3 ms·
> The incentives of the credit reporting agencies are fundamentally unbalanced because the average consumer has no meaningful input into their behavior. Howso?
by judge2020 4y ago
> The incentives of the credit reporting agencies are fundamentally unbalanced because the average consumer has no meaningful input into their behavior.
Howso? The lenders are obligated to submit truthful data, and under FCRA you can dispute fraud and incorrect data. And the FICO/Vantage scoring system can be computed from the data, they don't just spit out a hidden number generated by some ML model.
- eternalban 4y ago> [In "God" we trust] Here you go: https://en.wikipedia.org/wiki/Credit_rating_agencies_and_the_subprime_crisis https://en.wikipedia.org/wiki/Credit_rating_agencies_and_the...
- gruez 4y ago"Credit rating agencies" =/= "Credit bureau". If you read your own link, you'd discover that the former refers to companies like "Moody's Investors Service, Standard & Poor's, and Fitch Ratings". Needless to say, those are separate entities from credit bureaus, which are companies like equifax, transunion, and experian.
- eternalban 4y agoYour reply is a good example of missing the forest for the trees.
- gruez 4y agoThe average HN reader isn't a telepath. If you posted a link about a different subject without any accompanying explanation about why it's related, and other readers get confused, that's on you. Your follow up reply makes the same mistake, but it's worse because it's obvious that at least one person isn't seeing the connection, yet you don't make an effort to provide such an explanation.
- gruez 4y ago>And the FICO/Vantage scoring system can be computed from the data, they don't just spit out a hidden number generated by some ML model. AFAIK those models aren't public. If you search around you'd find some vague factors and aproximate weights, but nowhere near enough data to reproduce the scores yourself. For the typical consumer and company, they're a black box just like a ML model.
- judge2020 4y agoChecked the Experian website (their "freecreditreport" product, aka the thing that tricks people into foregoing annualcreditreport) and they provide weights for how much each contributes: Payment History: 35% of score. in no particular order: - Late payments 30+ Days ("About 98% of FICO High Achievers have no missed payments at all") / Late Payments 60+ Days / Accounts Always Paid as Agreed / No Delinquent Accounts - Collections ("FICO® Score 8 only considers collections with an amount of $100 or greater. Virtually no FICO High Achievers have a collection listed on their credit report") / Late Payments 60+ Days / Derogatory Public Records Amount of Debt: 30% of score. in no particular order: - % revolving credit: ("For FICO High Achievers, the average ratio is less than 7%.") - Number of Accounts with Balances ("FICO High Achievers have an average of 3 accounts carrying a balance.") - Total balance on revolving and open-ended accounts ("Most FICO High Achievers owe less than $2,500 on revolving and/or open-ended accounts such as credit cards, charge cards and department store cards.") Length of credit history: 15% of score - Average age of accounts ("Most FICO High Achievers have an average age of accounts of 9 years or more.") - Age of Oldest Account ("FICO High Achievers opened their oldest account 25 years ago, on average.") Amount of new credit: 10% of score ("FICO High Achievers opened their most recent account 2 years, 7 months ago, on average.") Credit Mix: 10% of score ("FICO High Achievers have an average of 11 revolving accounts, 5 installment accounts, 6 credit cards) Also, a trend I see is that almost every option says "Authorized user accounts aren't considered in the calculation of this attribute", which might be a big contributor as to why Credit Karma / Vantage Score numbers are almost always higher than FICO, which leads to people thinking they have a higher number when they walk into a car dealership / apply for a loan. But you're right in that I don't see any actual weights or FICO simulators.