Score impact of total balance in revolving accounts

[kilroy8]

To be clear this is not a question about utilization percentages. That is well documented and understood.

I am wondering about the score impact of total balances in revolving debt.

Example:

John has $1,000 in total revolving balances and $10,000 in total revolving credit lines. His utilization is 10%.

Mary has $10,000 in total revolving balances and $100,000 in total revolving credit lines. Her utilization is 10%.

Joe has $50,000 in total revolving balances and $500,000 in total revolving credit lines. His utilization is 10%.

We know that income is not a factor in FICO scoring. Because Mary owes 10 times as much as John - does she lose more points? What about Joe? He owes 50 times as much as John ! Assuming they all make the same amount of money (effectively what FICO does), Joe could be in big trouble - so his score should somehow reflect that risk.

[iv]

---

@kilroy8 wrote:

To be clear this is not a question about utilization percentages. That is well documented and understood.

I am wondering about the score impact of total balances in revolving debt.

Example:

John has $1,000 in total revolving balances and $10,000 in total revolving credit lines. His utilization is 10%.

Mary has $10,000 in total revolving balances and $100,000 in total revolving credit lines. Her utilization is 10%.

Joe has $50,000 in total revolving balances and $500,000 in total revolving credit lines. His utilization is 10%.

We know that income is not a factor in FICO scoring. Because Mary owes 10 times as much as John - does she lose more points? What about Joe? He owes 50 times as much as John ! Assuming they all make the same amount of money (effectively what FICO does), Joe could be in big trouble - so his score should somehow reflect that risk.

---

Nope. What FICO does is more like assuming that total credit lines scale perfectly with income.

There are other scoring models that do work as you suggest they should - where the actual dollar values matter.

Just not the well-known FICO models...

[Anonymous]

Hi iv.  I am not sure we have had a carefully designed test that rules out dollar values mattering.  A test like that could be designed.  It wouldn't be too hard.  But I don't think one has ever been done.

Some people on the forum (like Thomas Thumb) are open to the possibility that it might be a factor in one or more FICO models.

One reason I am myself open to that possibility is that some people have claimed that they get some advantage from going from 8% total U to a smaller amount (like 1%).  Other people are completely unable to replicate that claim (they get no advantage).  When I have tried to probe more deeply into the profiles of the people who do note a difference, they almost all have huge total credit limits (like SouthJamaica, for example).  Thus when SouthJ goes from $45k of debt to $45 of debt, he's seen a difference (while staying at under 8.99% total U in both cases) -- if dollar values matter then that would explain it.

[iv]

---

@Anonymous wrote:

Hi iv.  I am not sure we have had a carefully designed test that rules out dollar values mattering.  A test like that could be designed.  It wouldn't be too hard.  But I don't think one has ever been done.

---

Given that it is possible to see no significant score changes between $1 reporting and many 10s of thousands reporting... I'd feel quite confident is asserting that there is no dollar value scoring factor in standard FICO models. If it did exist, we would see dramatic shifts based on reporting/payoff for those with very high limits. But we don't - we see no change, or at best, noise.

---

@Anonymous wrote:

One reason I am myself open to that possibility is that some people have claimed that they get some advantage from going from 8% total U to a smaller amount (like 1%).  Other people are completely unable to replicate that claim (they get no advantage).  When I have tried to probe more deeply into the profiles of the people who do note a difference, they almost all have huge total credit limits (like SouthJamaica, for example).  Thus when SouthJ goes from $45k of debt to $45 of debt, he's seen a difference (while staying at under 8.99% total U in both cases) -- if dollar values matter then that would explain it.

---

The easier explanation there is that the "common wisdom" of 8.99%, 28.99%, etc... is correct, but incomplete, and there are simply other utilization breakpoints in the 0-9% range, at least for some scorecards.

Years ago, I also saw very minor (noise-level) score differences for utilization in the sub-9% range (# of card reporting balances could also have been a factor). That was with sub-100k limits. Today (with much higher limits), I can see zero impact, or noise-level impact from 50k+ changes in reported balances.... (PIF after statement cut can report wild +/- swings some months.)

If that kind of dollar-amount swing doesn't show a real impact... what would?

