Credit Scoring Primer, pub.5.17.20

[Curious_George2]

---

@Anonymous wrote:

---

---

But as I said you're not being penalized for not having a mortgage, however you're leaving points on the table, because you could accrue additional bonus points by having a mortgage at the right levels.

Keep in mind  negative scoring factors are there to tell you that you could be earning at least one more point by taking some action; that doesn't mean you're losing points by not doing so. 

in other words not having a mortgage doesn't penalize you, but if you did have one at appropriate levels it could offer additional points, imho.

---

It's an interesting theory, but I'm not persuaded. The reason code says lack of mortgage. If it's present, the score is lower than it would be if it were absent. 

I didn't get the sense from Tom Quinn, either in his written responses in the AMA thread or in hearing him interviewed on the podcast Cassie linked to somewhere, that he would make a distinction between a scoring penalty and a missed opportunity to gain points. That's the kind of distinction we hobbyists make when trying to create logical narratives. I have used exactly the same semantic trick to get myself somewhat comfortable with the all-zero penalty. But that's not how FICO insiders talk. They don't deny the existence of the all-zero penalty. Instead they admit that people with recent credit usage score higher than those without and then explain why. (Link)

I am struggling to accept that Mr. Quinn would say there's no penalty for not having a mortgage while secretly meaning what you suggest he meant. From a scoring standpoint, it's a distinction without a difference.

[Anonymous]

---

@Curious_George2 wrote:

---

@Anonymous wrote:

---

---

But as I said you're not being penalized for not having a mortgage, however you're leaving points on the table, because you could accrue additional bonus points by having a mortgage at the right levels.

Keep in mind  negative scoring factors are there to tell you that you could be earning at least one more point by taking some action; that doesn't mean you're losing points by not doing so. 

in other words not having a mortgage doesn't penalize you, but if you did have one at appropriate levels it could offer additional points, imho.

---

It's an interesting theory, but I'm not persuaded. The reason code says lack of mortgage. If it's present, the score is lower than it would be if it were absent. 

I didn't get the sense from Tom Quinn, either in his written responses in the AMA thread or in hearing him interviewed on the podcast Cassie linked to somewhere, that he would make a distinction between a scoring penalty and a missed opportunity to gain points. That's the kind of distinction we hobbyists make when trying to create logical narratives. I have used exactly the same semantic trick to get myself somewhat comfortable with the all-zero penalty. But that's not how FICO insiders talk. They don't deny the existence of the all-zero penalty. Instead they admit that people with recent credit usage score higher than those without and then explain why. (Link)

I am struggling to accept that Mr. Quinn would say there's no penalty for not having a mortgage while secretly meaning what you suggest he meant. From a scoring standpoint, it's a distinction without a difference.

---

@Curious_George2 I respect that but let me put it in a little different perspective and see if that helps. Each scoring category is independent and has things that can help or hurt it.

For example, an inquiry is a penalty. However if you don't take any inquiries, then I guess you could consider it a bonus.

It's all semantics, but the algorithm is designed to take away points for an inquiry and it's designed to add points when you have loans with certain attributes. 

we can look at it as half full or half empty but an inquiry is a penalty by your action and a partially paid down loan is a bonus by your action of earning it. 

all he was saying is they don't penalize you for not having a mortgage. That's not tantamount to there is no additional reward for an aged paid down loan.

[Anonymous]

@Curious_George2 Another quick thought. Look at the example scorecard in post one under Scorecard Basics.

Understand each category can offer a certain amount of points, but if you don’t have your profile in the correct order, then you’re not gonna get the maximum points for each category.

Just cause you don’t get the maximum amount doesn’t mean you’re being penalized, it just means you didn’t do the things necessary to get those extra points.

However there are things you can do to lower your score with true penalties like opening a new account or getting an inquiry. Those penalize you; simply failing to maximize a category is not tantamount to a penalty, imho.

[Anonymous]

---

@Curious_George2 wrote:

---

@Anonymous wrote:

---

---

But as I said you're not being penalized for not having a mortgage, however you're leaving points on the table, because you could accrue additional bonus points by having a mortgage at the right levels.

Keep in mind  negative scoring factors are there to tell you that you could be earning at least one more point by taking some action; that doesn't mean you're losing points by not doing so. 

in other words not having a mortgage doesn't penalize you, but if you did have one at appropriate levels it could offer additional points, imho.

