provide such a service are struggling.
So would Facebook be asking people to share credit scores constantly? How do they plan to get said credit scores? As far as I know, no one can legally access my credit report without my permission. For people with multiple accounts, which Facebook account do they use? What if I don't have an account? Will I just be denied all loans?
If they do manage to get the data, what makes them think this is a valid way to rate people's creditworthiness? It seems like an extremely random metric that measures nothing but is convenient for Facebook because of the type of data they collect. I don't see the moral angle because this is far-fetched speculation not rooted in reality. There are other companies who have tried to replace the current credit scoring with systems based on social data and they haven't gotten anywhere. Perhaps that might be because the best indicator of financial performance is past financial performance and nothing else?
I met a guy at a business networking event last year who worked for a company that generated loan scores from the text in your Facebook posts. He claimed they had a higher score quality than the old style credit scoring agencies. Note this is just the text you have typed in over time, not your set of friends.
I understand FCRA and ECOA very well and it is very unlikely we are going to get a different interpretation of these regs just because Facebook wants to get into the loan origination game. I see a number of issues:
1. Reason Codes: FCRA reason codes that the lender has to generate will be a PR nightmare if they decline the loan application based on Facebook data -- imagine a consumer getting a rejection reason as 'Your fiends have poor credit rating so we don't want to give you credit'.
2. Discrimination / Disparate impact: Most people who are poor also have poor friends and this type of analysis will lead to disparate impact that the article talks about and there is no way around it -- even if they get a slightly different interpretation of FCRA they are not going to solve disparity easily.
3. CRA Regulation around data quality: Credit Reporting Agencies (CRA) has specific mandates around providing accurate information about the the borrower. If a borrower thinks it is wrong, they can request the CRA to fix it through a dispute process -- this is a very time-intensive, expensive process for a CRA. I doubt Facebook would want to build a process for disputes and the disputes are going to be worse -- before I will say, I paid my credit card on time and you got it wrong to the credit bureau. Now I may say, I don't think my friend's rating is low and Facebook has to due diligence before they can claim it is correct.
I can see a few ways they might go about this:
1. Second shot at approval as opposed to decline: A lender may reject an application and may choose to use Facebook data to give the borrower a second chance. This is better for the borrower and the regulators may be ok with this type of approach. Although they need to be careful with banks not suddenly changing their FICO Score cut off to something very high so they can fall back on this.
2. Only make it available internationally: not all countries have such strong protections around consumer credit data. Also, many countries don't have a robust credit bureau and this may actually help consumers.
3. Focus on alternative payday lending: CFPB is keen on opening up newer sources of data for certain segments that do not get any credit today. They may be open to this approach if fb can show that it will allow lenders to provide credit to people who they wouldn't have before.
I have strong views about FCRA / ECOA and I feel most of the provisions in there are actually beneficial to the consumers. Even though my startup SimplyCredit is in the fintech space and would greatly benefit from relaxed regulation, I do think it is good to have strong consumer protection and that's why we don't use anything outside of what will be considered as 'payment behavior / financial management'.
I cringe every time a new fintech company claims they monitor how fast you can type or analyze your github profile to give you credit -- either they say these things for marketing but really using it for underwriting or actually using it and breaking many provisions of FCRA. The last thing I want to do is decline a person with disability for typing slowly -- especially when the focus is to really help consumers and not to showoff your power to analyze such realtime data!
Instead of data sources like Facebook, I would welcome a cleaner source of data with verified income and employment.
76 comments
[ 0.26 ms ] story [ 143 ms ] threadThing is, sometimes those friends (or other folks) have friends who enable them. Maybe it's to go out to the bars at night or buy a new game so they can play together. It's hard to predict what their impact will be, at least with the information most of us have to work with.
Now, suppose I did have a considerable amount of information, such as Facebook would. I could make more accurate predictions, but there will always be gaps. I always feel uncomfortable having personal conversations over Facebook because I always feel like someone's looking over my shoulder. Or maybe I don't want the communication remembered. Either way, it's a blank spot that can't be confirmed or denied, so it will always be an imperfect, some might say discriminatory measure.