[kilroy8]

When you hover over the "?" next to the total amount of revolving debt on the MyFico score page, a little box comes up that says:

"Most FICO High Achievers owe less than $2,500 on revolving accounts such as credit cards and department store cards."

This is on the FICO 8 score page pertaining to Amount of Debt.

[Anonymous]

---

@kilroy8 wrote:

When you hover over the "?" next to the total amount of revolving debt on the MyFico score page, a little box comes up that says:

"Most FICO High Achievers owe less than $2,500 on revolving accounts such as credit cards and department store cards."

This is on the FICO 8 score page pertaining to Amount of Debt.

---

Yep.  This is one of the reasons to think that some FICO models might consider dollar values -- because FICO itself has written something that implies this could be the case.  There are other places where FICO says stuff that suggests this.  (E.g the "Amounts Owed" page under Credit Education on this website.)

I remain agnostic (as I said earlier) but would like to see some well designed tests, at which point I would be more likely to have an opinion one way or the other.

[Anonymous]

In my research ive found that total $ amount of balances do in fact matter in certain (if not all FICO models) it certainly does in VS3.0 as well. Here is a chart from another poster on the matter.

The way I understand it based on which FICO version it works one of two ways:

Model 1: Average revolving acct balance (0 to $250, $250 to $500, $500 to $1,000 etc)

Here is another graph link:

http://s159.photobucket.com/user/BobWangPhotos/media/BalanceampUtilization_zps0b70a7f5.jpg.html?sort...

[Anonymous]

To further evidence the balance factor, MyFicos own score estimator has the following break lines: 0 to 500, 500 to 1000, 1000 to 5000, 5000 to 10,000 and 10,000 to 20,000.

They deduct about 10 points for 20k+ and 5 points for anything between 1000 to 20000

[Revelate]

I know I'm beating a dead horse on this one however this is another instance:

You cannot trust ANY third party presentation and interpretation of your credit report as gospel for the FICO algorithm (specifically I'm talking about their mainstream models, they have a bunch of one-offs which we never will see and maybe it's different there to one of iv's excellent points on this topic).  Score shifts, and reason codes, and that's it for trying to analyze it.

Not even myFICO: to wit, I've sent in multiple comments for mistakes in their products and they check with their internal people and have fixed the issues.  The people in FICO Consumer are not the same ones that work on the algorithms, and that's a tightly controlled piece of IP even inside the company, always has been.

We've tested this, even I have, with 20k of balances and nothing budged.  What did?  Vantage, VS 3.0 at least absolutely looks at your aggregate revolving debt, FICO does not through FICO 8 and I don't think FICO 9 either though I don't track those scores well, wish Penfed did more than every 3 months and had a historical credit page like almost everyone else does... Penfed /sigh.

I agree it doesn't make that much sense with how credit limits have exploded over the past 5 years, and maybe FICO X or whatever it's going to be called might switch to an aggregate debt model, but FICO explicitly leaves DTI calculations to the lenders and as such it makes absolutely no judgement (perhaps rightly) because incomes are not on one's credit report and that's the only information FICO algorithms look at.

Today anyway.

Also, I would note that Vantage is seemingly not making further inroads: FWIW currently the overwhelming majority of lenders agree with FICO's algorithm design decisions.

[Revelate]

Well jeez, nothing like pulling a 1B Experian report to see that in fact I was mistaken.

EX FICO 2: explicitly a two decade old FICO 98 model still in use for the mortgage trifecta:

• 4. The balances on your non-mortgage credit accounts are too high.

That is from a HELOC.  And it's a double-whammy too as get this:

• 1. You've made heavy use of your available revolving credit.

EX FICO 3 that I don't know of a single lender that ever used:

• 2. The amount owed on your revolving and/or open-ended accounts is too high. 

For reference the HELOC is sitting at $18,889 / $27,416, high balance according to the credit report of $19,047 and to be fair I'm not 100% certain how HELOC's are scored under FICO 98 I assumed standard revolving with the limit taken into effect, I know they aren't excluded like they are in EQ FICO 5 / TU FICO 4 and all FICO 8/9 models.

Interestingly appears I might have gotten a similar +30/40ish points with October similar to EQ FICO 5, wish I'd looked at this a few days ago to see if that happened early as it appeared to on EQ.

ETA: the only other revolving balance on there is $1,930/$43,500 which presumably is a non-issue, everything else $0.