---

It's an interesting theory, but I'm not persuaded. The reason code says lack of mortgage. If it's present, the score is lower than it would be if it were absent. 

I didn't get the sense from Tom Quinn, either in his written responses in the AMA thread or in hearing him interviewed on the podcast Cassie linked to somewhere, that he would make a distinction between a scoring penalty and a missed opportunity to gain points. That's the kind of distinction we hobbyists make when trying to create logical narratives. I have used exactly the same semantic trick to get myself somewhat comfortable with the all-zero penalty. But that's not how FICO insiders talk. They don't deny the existence of the all-zero penalty. Instead they admit that people with recent credit usage score higher than those without and then explain why. (Link)

I am struggling to accept that Mr. Quinn would say there's no penalty for not having a mortgage while secretly meaning what you suggest he meant. From a scoring standpoint, it's a distinction without a difference.

---

@Curious_George2 There is most certainly a difference between a penalty and a missed opportunity to gain points. So you're telling me that you're penalized for not having a loan at under 9%? Because you will have that reason code until you do. That doesn't mean you're being penalized; that means you have an opportunity to improve your profile. 

likewise if you only have a loan on your profile with no revolvers you're not being penalized for not having revolvers, but you're not gonna get points for diversity credit mix until you do. Nor will you get scoring factor points for having a certain number of revolvers via mix, until you acquire those revolvers

That's why we build credit. We don't start off with an 850 and then they penalize until you get your score. No, your profile is analyzed to see how many points are earned in each category, or penalized for things like new accounts and inquiries. It's not really an enigma, imho.

An inquiry you're penalized for; the lack of a mortgage you're not penalized for. The lack of inquiries would result in a lesser penalty, just like a mortgage at proper levels would result in a higher award for certain categories.

[Anonymous]

@Curious_George2 I apologize for the multiple posts but my mind is everywhere and if I were to go back and edit there's a chance you would miss it.

I wanted to explain about the all zero penalty. Believe it or not it's not a penalty. lol, I know, many will disagree, but hear me out. 

Aggregate revolver balance is a characteristic. For the value of zero you get zero points. For the value of $1 - $499, you might get 13 points, scorecard and profile depending. For $500-$999, you might get 12 points; for $1000 to $4999 you might get 10 points; between $5000-$9999, you might get 8 points; $10,000-$24,999, maybe 6; and over $25,000, maybe 4 points. 

now, that's just aggregate revolving balances. each characteristic has a similar way points are calculated for it, I believe. 

Now you've got a look at number of accounts with a balance, percentage of revolvers with a balance, aggregate and individual revolving utilizations. Then you've got to do balances and utilization for installment, ect...the total of all your points will be the points contributed from the amounts owed category. combine that with the points from the other four categories and you have your total score. 

look at the example scorecard and see if that makes sense? Because although they are giving a simplified scorecard with only one characteristic for each of the five categories, there are actually many characteristics for each category.

[Curious_George2]

---

@Anonymous wrote:

---

@Curious_George2 wrote:

One theory I have is that these factors could be related to how credit mix is evaluated within the scoring system, but perhaps they are never actually used as reason codes. In other words, maybe they exist behind the scenes but are never deployed. That seems odd, but possible.

I don't think it's odd at all. All of those reason codes show us what questions they are asking (characteristics) about profile data, and by monitoring when they appear, we can figure out some rough thresholds or ranges (attribute range).

I may never see 'Lack of recent reported morgage loan information' or 'Lack of recent revolving HELOC information', but the code was definitely looking for a mortgage or HELOC on my file. No mortgage, no need to check for 'too many mortgage loans with balances'.

---

The reason I thought this was odd is that the score factors we are talking about are outputs. They aren't inputs or values derived and used solely within the scoring software. 

But I think I've found an answer to my own question. Somewhat surprisingly -- to me anyway -- there is no reason code for credit mix. I expected to see one that said something like "Insufficient variety of credit account types." But it's not there. Not in any of the five tables you've posted so far in the Score Factors thread. Weird, right? We know this is a scoring factor, so why isn't there a single reason code for it? I think the answer is that there are several reason codes for it... and my bête noire du jour, 79, is one of them. So, if a consumer has only revolvers on their report, they would lose points for credit mix and the reason codes that would be returned by the FICO TU8 and EX8 software boxes (though maybe not revealed to the consumer for reasons we all understand) would presumably include some or all of 79 (lack of mortgage), 32 (lack of installment loan), maybe 59 (lack of HELOC) and maybe even 50 (lack of retail accounts).