I'm reminded of the scene in Gattaca when the protagonist is born and instantly a doctor preforms a DNA test to show all the disadvantages he's statistically likely to experience in life: asthma, ADHD, and so on. I'm starting to feel like that's becoming the case today, but with personal history instead of genetics. You can't escape your genome but you can grow up and change yourself from the immature person that made mistakes as an adolescent. Still, the record remains.
Something doesn't feel right about all of this.
I think that's because our societies have evolved over many centuries with the basic truth that past indiscretions would tend to be forgiven and forgotten in time. For the first time ever, in literally the space of one generation, we have created both mainstream infrastructure that allows unprecedented monitoring of individuals and mainstream infrastructure that never forgets.
It's going to take further generational change to adapt to those new possibilities and decide how -- and even if -- we should use that technology. In the meantime, sadly, varying degrees of bad things are going to happen to a lot of people with varying degrees of innocence through little or no fault of their own.
My greatest concern is that many people in our societies who have the loudest voices in the media and the greatest power through running governments and to some extent big businesses also have very little understanding of the technology and its implications. They have bizarre incentives that promote doing some bad things with it, and little accountability for (or maybe even awareness of) the potential risks they take on our behalf, and worst of all they're convincing ordinary people who don't work with the technology and understand its implications that this is all OK and to be expected.
Alternately, market collusion and the unquenchable desire of people to sell their information in exchange for Javascript-enabled friendships will force us into a grim future where the mistakes of the people we love & an assortment of bureaucratically-mandated sensors impede our pursuit of happiness.
Perhaps it's time to draw our lines in the sand.
I'm still holding out for a privacy-centric social network. Only problem is I won't join it unless enough of my friends do, so I guess we'll see.
I hardly use Facebook anymore anyway.
The problem is for most things we're not doing any deciding. We just invent new technology, and then we have to live with its consequences, while never stopping and thinking about it or actually and democratically deciding if we should use it and how we should restrict it.
All that with the sole exception of when there's comercial interests at stake. Then we go out of our way to prevent the technology from messing with them, e.g. adding DRM to help with copyright. But for everything that can mess up the lives of millions of people (be it because of lifestyle changes, environmental damage, invasion of privacy, etc) we either never get to decide and restrict consequences, or we do very little of it.
And I'm not sure that's a bad thing. Being young and male, I pay for the sins of my demographic category in car insurance premiums. I'd much rather pay for my actual driving style. Similarly, credit risk can probably be assessed by actual spending behavior rather than history.
If you find it uncomfortable that lending institutions and insurance companies might act in the most rational, efficient manner, consider whether you're really comfortable with capitalism.
Does your insurer offer lower premiums if you get a "black box" attached to your car and can demonstrate you drive safely? Asking because I'm curious -- car insurance rates for young men are very high and this was mentioned as a possible mitigation.
Statistically, you're extra experience (and safe driving record) makes you a much, much safer driver and insurance companies know this.
The problem is that you can't eliminate risk[1][2], you can only shift it to someone else. But, we now live in a time in which people with the most powerful computers can shift all of the risk away from them. Consider where that risk ends up getting shifted.
[1] http://c2.com/cgi/wiki?YouCantEliminateRisk
[2] Well, science probably does eliminate risk and uncertainty, but that's a totally different enterprise.
Now, unless they choose just to not pay out, they can still reduce their risk levels below threshold of the industry. That's what actuarial science is founded on.
Oh, you mean "that branch of science that everyone ignores because it's politically unpalatable?"
"Gee, we promised the moon for our pension obligations and funded them at an absurdly low level, and now we're bankrupt? I don't know how this could have happened!" ~ Municipal governments
Yes, the pool is not always large enough, so you can throw it into a reinsurer from the insurance company. Yes, the reinsurance pool isn't always large enough, so you can retrocede to a few other reinsurers for diversification.
