The problem with trolling my FB account for understanding my behavior is that people fall into roughly 3 categories:

- Old work associates

- School friends

- Family and close friends

If I have a deadbeat cousin who also happens to be the black sheep of the family, they may penalize me unfairly for his behavior. Or if I used to be in school with someone who is now doing poorly financially, I get whacked. Now consider that this person might have been my close friend in 6th grade, but we grew in different directions. Where is the relevance here? Where's the logic in that?

Pulling this info from social graphs as augmentation is a whole lot different than simply creating a full profile from your graph in the absence of real financial data.

Lagging or not, how I paid my credit cards over the past 5 years is going to tell you a lot about how I treat money. Not so much about what I'm tweeting tomorrow.

Social grading gives them free reign near-uniformly depress credit scores to make everyone appear less credit-worthy.

This allows them to increase average rates and maintain low lending volume, under the cover of credit scores that appear low in historical context. When it behooves them to begin lending again, their lending to people with low(-in-historical-context) credit scores will be paraded about as virtue.

There is simply no-one who is that far removed from a black sheep or friend/family member who's fallen on hard times. Not even during good economic times. Certainly not given the way people friend one another on these sites.

Any grading of people based on social connections will drag nearly everyone down, potentially in a self-reinforcing cycle.(higher rates push more people into worse economic situations, lowering their credit score, reflecting negatively on their friends who now pay higher rates, etc)

If a bank wanted to raise rates, the bank could just raise rates. They don't do that because, if they raise them excessively, competing banks will undercut them. This same reason prevents them from using Snidely Whiplash Credit Scores ("Everyone gets a 20! BWA HA HA!"): if SWCS predicts defaults worse than FICO does, then competing banks will use FICO and steal the borrowers for whom FICO is more predictive than SWCS, letting EvilBank attempt to raise rates on the dregs of the barrel.

Yes, it's only an issue inasmuch as it's (nearly) uniformly applied. i.e. the risk is mostly of FICO scores being weighted by 'social' factors.

And my optimism that new more-accurate ratings agencies will appear to compete with profitably-wrong ratings agencies went out the window with the CDO mess.

Well exactly.

Let them use whatever the hell they want. Either they'll be right (and have a useful credit rating score) or they'll be wrong, and lose custom to other banks. That's how a market works.

You have no God-given right to credit. You have access to as much credit as someone wants to offer you, and they compete each other such that the rates you get offered reflect the market-thinking of normal profit (as compared to super-normal profits) + default costs.

Of course you have no god-given anything, that's ridiculous. And so is the concept of rights.

But you do have a reasonable expectation as a citizen of a country that your common tax/output funded currency will be available to you on the same terms it is to others. Otherwise, why would you contribute to/use it?

This is already not the case in most places as all lending is done through predatory monopolies with privileged access to the central bank, and a central bank financed via government subsidy but outside of government (the people's) control.

Already we've got a toxic system that's not of much use to the people who pay for it.

part of it is advertising...this way they can advertise 1.9% APR* (with good credit) and give everyone else 9.5%

9.5% is pretty damn good in Europe, I haven't really seen low APRs here. Plus, grace periods are a rarity(more exactly I do not know of any Nordic nor Eastern European banks which would offer them). You use the credit card and you pay the interest from the moment you put something on the card.

I think US has the most competitive credit card enviroment.

I think you'd have to compare the various fees and fine print as well. The US has a great credit card environment for people who don't actually ever need credit.

It becomes considerably less great for people who use those cards as anything other than a lower-risk debit card.

> they may penalize me unfairly for his behavior.

What you are describing is just a fundamental quality of these types of risk assessment algorithms. For example, males pay more for car insurance than females, because there is a demonstratable correlation between being a male and the insurance company paying out more money to you.

This is true even for an individual that has been in absolutely no accidents. Statistically it is true that a male that has had no accidents still has a higher expected value for insurance payouts in the future than a female that has been in no accidents.

You as an individual is an unknown quantity that they cannot absolutely determine. There may have been completely legitimacy reasons why you failed to pay credit cards that cannot possibly occur again in your life, you should equally complain about that not being taken into account.

