topic Re: VS 3.0 vs VS 4.0 in General Credit Topics

VS 3.0 vs VS 4.0

Anonymous — Fri, 06 Sep 2019 17:58:18 GMT

While not "Fico" scoring, I haven't really yet seen a discussion on this topic. I haven't seen TT post in a while and I know he's pretty up on VS, so it would be great if he stopped in. Anyway, I was wondering if anyone knew any of the differences implemented for VS 4.0 relative to the all too common VS 3.0? I ask because Synchrony recently moved from offering a Fico score (TU 08) monthly on their accounts to offering a TU VS 4.0 instead. While overall this change isn't favorable, it was the first time I've seen my VS 4.0.

My TU VS 4.0 is 850, where the highest TU VS 3.0 I've ever had was 839; Typically my TU VS 3.0 varies 836-839 and only drops as low as 831 if I report a high balance on a card with several other balances reported. I believe TT once said that the highest reported VS 3.0 was 840-841 (which perhaps he had) so it seems to me that one change with VS 4.0 is that it's more forgiving of certain factors, as evidenced by the fact that a perfect score can be achieved there.

Re: VS 3.0 vs VS 4.0

BallBounces — Fri, 06 Sep 2019 18:45:24 GMT

The topic must at least be a little entertaining then, BBS ....

I will be interested in informed opinions. I have none I would like to share, beyond the fact the VS4 is a lot more FICOish than VS3.

Re: VS 3.0 vs VS 4.0

Anonymous — Fri, 06 Sep 2019 19:10:44 GMT

I don't know about entertaining, but I find it interesting because it allows me to learn something new.

I do not find the routine tracking of VS 3.0 scores to be entertaining at all, especially when it leads to "Why did my VS 3.0 change and Fico didn't" type posts all the time.

Re: VS 3.0 vs VS 4.0

Anonymous — Sat, 07 Sep 2019 09:21:28 GMT

It's going to get a lot more interesting soon: ( housingwire.com )

"In a rather surprising development, the Federal Housing Finance Agency announced Tuesday that it will allow Fannie Mae and Freddie Mac to consider using VantageScore as an alternative to their current FICO credit scoring model, a dramatic reversal from a proposed rule issued late last year."

They are definitely submitting VS4 for consideration.

It looks a lot like FICO 9, but with trended data and of course, the ever-popular 'leverages machine learning' feature.

"This rich new data source [trended data] helps VantageScore 4.0 refine its scoring accuracy, particularly among high-scoring consumers with well-established credit habits." vantagescore.com

Re: VS 3.0 vs VS 4.0

Anonymous — Sun, 08 Sep 2019 04:10:06 GMT

I'm not understanding how VS 4.0 is using trended data if it isn't available on a CR. I believe that TU does harbor trended data, so if VS 4.0 is able to use that in its score generation that's a good thing, but AFAIK the other 2B (EX, EQ) were not keeping the data that could be used by a scoring model for "trended data" purposes. Those more in the know than me, definitely provide some insight.

If VS 4.0 is using (TU) trended data, that could explain a major difference between VS 4.0 and VS 3.0, the purpose of this thread. That being said, it would make sense that my VS 4.0 score is better than my VS 3.0 scores, as when looking at my trended data from the last (say) 2 years the algorithm would see I'm a strict Transactor that has always PIF on every single one of my accounts, something that should signify a sign of extremely low risk.

Re: VS 3.0 vs VS 4.0

iv — Sun, 08 Sep 2019 05:42:10 GMT

@Anonymous wrote:
I'm not understanding how VS 4.0 is using trended data if it isn't available on a CR. I believe that TU does harbor trended data, so if VS 4.0 is able to use that in its score generation that's a good thing, but AFAIK the other 2B (EX, EQ) were not keeping the data that could be used by a scoring model for "trended data" purposes. Those more in the know than me, definitely provide some insight.

All three CRAs store the same trended data fields.

But... only if that data is actually provided from the lenders.

And not only do not all lenders report that data - some that previously did report, stopped doing so a few years ago.