That pool isn't aways large enough, so you can securitise away the tail risks and aggregate loss risks into a cat bond, or a pandemic bond, or a longevity swap, or something fun.
These will then in turn be bought up by pension companies, in order to diversify against their market risks, and the needle returns to the start of the song, and we all sing along like before.
To your specific point, in my country we have the concept of a "no claim bonus" where the insurance premium is drastically lower if there is no claim made in a year - not sure whether there is a similar practice in the US.
The whole point being that just because data is available does not mean I can use it even though it may be statistically significant for me to use it.
Again, I am not sure of insurance regulations so there may be a possibility around using more data, if available, in insurance models.
You can clock up several years of no claims bonuses, and the discount increases with each year.
Interestingly, the no claims bonus can be insured. Now, generally only risk averse people will insure their no claims bonus - and these people also tend to drive more risk aversely and hence have less accidents (or perhaps less likely to make claims - who the hell knows).
So, the very fact that somebody wanted to insure their no claims bonus should in theory give them a discount on their premiums. But you cannot offer a discount for an extra feature! This means you have to dive into your generalized linear model and fiddle with a couple of parameters to make sure there is some sort of nominal charge for the insurance on the no claims bonus.
My favourite thing about capitalism is that you actually only have to consider whether or not you're comfortable with this policy. If you don't like it, by no means do you HAVE to take a loan. You DO have to buy insurance in many situations, but at that point it's no longer a true free market anyway.
This is blatantly apparent in the health industry, where if you're a cash patient that knows the right magic incantations, you can get 90% of the bill waived, and if you don't know those incantations, you're forced into bankruptcy with hundreds of thousands in medical debt. The American health system is despicable and must be rectified.
The specific incantations are different for every provider or hospital. Sometimes you can get it just by saying up front "I'm paying cash today." Sometimes you have to go through a complicated Patient Financial Services system to get the bill reduced, and then you have to say the right words to the right people if you want reduction.
A massive bulk of "medical" labor goes into these cat and mouse games between patients, insurers, providers, and facilities. They're all trying to figure out how to maximize their profit and/or savings and it gets very complicated.
If insurers learn that other people, even cash patients, are getting "a better deal", they'll punish the provider and/or facility by removing them from the network, thus cutting the patient traffic. When providers and facilities bill insurers, they expect to get maybe 10%-20% of the billed amount, so they rack the cost of everything up through the roof so that they'll be OK with the pittance they get in return. And that's the tip of the iceberg. There are a lot of people whose jobs are dedicated to nothing more than these billing shenanigans.
Government-mandated insurance is a large-scale perverter of the "market".
FTFY
But that's all the more reason why insurance companies want (and IMO should have) better insight into who you are: so the risk is more accurately quantifiable.
While this particular application mentions loans, credit scores are used for a lot more things than loans. It would be interesting to live in a world where you can't get a roof over your head because your deadbeat brother has a terrible credit score.
At a conference in 2007, I learned that health care companies were already exploring data they could get from those free get-a-discount grocery store cards. Unfortunately that's about as much as I know about it. I wish I would have asked the lecturer for more information at the time.
This is already happening, I met the founder of http://wesavvy.com/ recently which uses your phone to track exercise and reduce your insurance premium.
My understanding of the idea, which may be rubbish, runs as follows.
1. Find a way to identify good risks - in this case regular teeth brushers, or occupants of a crack-house that share a toothbrush.
2. Hence, offer lower premiums to low-risk customers than those companies that charge an average premium to average-risk customers.
3. Get a few customers together and make sure the shareholders of the polyopoly dental insurance mammoths get wind that you are thieving their customers. Panic sets in among the incumbents. Innovation!! Drat!!
4. Get bought out for a Unicorn bill before you even pay a dental claim or cover your costs.
When you use a phone to track exercise and reduce insurance premiums, that's a good thing, it encourages (or at least rewards) people who exercise and become healthier and might need less medical care.