It's not that any one of those things is required, it's just that they've chosen to describe a poor mix by itemizing the credit types that are not represented. Slightly counterintuitive, but I guess it works and it's perfectly consistent with what Tom said in the AMA.  

Here's an interesting twist on this: consumer finance accounts might contribute to credit mix! Over at the official myFICO Credit Education entry for What Does Credit Mix Mean? it says the mix can include "credit cards, retail accounts, installment loans, finance company and mortgage loans." It seems likely that the negatives of a CFA would far outweigh this one little positive, but this would explain the existence of seemingly contradictory reason codes 99 and 6 in EX8 and TU8.

[Anonymous]

And just to draw a contrast for example, the characteristic number of inquiries:

0 would equal maybe +25,
1 would equal maybe +21,
2 maybe +17,
3 maybe +17,
4 maybe +12,
5 maybe +8,
6 maybe +8,
7 maybe +5,
8 maybe +2 and
9 maybe zero.

So we see with inquiries that your action of acquiring them penalizes you.

Let’s do the same example with amounts owed, the characteristic I did above, revolving balances:

$0 maybe 0,
$1 - $499 maybe 13 points,
$500-$999 maybe 12 points,
$1000 to $4999 maybe 10 points,
$5000-$9999, maybe 8 points,
$10,000-$24,999, maybe 6; and
$25,000+ maybe 4 points.

Again you are rewarded or penalized based on your actions, not your lack of actions. We can debate the reason for the so-called all zero penalty all day long, but the fact is that occurs.

My humble opinion, it occurs because you’re not using revolving credit, so how can it be evaluated and how can you be awarded for it? But let $1 report and you’re using revolving credit and therefore showing that you can manage it and so therefore points are given. However, if you use it to excess, less points are awarded.

Now, let’s flip to the aggregate installment utilization characteristic of amounts owed.

0% = 0 points
1%-9% =35 points
10%-64% = 12 points
65%-100% = 4 points

Take no action installment-wise, you get no points. Have an open one, get some points, pay it down, get more points the further it goes down. Pay it off, again you get zero.

In my opinion whether or not it’s a penalty or an award depends on what your actions can do. You’re not being penalized for not having a mortgage but you’re definitely being rewarded if you have one and it’s paid down or has been open for an extended period of time. To further buttress the argument, a perfect score is possible without a mortgage with that code still showing. This is true for many characteristics.

So @Curious_George2 , your statement: “If it's present, the score is lower than it would be if it were absent,” is not true, because a person can have a perfect 850 without a mortgage, with that reason code. And, although the code is present, the score is not lower than it would be absent the code.

So I posit that you are not penalized for the lack of a mortgage but the presence of one could offer more points.

[Anonymous]

---

@Curious_George2 wrote:

---

@Anonymous wrote:

---

@Curious_George2 wrote:

One theory I have is that these factors could be related to how credit mix is evaluated within the scoring system, but perhaps they are never actually used as reason codes. In other words, maybe they exist behind the scenes but are never deployed. That seems odd, but possible.

I don't think it's odd at all. All of those reason codes show us what questions they are asking (characteristics) about profile data, and by monitoring when they appear, we can figure out some rough thresholds or ranges (attribute range).

I may never see 'Lack of recent reported morgage loan information' or 'Lack of recent revolving HELOC information', but the code was definitely looking for a mortgage or HELOC on my file. No mortgage, no need to check for 'too many mortgage loans with balances'.

---

The reason I thought this was odd is that the score factors we are talking about are outputs. They aren't inputs or values derived and used solely within the scoring software. 

But I think I've found an answer to my own question. Somewhat surprisingly -- to me anyway -- there is no reason code for credit mix. I expected to see one that said something like "Insufficient variety of credit account types." But it's not there. Not in any of the five tables you've posted so far in the Score Factors thread. Weird, right? We know this is a scoring factor, so why isn't there a single reason code for it? I think the answer is that there are several reason codes for it... and my bête noire du jour, 79, is one of them. So, if a consumer has only revolvers on their report, they would lose points for credit mix and the reason codes that would be returned by the FICO TU8 and EX8 software boxes (though maybe not revealed to the consumer for reasons we all understand) would presumably include some or all of 79 (lack of mortgage), 32 (lack of installment loan), maybe 59 (lack of HELOC) and maybe even 50 (lack of retail accounts).