Pull your annualcreditreport.com reports (or the state-level equivalents, or the paid full reports, not the "fluffy" ones), to see the full details currently provided and stored. Paper reports are probably the best for this, but the online ACR does show this fairly well.

All three CRAs have monthly tracking fields for Balance, Scheduled Payment, Payment Date, Amount Paid/Actual Payment, High Credit, Credit Limit, and Paid/30/60/90/etc.

In general, installment loans, mortgages, even HELOCs report full data for trending purposes (balance, scheduled payment, payment date/actual payment).

Revolving accounts vary quite a bit by lender:

Amazon/SyncB: Balance/Scheduled Payment/Actual Payment Date/Actual Payment Amount

American Express: Balance/Scheduled Payment

Bank of America: Balance/Scheduled Payment/Actual Payment Date

Citi: Balance/Scheduled Payment/Actual Payment Date

Discover: Balance/Scheduled Payment/Actual Payment Date

JPM Chase: Balance/Scheduled Payment/Actual Payment Date

Wells Fargo: Balance/Scheduled Payment/Actual Payment Date/Actual Payment Amount

What's interesting is that Bank of America, Chase, and Citi previously did report Actual Amount Paid... but stopped (around 2015-ish), while SyncB started reporting full data in 2016. (No idea why Amex doesn't report actual payment date - but not that long ago they didn't even report scheduled payment amounts.)

Given how sparse the full trended data is, I'd guess that analysis of balance direction (increase/decrease) makes up the majority of the "trended" part of newer scoring models. I'm sure score developers would love to be able to depend on "Actual Payment" data... but most major lenders appear to have decided that sharing that level of data is not in their best interests.

Re: VS 3.0 vs VS 4.0

Anonymous — Sun, 08 Sep 2019 11:02:40 GMT

@iv wrote:
@Anonymous wrote:
I'm not understanding how VS 4.0 is using trended data if it isn't available on a CR. I believe that TU does harbor trended data, so if VS 4.0 is able to use that in its score generation that's a good thing, but AFAIK the other 2B (EX, EQ) were not keeping the data that could be used by a scoring model for "trended data" purposes. Those more in the know than me, definitely provide some insight.
All three CRAs store the same trended data fields.

But... only if that data is actually provided from the lenders.

And not only do not all lenders report that data - some that previously did report, stopped doing so a few years ago.

Pull your annualcreditreport.com reports (or the state-level equivalents, or the paid full reports, not the "fluffy" ones), to see the full details currently provided and stored. Paper reports are probably the best for this, but the online ACR does show this fairly well.

All three CRAs have monthly tracking fields for Balance, Scheduled Payment, Payment Date, Amount Paid/Actual Payment, High Credit, Credit Limit, and Paid/30/60/90/etc.

Great post, iv. How do we absolutely know that some lenders stopped reporting the full set of data?

Just because it's not on the full credit report doesn't mean they haven't had it stored in a database for several years.

Look at this bit of information from 6 years ago: ( https://www.creditcards.com/credit-card-news/credit_reports-reveal-card-bill-paying-data-1270.php )

"Why is the payment detail being added to credit reports now? The “Metro 2” format that lenders use to report data to credit bureaus already includes payment details, so the information is not new. But now credit bureaus say they are devoting the computer resources to maintaining the figures on millions of credit reports, after testing the data to meet FCRA standards." (emphasis mine)

That suggests to me that our credit reports are just a database view representing a subset of the data. We all know the CRA's collect an enormous amount of data extending far beyond simple tradeline data, for use in their various marketing products and not necessarily credit scoring.

I would probably be horrified after running a SQL query that joins all their various tables on my primary key.

Re: VS 3.0 vs VS 4.0

iv — Mon, 09 Sep 2019 01:22:08 GMT

@Anonymous wrote:
Great post, iv. How do we absolutely know that some lenders stopped reporting the full set of data?

I watched it happen on my reports at the time... the fields have been there, visible for many years, and multiple banks had been providing the data.