When you use social networks to track and monitor who your friends are in order to approve you for credit, that causes a different set of rewards and incentives. It rewards you for choosing who you are willing to be friends with based on their creditworthiness. It punishes people who have associations with people with poor credit.
It essentially keeps good people down too. If you are a role model in your group of peers. If you're the guy out of your group of friends who got an education and got a steady job. If you pay your bills and don't carry debt, and you want to buy a house and have the money and responsibility to do so, you should be allowed to. When you do that, it shows those friends what you can do, and it allows you to move forward.
On the other hand, if you get an education, get a job, pay your bills, and don't carry debt, but when you go to buy a house your mortgage application is denied simply because you have some friends who are struggling on facebook, what is the impact of that? It just causes you to stop moving up. Maybe it discourages you. Maybe it makes you realize that the world doesn't really want you to be better than what your friends have become.
What the problem is has less to do with being judged by your actions, but rather being judged by the actions of those you associate with.
The trouble is that the situation you describe is idealised and probably unachievable. In the real world, it seems more likely that a different, more complicated set of strict automated rules will be created, and it will simply create different loopholes through which some innocent people fall and get treated unfairly as a result. There will just be a lot more loopholes with fewer people falling through each one, making the system as a whole less accountable.
I learned a useful lesson just after I graduated from university. I had a decent job offer, I wanted to consolidate my modest amount of student debt into a sensible place as the student banking perks were expiring, and I wanted a small loan to buy a car to commute to that new job (which was far cheaper than either moving or taking public transport). I went to the bank I'd been with for years as a student, with all the paperwork including the signed contract for the new job, and explained what I was looking for. The guy I was speaking to looked a little sheepish, said that he thought what I was asking to do seemed perfectly reasonable... and then off the record suggested I didn't formally apply for it, because he already knew the computer would say no. My new job didn't start until the following week, and apparently that would mean the computer would have considered me to have no income at that time and so all the application would have done was get an automatic refusal and a mark as such on my credit record.
I think the moral of the story is that more data is only as useful as the analysis you make of it, and trying to codify and automate matters as complex as a typical adult's financial affairs is unlikely to lead to a reliably good analysis. In fact, I think it's even the law in my country these days that any such automated decision has to be reviewed by a real human on request, which might or might not be worth much in practice but does give some idea of the scale of the problem.
Why? Because I'd never gotten a credit card before. Had I simply moved existing purchases onto it, and taken the same (zero) revolving debt, I would have appeared as much less risky. [1]
What's worse is that when I went to the "what's my FICO"/credit report sites, they would consistently say that my credit score should be 740, even though this never translated into any observable consequences in terms of ease of getting loans.
And what made that even worse was that there were actual critics of the subprime practices then that refused to believe this was happening, that anyone classified as subprime must have deserved it in some way, even after I offered to forward this information over!
[1] Yes, I know, I should have known to pointlessly get a credit card and switch which buckets expenses go in, which is somehow informative about my credit risk.
Doesn't the ACA prohibit this. I thought they were now only allowed to base price on Age, Sex, and smoking status.
(And not to belabor your example, but the person in the wheelchair has to actually be unable to do the job even with "reasonable accommodations" from the employer. It's not enough to say "it seems like it might be dangerous" or "we don't have ramps")
Basing credit scores off of other people's credit scores is about as rational and efficient as basing my temperature and blood pressure readings on the average of my friends' readings.
You surely can find other cases in which you end up paying more than the average. In the end, what matters is information asymmetry in any negotiation. In those terms, the consumer seems to be in an increasingly weaker position with the recent technological changes, social media, etc.
As a consumer I don't think that allowing market micro-segmentation is a sensible strategy. It's more something that I don't know very well how to stop, and it worries me.
It seems incredibly likely to me that one's social set are going to influence one's actions, e.g. the type of food one eats, the fashions one wears—or one's likelihood of defaulting on a loan.
What would be surprising would be were it otherwise.
