I said no to mortgage backed securities in 2006. Beginning to think about targeted advertising disguised as dopamine driven feedback loops along the same lines.
Sometimes I wonder if people over-estimate the effects of AI/ML. There is a whole industry out there trying to figure out effective lending practices for centuries now. But, it seems people thinks huge amount of data crunched along with AI/ML is enough to solve this problem.
In the context of data collection and prediction, what does it mean to be irrational? Truly random? Do people actually behave in truly random ways? Or are they predictable?
In theory you’d want the algorithm to determine how rational someone is likely to be, assuming that is correlated with debt repayment. If you can say with some degree of confidence that a person is likely to act irrationally and that people who act irrationally don’t pay back debt as often, then that’s a signal to consider when deciding to lend. Not sure exactly how you’d determine that but I guess the hope of collecting such varied data points as phone battery level, typing speed, and food delivery preferences are meant to provide some amount of signal there.
What do rational/irrational mean, quantitatively speaking? Before we even write the algorithm, we've got to come up with a formal definition and I can't even begin to think about how you would do that.
Future behavior isn’t entirely predictable by past behavior. Sure. External events might put a particular individual in arrears, even though she/he has never defaulted in her/his life.
Machine learning however, can use past behavior to make an educated guess, regarding how/when/if an individual will repay and/or how vulnerable said individual is to unforeseen external events.
Machine learning might not be able to go as deep as evaluating someone’s chances of getting hit by lightning, for the purposes of credit ratings. However it does figure out if an individual has the chance to get out of sticky situations.
Summarizing, it’s not about predicting irrational behavior. It’s about using past behavior to make a better educated guess than was possible before.
It would definitely seem like ML could improve performance here, if they fed it all the right features.
I feel like outstanding debts (amount, time since last payment) could be really helpful in this regard so people can't rack up huge debts from multiple lenders, but that would also be a privacy nightmare.
A bank in Belarus scrapped ML model that was used to issue consumer loans after a year. Initially the model gave better returns, but after a year it was underperforming. The bank went back to a weighted average of few factors like it was before with humans reviewing the weights. In retrospect this is not surprising. The model tries to predict long-term behaviour, but besides few stable factors like profession, income, education human life still rather unpredictable even after averaging hundreds of thousands cases.
Non-traditional credit-worthiness data is mostly only possible in China at this stage. Looking at the Fico XD score which enhances the traditional credit bureau score with Lenddo information is quite a game-changer in terms of banking for the non-banked. There will be people like "Mr. Bai" who start to gamble with this system and fail. But these people exist everywhere.
Slightly over a year ago there was an interesting post up on HN where lenders used nude pictures as IOU¹. Wonder if this trend progressed with the companies they write about in this piece.
ironically this was not a problem to small banks in small towns before the conglomerates took over with credit scores.
this is trying to shoehorn the wrong solution to a problem. a local bank would see very easily that the person the article opens with was developing a borrowing addiction. they would also not lend to too many people investing on the same market in the same region, etc. all things that a centralized system can't kown very well, even if they know everything about your past.
In contrast to the other comments here, this isn't a machine learning problem. Also, not a lending problem. The big issue is privacy.
As I was reading the article, everything seemed fine until I read the lending companies were stealing private info. Calling contacts, threatening violence with location tracking.
If lenders want to use battery life and typing speed as as credit indicator, that's up to them. (By the way, the NY Times probably has that wrong. Battery life and typing is likely used for for user tracking/fingerprinting. Imagine a bot borrowing small amounts 10,000 times pretending to be 10,000 different people. Lender's trying to combat something like that.)
Over-borrowing is a user error. Mr. Bai needs to take some responsibility. (I fully realize people blame the financial institutions, including for the credit crisis in US in 2008... we can agree to disagree on that one). Also, there are ways to protect over-borrowers, such as bankruptcy laws and statue of limitations. Bankruptcy laws can allow them to start over. Statue of limitations can put an end to the credit collector hounding.
When it comes to privacy, this article will be an example I'll be mentioning to people when they say "I don't mind the surveillance, I have nothing to hide. I'm not breaking any laws".
Actually, the battery life thing is correct. Shady lenders in other countries also use that metric.
If you're the sort of person who keeps their battery topped off, it's a signal that you're responsible or perhaps have a steady job where you can plug in.
If you're the sort of person who's always scrounging for a plug to juice up a few percent here and there, it indicates a more risky lifestyle.
>One Paipaidai borrower, a man named Lin in a small town in Fujian Province called Quanzhou, said he had racked up about $75,000 in loans from 30 different platforms for living expenses and an investment in a shoe store. Mr. Lin, who asked that his full name not be used for fear of reprisal from debt collectors, said he received multiple calls a day from them.
Quanzhou has a population north of 8m [0]. Could the author have meant to mention a town/region inside Quanzhou?
24 comments
[ 0.21 ms ] story [ 799 ms ] threadSometimes I wonder if people over-estimate the effects of AI/ML. There is a whole industry out there trying to figure out effective lending practices for centuries now. But, it seems people thinks huge amount of data crunched along with AI/ML is enough to solve this problem.
No amount of machine learning or artificial intelligence can change that.
Machine learning however, can use past behavior to make an educated guess, regarding how/when/if an individual will repay and/or how vulnerable said individual is to unforeseen external events.
Machine learning might not be able to go as deep as evaluating someone’s chances of getting hit by lightning, for the purposes of credit ratings. However it does figure out if an individual has the chance to get out of sticky situations.
Summarizing, it’s not about predicting irrational behavior. It’s about using past behavior to make a better educated guess than was possible before.
The goal isn't to solve the problem, as in predict with 100% accuracy. The goal is to improve by a few percentage points.
Since categorization is what ML does well and credit histories are great annotated training sets, it seems like a perfect fit.
I feel like outstanding debts (amount, time since last payment) could be really helpful in this regard so people can't rack up huge debts from multiple lenders, but that would also be a privacy nightmare.
Remember last time you heard China has a high saving rate?
How time has changed.
[1] https://news.ycombinator.com/item?id=13003838
this is trying to shoehorn the wrong solution to a problem. a local bank would see very easily that the person the article opens with was developing a borrowing addiction. they would also not lend to too many people investing on the same market in the same region, etc. all things that a centralized system can't kown very well, even if they know everything about your past.
As I was reading the article, everything seemed fine until I read the lending companies were stealing private info. Calling contacts, threatening violence with location tracking.
If lenders want to use battery life and typing speed as as credit indicator, that's up to them. (By the way, the NY Times probably has that wrong. Battery life and typing is likely used for for user tracking/fingerprinting. Imagine a bot borrowing small amounts 10,000 times pretending to be 10,000 different people. Lender's trying to combat something like that.)
Over-borrowing is a user error. Mr. Bai needs to take some responsibility. (I fully realize people blame the financial institutions, including for the credit crisis in US in 2008... we can agree to disagree on that one). Also, there are ways to protect over-borrowers, such as bankruptcy laws and statue of limitations. Bankruptcy laws can allow them to start over. Statue of limitations can put an end to the credit collector hounding.
When it comes to privacy, this article will be an example I'll be mentioning to people when they say "I don't mind the surveillance, I have nothing to hide. I'm not breaking any laws".
If you're the sort of person who keeps their battery topped off, it's a signal that you're responsible or perhaps have a steady job where you can plug in.
If you're the sort of person who's always scrounging for a plug to juice up a few percent here and there, it indicates a more risky lifestyle.
Quanzhou has a population north of 8m [0]. Could the author have meant to mention a town/region inside Quanzhou?
[0]https://en.wikipedia.org/wiki/Quanzhou