It's not that any one of those things is required, it's just that they've chosen to describe a poor mix by itemizing the credit types that are not represented. Slightly counterintuitive, but I guess it works and it's perfectly consistent with what Tom said in the AMA.  

Here's an interesting twist on this: consumer finance accounts might contribute to credit mix! Over at the official myFICO Credit Education entry for What Does Credit Mix Mean? it says the mix can include "credit cards, retail accounts, installment loans, finance company and mortgage loans." It seems likely that the negatives of a CFA would far outweigh this one little positive, but this would explain the existence of seemingly contradictory reason codes 99 and 6 in EX8 and TU8.

---

@Curious_George2 yes you are correct that "lack" would point to an opportunity to gain additional points through both credit mix and other characteristics.

yeah CFA is another category, but I don't think it's one that offers points. As far as I'm aware, revolving and installment are the two categories that offer points for mix diversity, open-ended is still out with the jury, but the number of revolvers for instance is a mix scoring factor and dictates a point award. The ratio of revolvers to loans is a mix scoring factor as well. 

[Anonymous]

---

@Curious_George2 wrote:

 Somewhat surprisingly -- to me anyway -- there is no reason code for credit mix. I expected to see one that said something like "Insufficient variety of credit account types." But it's not there. Not in any of the five tables you've posted so far in the Score Factors thread. Weird, right? We know this is a scoring factor, so why isn't there a single reason code for it? I think the answer is that there are several reason codes for it... and my bête noire du jour, 79, is one of them. So, if a consumer has only revolvers on their report, they would lose points for credit mix and the reason codes that would be returned by the FICO TU8 and EX8 software boxes (though maybe not revealed to the consumer for reasons we all understand) would presumably include some or all of 79 (lack of mortgage), 32 (lack of installment loan), maybe 59 (lack of HELOC) and maybe even 50 (lack of retail accounts).

That is exactly how it works. You just found it on your own through analytical reasoning. You're right.

Credit mix is a single category. A complete rating for credit mix comes from the answers (attributes) to multiple questions (characteristics) related to account types. These characteristics may also be influenced by scorecard and even other characteristics in a separate category.

That isn't a guess about FICO's algorithms either - it's just a part of general data science and logistic regression.

(Overview here: https://towardsdatascience.com/intro-to-credit-scorecard-9afeaaa3725f )

What we are all guessing about is whether these scoring models use a top-down or bottom-up approach in point calculations. The individual characteristics could start with maximum points and subtract for anything deemed negative, while the overall integration starts from a minimum. Everything could very well start from a maximum 850.

FICO examples for various characteristics even show that top-down approach in several 'reason computation' slides. One of them shows AoYRA maximum points at 45 (24 or more months, example points of course), with deviation from maximum contributing to the order of reason codes on our reports. Higher difference = Higher ranking.

Every reason statement that begins with 'Length of time' and doesn't specifically refer to 'open accounts', mentions that a score could be impacted by age of oldest and/or average ages. So yeah, Average Age of Mortage Accounts (AAoMA) is one of at least 2 attributes there, but how much does it influence that particular 'Length of time' characteristic? That would pretty hard to figure out.

Maybe we do all start with an 850 inside the algorithm and it scores from there. I don't have a problem looking at it that way. In that case, 'Lack of consumer finance accounts' might return maximum points for that part of the scoring (ranking value of 0, so it's waaay down there), since those accounts were found to appear more on the riskier profiles.

[Anonymous]

I totally agree, but I believe it’s bottom up; otherwise Quinn would be lying. If it was top down, you would be penalized for the failure to have a mortgage.

In my opinion, each characteristic can have award/penalty values that go from zero to a negative number, or from zero to a positive number, depending on the characteristic.

I think your actions determine the award/penalty, not your omissions. However your omissions can hurt you, but that’s like saying I would’ve gotten that promotion had I done this this this and this. Are you being penalized for your failure to do those things? Or are you simply not being rewarded because you did not earn them?