Yup, but so too do credit scores in general. If Harry was unlucky, lost his job and missed a few credit-card payments, then his credit card will suffer regardless of whether he is now actually likely to default; the trouble is that he is the type of person who will default.
It's unfortunate, but that's what happens whenever one tries to categorize people. And it's unavoidable, even with extensive work in-person, as anyone who's ever been deeply disappointed or pleasantly surprised by a family member or close friend can tell. If Pete couldn't tell that his brother would steal his car, can he really be certain that a stranger won't default on his loan? All any of us can do is play the probabilities.
While it is true but that's completely under Harry's control. Consumer unsecured credit is a volume game so it is not going to be possible to underwrite based on personal history. So you need to look at the historical behavior of similar people and the data will show he has a higher risk of default.
That said, Harry can definitely improve this situation by getting a secured credit card from a credit union and build his credit back up. However, if he gets dinged for his friend, he has no way out of it without unfriending everyone from his prior life on Facebook.
Lending can be more personal but the economics of unsecured lending makes this challenging. Credit unions get closest to this but the smaller ones who can provide such a service are struggling.
If they do manage to get the data, what makes them think this is a valid way to rate people's creditworthiness? It seems like an extremely random metric that measures nothing but is convenient for Facebook because of the type of data they collect. I don't see the moral angle because this is far-fetched speculation not rooted in reality. There are other companies who have tried to replace the current credit scoring with systems based on social data and they haven't gotten anywhere. Perhaps that might be because the best indicator of financial performance is past financial performance and nothing else?
1. Reason Codes: FCRA reason codes that the lender has to generate will be a PR nightmare if they decline the loan application based on Facebook data -- imagine a consumer getting a rejection reason as 'Your fiends have poor credit rating so we don't want to give you credit'.
2. Discrimination / Disparate impact: Most people who are poor also have poor friends and this type of analysis will lead to disparate impact that the article talks about and there is no way around it -- even if they get a slightly different interpretation of FCRA they are not going to solve disparity easily.
3. CRA Regulation around data quality: Credit Reporting Agencies (CRA) has specific mandates around providing accurate information about the the borrower. If a borrower thinks it is wrong, they can request the CRA to fix it through a dispute process -- this is a very time-intensive, expensive process for a CRA. I doubt Facebook would want to build a process for disputes and the disputes are going to be worse -- before I will say, I paid my credit card on time and you got it wrong to the credit bureau. Now I may say, I don't think my friend's rating is low and Facebook has to due diligence before they can claim it is correct.
I can see a few ways they might go about this:
1. Second shot at approval as opposed to decline: A lender may reject an application and may choose to use Facebook data to give the borrower a second chance. This is better for the borrower and the regulators may be ok with this type of approach. Although they need to be careful with banks not suddenly changing their FICO Score cut off to something very high so they can fall back on this.
2. Only make it available internationally: not all countries have such strong protections around consumer credit data. Also, many countries don't have a robust credit bureau and this may actually help consumers.
3. Focus on alternative payday lending: CFPB is keen on opening up newer sources of data for certain segments that do not get any credit today. They may be open to this approach if fb can show that it will allow lenders to provide credit to people who they wouldn't have before.
I have strong views about FCRA / ECOA and I feel most of the provisions in there are actually beneficial to the consumers. Even though my startup SimplyCredit is in the fintech space and would greatly benefit from relaxed regulation, I do think it is good to have strong consumer protection and that's why we don't use anything outside of what will be considered as 'payment behavior / financial management'.
I cringe every time a new fintech company claims they monitor how fast you can type or analyze your github profile to give you credit -- either they say these things for marketing but really using it for underwriting or actually using it and breaking many provisions of FCRA. The last thing I want to do is decline a person with disability for typing slowly -- especially when the focus is to really help consumers and not to showoff your power to analyze such realtime data!
Instead of data sources like Facebook, I would welcome a cleaner source of data with verified income and employment.
I don't like it, but i'm a responsible middle aged white male so I probably won't suffer.