Crossed Two Utilization Thresholds With No Score Change

[Anonymous]

Hi TT.  Thanks for producing that table.  Many of the newcomers here may not be aware that trended data like these (e.g. one's balance and amount paid on that balance for each months strecthing back 24 or more months) are reported to the CRAs -- i.e. not merely the most recent balance.

But again, this is a fairly recent thing in the scheme of credit reporting (last few years).  Certainly when FICO 8 was developed (and in the few years after as well) these data were not being reported to the bureaus.  Therefore the FICO 8 algorithm could not have been designed to detect whether a card had been used multiple times in the last year.  As I mentioned earlier, the best the algorithm could do is to use the DOLA to detect the most recent time that the card had been used. 

[Thomas_Thumb]

CGID -

Balances are fundamental to Fico determining individual and aggregate utilization. I don't have any credit reports from 2010 to see if how far back monthly balances are maintained. The 1st report I ever pulled was from EQ in 2014 and it listed monthly balances for the prior 3 years as I recall. I may check hard copies tonight.So, looking at prior balances certainly could be built into Fico 8. Positive balances could be used as an indicator of an account being in use without resorting to trending. If such a thing were done then we have one of the few opportunities where PIF after statement adherants could have an edge over PIF before statement adherants.

The elements of amount paid and past due was reported monthly by many CC companies starting in 2014 and continued thru mid 2015. It since fell out of favor and has been all but abandoned.That additional data is fundamental to transactor/revolver behavior identification and provided some trending analysis opportunities that could improve risk analysis.

With regard to individual card utilization and impact on score regularity of card use is likely overshadowed by other things such as:

 - card being new (say 3 months old) going to high utilization vs established.card (lets say over 36 months old) going to high utilization

- card being active and going to high utilization vs inactive card becoming active and going to high utilization

- dirty vs clean profile

- file age and file AAoA

[Thomas_Thumb]

---

@Anonymous wrote:

TT, do we have any data points regarding the "highly established" card with high utilization relative to one that isn't highly established?  I suppose the best way to test this would be the following and I guess my question is whether or not anyone has done (and posted on here) doing this:

Someone has a bunch of CCs.  They are at AZEO except 1 card.  That one card is reporting high utilization, say 70%+.  That card is a highly established card, which you define as 3+ years old.  The next cycle, they bring that card down to a $0 reported balance and move their 70%+ utilization reported balance to a different card, except this card is not highly established.  For the sake of this illustration let's say it's < 1 year old.  Of course this excercise could very well work in the reverse, where the younger aged card is used first, followed by the aged card second.

Has anyone done this and reported scoring results?

---

BBS - Unfortunately no. All I know is that a few posters with young profiles have reported some rather pronounced shifts in Fico 8 score due to changes in revolving credit utilization on recently opened accounts.

The candidate for your test would need to have two low limit cards (a new account - say 2 to 4 months old and an aged account - at least 36 months old), a high aggregate credit limit, say of $100k, a credit age of 10 years or more and an AAoA of at least 2 years (AAoA of 5 years or more would be best).

I'd be interested in seeing some data along the lines of the table below for different types of profiles. The ideal candidate would have a minimum $60k aggregate CL, say 8 to 12 cards a lengthy credit history and AAoA of 5 years or more and have an established low limit account to test. The minimum 8 accounts provides a margin of safety becauses inadvertantly allowing a 2nd ("high limit") card to report a small balance should not influence results. The Alternate comparison candidate would have an AAoA under 2 years and test a new low limit account.

I would anticipate results would vary depending on if the profile has any derogatories. Thus, need data from both clean and dirty profiles.

Here is a test plan that would provide some valuable data:

Test #	$500 CL card	UT% test	Fico 8 Score
1	$460	above 89%
2	$350	above 69%
3	$260	above 49%
4	$160	above 29%
5	$40	below 9%

[Anonymous]

For the use of trended data to have been included in FICO 8 (specifically the history of monthly balances stretching back 24 months) these data would have had to have been available when FICO was developing the model.  I.e. in 2008.  Monthly balances were not being submitted back then.

And as you observed earlier, a series of $0 balances would not have implied that the card was unused, and so even if these data had been avaiable, it's doubtful that the developers would have used them to include a brand new scoring factor ("Has this card been used multiple times in the last year?").  And if they had wanted to do so, it is doubtful that they would have done so until most CC issuers were reporting them to all three CRAs, which was certainly not happening in 2008.