Hacker News Reader

Comment by scarab92 | Hacker News Reader

Credit scores are a good system actually.

They significantly improve the ability of a lender to model the default risk associated with a borrower, which results in lower borrowing costs for higher quality customers, as well as stopping debt spirals for people struggling.

In the absence of credit scores, higher quality borrowers will be charged higher interest rates or require higher collateral, since they are less differentiated from average quality borrowers, and/or access to credit will be restricted to a narrower proportion of the population.

View on HN

Credit scores often oversimplify risk, ignore nuanced factors, and can worsen financial struggles for those with low scores by pushing them toward higher costs. Countries without credit scores don't have higher interest rates or higher rates of defaulting. If we really want to, creditworthiness can be predicted more accurately and fairly with AI models, although that brings another level of bias.

> and can worsen financial struggles for those with low scores by pushing them toward higher costs.

Shouldn’t those costs be higher because of the increased risks they represent? If someone has a habit of not paying their debts, why would a lender take on the higher risk without getting paid more? They wouldn’t, so they simply won’t loan the money.

However on the other side of this, nobody has to borrow money. If you don’t borrow, credit scores are irrelevant.

> nobody has to borrow money. If you don’t borrow, credit scores are irrelevant.

Except if they want their own place to live.

And for the period of about a decade or two, if they want to buy anything over the Internet. For some reason, e-commerce in its early years would only accept credit cards. Took quite a while before debit cards started working for on-line shopping, and by that time, the damage was already done - credit cards got a boost in popularity both in the US and worldwide.

> creditworthiness can be predicted more accurately and fairly with AI models, although that brings another level of bias

I am used to the word "bias" meaning a specific kind of inaccuracy.

> In the absence of credit scores, higher quality borrowers will be charged higher interest rates or require higher collateral, since they are less differentiated from average quality borrowers, and/or access to credit will be restricted to a narrower proportion of the population.

Given we're talking about alternatives here, I think this should be: "In the absence of credit scores, or some other mechanism for comparing potential borrowers". There's lots of ways to compare borrowers, and if you assume credit scores are the only way to do it, all your solutions are going to involve them.

The term 'credit score' is sufficiently general to encompass literally all methods of comparing borrowers. You could certainly take issue with some specifics of how particular agencies calculate it, but the idea that there is some "alternative way of comparing borrowers" then I'd invite you to invent another formula for determining loan parameters, that does not boil down to a scalar value.

Loans involve the calculation of parameters. You can either choose those implicitly through personal knowledge, or explicitly through a scalar metric (credit score). There is no viable third option, and the first option is just a bad version of the second, in the end.

Yep.

It's worth pointing out that credit scores actually are actually just the P(^default) expressed on an integer rather than fractional scale.

There's also multiple credit scores, there are the scores computed by credit bureaus which look at your P(default) against all lenders, but many lenders also compute their own internal credit scores using models trained against their own customer base (and possibly also taking into account additional data that they hold about you).

Wasn't the current system used in the US developed as a hack for a particular situation that then got extended for things it wasn't actually meant to do? I remember reading something about it in the Big Short.

Reply on news.ycombinator.com