Those fields are still there on your full reports, but many of those banks stopped providing that data (right around when T/R-based scoring started really making waves). They now show as $0 or blank for the lenders that don't report those fields.

I've got copies of annual reports in the 2014, 2015, 2016 range, showing banks dropping the reporting (and at least one starting, where they hadn't reported those details previously). It's very clearly choices made by individual lenders to report or not report - and the CRAs are just displaying whatever subset of data they are actually getting from each lender. Pull a full ACR and see which of your lenders supply full vs partial monthly data.

@Anonymous wrote:
Just because it's not on the full credit report doesn't mean they haven't had it stored in a database for several years.

Failing to provide that data (if stored on your report in the CRA's database) on the full annual report would be a violation of Title 15 US Code Section 1681g(a)(1), which requires "All information in the consumer's file...".

@Anonymous wrote:
That suggests to me that our credit reports are just a database view representing a subset of the data. We all know the CRA's collect an enormous amount of data extending far beyond simple tradeline data, for use in their various marketing products and not necessarily credit scoring.

I would probably be horrified after running a SQL query that joins all their various tables on my primary key.

Ah... but the catch there is it's not technically in the same CRA database!

That additional marketing, etc data... is in the databases of the wholly-owned subsidiaries of the CRAs established for the purpose, most of which manage to not legally count as "consumer reporting agencies". They may share ownership, management, and infrastructure... but legally they are separate.

(Did you check out your Acxiom profile back when it was available on aboutthedata.com? That was both horrifying and amusing.)

Re: VS 3.0 vs VS 4.0

Anonymous — Mon, 09 Sep 2019 02:29:02 GMT

I feel like the last time I did look at my real hard copy credit reports that the TU report had more data on it such as amounts paid and such each cycle, where EX/EQ lacked the same data from the same lenders. It's been probably a good year or longer since I've pulled them, so perhaps it's worth another look.

Re: VS 3.0 vs VS 4.0

iv — Mon, 09 Sep 2019 02:48:02 GMT

@Anonymous wrote:
I feel like the last time I did look at my real hard copy credit reports that the TU report had more data on it such as amounts paid and such each cycle, where EX/EQ lacked the same data from the same lenders. It's been probably a good year or longer since I've pulled them, so perhaps it's worth another look.

I've got my ACR reports from 2014 to 2019 in front of me right now... and all three show those fields, in slightly different display formats.

In my case, if a lender was reporting a certain data field to one CRA, they were reporting it to all. YMMV, depending on your lenders.

Re: VS 3.0 vs VS 4.0

Anonymous — Mon, 09 Sep 2019 12:05:56 GMT

There is more to this that we're not seeing, and I say this because VantageScore has been promoting use of their VS4 model using statements like this: "The VantageScore 4.0 model takes advantage of trended credit data newly available from all three national CRCs."

And then they specifically mention payment behavior, so we know it's not just utilization rate: "By capturing the trajectory of borrowing and payment behaviors, trended credit data provides a more precise, holistic view of consumer habits."

And all 3 CRA's have various products they have been selling for many years that use trended data - with payment details - for several years now.

It's clear they have the payment details. They also need at least 2 years of this data to build a new scoring model, so they had to have been collecting it for a while now. And Fannie Mae has been using it in underwriting for several years now.

I wonder if this 'payment amount/payment data' information is exempt from some reporting requirements because it's not currently being used in any credit scoring model.

Re: VS 3.0 vs VS 4.0

iv — Mon, 16 Sep 2019 00:17:58 GMT

@Anonymous wrote:
There is more to this that we're not seeing, and I say this because VantageScore has been promoting use of their VS4 model using statements like this: "The VantageScore 4.0 model takes advantage of trended credit data newly available from all three national CRCs."

They can only take advantage of what's being reported...

@Anonymous wrote:
And then they specifically mention payment behavior, so we know it's not just utilization rate: "By capturing the trajectory of borrowing and payment behaviors, trended credit data provides a more precise, holistic view of consumer habits."
And all 3 CRA's have various products they have been selling for many years that use trended data - with payment details - for several years now.