No one can perfectly predict the future (if they could, insurance would be pointless). They can only use information that they have to imperfectly model the future; both who you tweet and previous credit card payments are imperfect signals for what will happen in your future. It is true that how you paid your credit cards in the past is a much stronger singal; it is absolutely impossible that any credit evaluation is taking who your cousin is and your personal history as equals. Their models take into account that who you tweet is a much lesser indicator.

>You as an individual is an unknown quantity that they cannot absolutely determine.

well, we can imagine that given more info and better algorithms they would narrow the "unknown quantity" into the range much more narrow and thus less populated than "male, 22 years old, 00000 zip code". From a hundreds thousands peers to just a hundred of peers in the same risk level pool - it would be very different "unknown quantity" then.

The better they differentiate the higher profits they will get by offering lower quotes to no-risk ones while more intensely screwing ones with the risk. This is wet dream of the retail, insurance, etc... business - custom targetted offerings. Lower prices when it is really neccessary to make the sale and screw the customer by jacking up the price when it is possible. To do this they need to _know_ the customer. Thanks Facebook.

Excellent points, I could see a big benifit of these types of algorithms for people that don't have established credit scores. If as a bank seeing that a high school graduate knows many of my good customers, this would make me more willing to loan him money.

Social graphs can show these things. For example you deadbeat cousin who is a black sheep in the family probably has a very different circle of friends outside of your family than you or most of your family does. As such he could be identified as an out-lier. In fact I would not consider these algorithms worth while if they didn't do that.

I doubt this would replace a credit score but it may argument it. Or it might lead to a lender making further inquires. For example if you've been tweeting about how you "hope to blow tens of thousands of dollars in Vegas" it seams reasonable that would make a lender worry about lending you.

In the end I don't see these algorithms are very different from a bank in a small town where the manager personally knows everyone in the town.

If the algorithm is good, it will have a leniency for having a black sheep in your family- seemingly everybody does, and numerous demonstrably low-risk individuals I am sure have their black sheep as well.

If the algorithm is good.

"""If I have a deadbeat cousin who also happens to be the black sheep of the family, they may penalize me unfairly for his behavior."""

Yeah, like you're so much better than Matt "Glue Sniffer" Jameson, now that you made it in the big city. Really, is Des Moines as nice as you're making it out to be?

Anyway, when are you coming back to Idaho, Jim? We miss you!

P.S Matt is in jail again, if you can chip in for the bail, send me a message in FB.

I bet the people at working at The Onion do too.

Are you saying you fear they will develop algorithms that are non-discriminatory but produce results that essentially mirror discriminatory results?

"Fear" wouldn't be the word I would use, and I wouldn't say they're mirroring discriminatory results so much as approximating reality with some level of error. Reality is not egalitarian. Good approximations of reality will probably not be egalitarian, either. A nakedly discriminatory heuristic might be a better approximation of reality than "everyone is equal." A non-discriminatory heuristic might be a better approximation than that and also have the side effect of falling very unequally across arbitrary groupings of people.

For example, according to actual data I've seen, poor people really are disproportionately likely to a) have missed payments before, b) not repay your loan, and c) belong to certain ethnic groups. Some people may fit into one of these categories and not the others, but the correlation exists, so you can't clamp down on one without also measurably reducing the others unless you go out of your way to do so.

(Don't know who's downvoting, but I'm just stating pretty well-known facts. Don't take this to indicate approval — I'm just saying that, statistically, some groups do get the short end of the stick. What you do with this information is a different question entirely.)

Here is some actual data comparing ethnic groups. It does indeed conclude that blacks are more likely to exhibit risky financial profiles than whites.

http://www.bos.frb.org/economic/wp/wp1992/wp92_7.pdf

However, the study also concludes that for a sample of black people statistically identical to whites, the approval rate is only 82% (compared to 90% for whites).

Interesting. That study was only shortly after the FICO scoring system was introduced, so I'm curious if anything has changed.

Why is more accurately assessing credit risk without discriminating by race the "ultimate nightmare"?

One example - down payment size for home mortgages. A low loan/value ratio strongly affects risk (a), it's verifiably blind to all protected classes (b) and is also correlated with certain protected classes (c).

Is it the "ultimate nightmare" for banks to use the loan/value ratio while evaluating creditworthiness of mortgage applicants?

Its days like these that explain why so many of us pay close attention to what you have to say. :)