While they would love to use all of the possible fields... notice the phrasing: "capturing the trajectory". The one data field that is reliably available across major lenders is monthly balance. You can do a lot with the trajectory of that balance - increasing, decreasing, or stable. It's not as good as having the actual monthly payment data, but it's still valid trend data.

Are the full set of data fields factored in for the few lenders that report them? Maybe? But trended data based on 24 to 30 months of balance history is still trended data. (Both for actual stats and for marketing purposes...)

@Anonymous wrote:
It's clear they have the payment details. They also need at least 2 years of this data to build a new scoring model, so they had to have been collecting it for a while now. And Fannie Mae has been using it in underwriting for several years now.

I disagree that it's clear. There is every indication that score developers and Fannie/Freddie WANT trended data. There is also solid evidence that most lenders are declining to provide it, even if they had previously done so.

@Anonymous wrote:
I wonder if this 'payment amount/payment data' information is exempt from some reporting requirements because it's not currently being used in any credit scoring model.

There is no exemption in 1681g based on whether or not a data field is used in scoring of any sort. If you see a blank field on your ACR-pulled reports, that field has no data.

Re: VS 3.0 vs VS 4.0

Anonymous — Mon, 16 Sep 2019 02:53:35 GMT

@iv wrote:
They can only take advantage of what's being reported...

I was thinking that the only way for me to know it's not being reported is to see an actual lender's report in Metro2 format where Field 16 - Actual Payment Amount is blank or something else being used as a 'null' value.

Your information makes sense to me, especially concerning FCRA requirements, but then I can find several dozen sources going back years that say they have been and/or are using the payment amounts in trended data from at least Equifax and TransUnion (Experian was exempt in 2016, but said they would start within a year.)

Like this source: National Mortgage Professional Magazine (March 2016)

"This expanded view of a consumer’s credit history can reveal trends and behaviors that could never be detected with the current credit reports that show only the most recent payments on the account...[These payment trends cannot be seen on standard versions of the consumer credit report] or picked up by any of the traditional risk scores that use the current non-trended data to calculate the score."

And then TransUnion itself, from October 2015 in a statement from their CEO:

"The majority of lenders have been reporting actual payment amount to TransUnion on a consistent basis, and we’re pleased to see this trend,” said Peck.

Also, CoreLogic: (PDF link.)

Q. Will Trended Credit Data be displayed on all active tradelines from Equifax and TransUnion?

A. Yes, if provided by the bureaus, trended credit data for the most recent 24 months will be displayed on all active tradelines from Equifax and TransUnion. Each bureau has exclusions for trended data display, [so it is possible for the information not to appear on every active tradeline].

And yet even with all those sources out there saying they have the data, one can come here to the forums and see plenty of reports from people that have pulled their full credit report where the payment amount is not shown.

On my own full report from March, the 'Actual Payment Amount' for my Citi issued Visa is blank or $0 on EQ/TU/EX.

My credit union issued Mastercard (processed by PSCU which handles hundreds of different CU cards), on EQ/TU anyway, includes paymount amount and exact date of payment.

It is possible that Citi, a major lender for sure, doesn't report 'actual payment amount' for any of its accounts. That just conflicts with so many sources out there, but maybe they just stopped doing it at some point, like you suggested.

Re: VS 3.0 vs VS 4.0

AverageJoesCredit — Mon, 16 Sep 2019 16:54:43 GMT

3.5

Re: VS 3.0 vs VS 4.0

Anonymous — Sun, 29 Sep 2019 07:33:46 GMT

Just a note, the quote from Corelogic stating that each bureau has its own exclusions could be a nice way of stating that if they don’t have it, it is excluded. 😉

I also wonder if it’s possible that the bureaus have wholly-owned subsidiaries created to receive the data from the lenders, and then only pass on the data they want to the actual databases at the CRA’s themselves? This way they could legally hold the trended data in separate databases separate from the actual CRAs and legally state the CRA’s neither have the data nor ever received it.

There are many ways to be legally creative, as iv suggested for the additional data thats held for marketing purposes.